Fuzzy logic-based induction motor protection system (PDF Download ...
Neural Comput & Applic DOI 10.1007/s00521-012-0862-0
ORIGINAL ARTICLE
Fuzzy logic-based induction motor protection system Okan Uyar • Mehmet C ¸ unkas¸
Received: 15 August 2011 / Accepted: 23 January 2012 Springer-Verlag London Limited 2012
Abstract The protection is very important to detect abnormal motor running conditions such as over current, over voltage, overload, over temperature, and so on. When a failure is sensed by the protection system, a time delay should be specified to trip the motor. In the classical systems, motors are stopped with the time delay, which is adjusted constantly without considering the fault level. This paper presents a fuzzy logic-based protection system covering six different fault parameters for induction motors. This paper focuses on a new time-delay calculation for stopping induction motor and improves the overall detection performance. The time delay is computed by fuzzy logic method according to various fault parameters when one of the failures occurs on the motor. This system is successfully tested in real-time faults on the motor, and it shows that it provides sensitive protection by fuzzy rules. Keywords Fault diagnosis Fuzzy logic Induction motor protection
1 Introduction Induction motors are widely used in many industrial applications because of their simple construction, high reliability, relatively low cost, and low maintenance O. Uyar Techical Education Faculty, Electronic and Computer Education, Selc¸uk University, 42003 Konya, Turkey e-mail: [email protected] M. C¸unkas¸ (&) Technology Faculty, Electric and Electronic Engineering, Selc¸uk University, 42003 Konya, Turkey e-mail: [email protected]
requirements. The long-term working without interruption is very crucial for production systems. Many electrical and mechanical errors such as a cable cut, phase failure, short circuit, or overload can lead to break down of induction motors or pose danger for the whole production line and for the people who work there. Motor operating life is reduced if these faults left undetected. In addition, it causes delay of production process and financial losses due to stopping motor. This is the reason why a motor protection is very import and in cases of faults or damage, every motor needs a reliable protection [1, 2]. The major faults of induction motors can generally be categorized as follows: [26] 1. 2. 3. 4.
stator faults resulting in the opening or shorting of the phase winding; rotor cage failures (broken rotor bars/end-rings); bearing and gearbox failures; air-gap irregularities (static/dynamic).
The surveys of faults of induction motors focus on monitoring of thermal, vibration, electrical, noise, torque, and flux so far. The electricity-related faults are one of the important problems that demands special attention. Various sensors are used to detect the distinctive signals resulting from these faults. To extract the particular features from the faults, different types of signal processing techniques are applied to these sensor signals [27]. Nowadays, microcontroller-based methods can monitor the operating induction motor continuously that does not require human inspection to detect faults in motor. There has been extensive research on detecting abnormal conditions of induction machine using microcontroller-based systems [17–21]. In the previous studies, the time delays are adjusted constantly without considering fault level. When an error
123
Neural Comput & Applic
was detected, the motor is disconnected from supply without waiting for recovering errors or after the constant time delay. This situation can lead disruption of the production line. Similarly, if protection system waits too long to stop motor in a crucial fault, it causes serious defects. This paper focuses on a new time-delay calculation for stopping induction motor and improves the overall detection performance. The protection system is proposed to trip the motor in which the time delay is calculated between 0 and 4.5 s according to the fault level using the fuzzy logic method. If the failure is not recovered within that time, the system sends the trip signal to the solid-state relay to stop motor. So the flexible control was performed for especially industrial applications in which the time delay plays an important role. In addition to the performance of the control, strategy has been verified by MATLAB/Simulink. The developed system offers comprehensive motor protection as well as automation, control, communication, and monitoring functions. The protection method can be adapted to more than one induction motor.
2 Related works Many monitoring and protection systems are proposed for commonly encountered failures today. Approximately, 30–40% of these faults consist of stator failures [3]. The diagnostic methods for electrical machines with special reference to induction machines are summarized under four main titles. These are electrical and mechanical failures, signal processing, and artificial intelligence techniques [4]. Some studies using stator currents can diagnose the mechanical faults such as broken rotor bar, bearing, stator, and armature failures. The conventional methods detect the mechanical failures of induction motors by means of mathematical models [5–10]. However, these methods may be insufficient for complex and nonlinear error conditions that cannot be modeled mathematically. Soft computing and other computing methods such as signal processing and fast Fourier transform are used to increase accuracy and efficiency of detection methods. Soft computing techniques provide more advanced solutions than classical methods in error detection and diagnosis problems. Artificial intelligence methods such as neural networks, fuzzy logic, genetic algorithms, and their hybrids are useful in this type of problems [11–16]. Mechanic and electronic motor protection relays are still widely used in the motor protection applications. Recently, another class of motor protection consists of PLC and microprocessor-based systems [17–19]. One of most important features of these microprocessor-based systems is their ability to communicate data back to master computer. So, the system can provide motor protection, control,
123
metering, diagnostics, statistical functions, and communication. Current, voltage, and temperature values are measured in online fault detection systems. Measured data are transferred to computer by means of PLC, microcontroller, or data acquisition card; so many parameters are monitored through the designed software. When an error detected, the motor has stopped automatically or with the help of operator [20, 21]. A new time-delay calculation for tripping induction motor due to the fault condition that are the subject of this study are briefly introduced in the next section.
3 Proposed protection system The protection system is to detect the failures in a reliable and quick manner that occurs in the operating motor. The system is expected to perform the protection functions against the following conditions [22]. Over voltage: It is a condition in which voltage is higher than the rated level that is most often used to refer to voltage values in power lines. The problem can affect electric and electronics devices in all industries connected to the power line. Over current: It is a condition in which current is higher than the rated current capacity that can be caused by a short circuit, loose connection, and ground fault. Winding temperature is one of the most important parameters in induction motors since every rise in temperature greatly reduces the expected life span. The heat occurs in the winding of the motors if they operate for a long time. This protection can prevent a premature insulation failure. Voltage unbalance occurs in a three-phase system where the magnitudes of phase or line voltages are different and the phase angles differ from the balanced conditions, or voltage unbalance generally causes torque, speed variations, additional losses, current unbalance, and noise in three-phase motors [23]. This paper uses the maximum deviation from the rated values of three line voltages. Thus, the voltage unbalance is computed in percentage as in (1). Vmax VN Vmax [ VN
min %V ¼ VmaxVV 100 N Vmax Vmin %V ¼ Vmax 100
ð1Þ
Vmin, minimum phase voltage; Vmax, maximum phase voltage; VN, full-load voltage. Current unbalance leads to higher losses, circulating currents in windings, and increased heat in the motor. The current unbalance stems from voltage unbalance, large and unequal distribution of phase loads, and the circulating currents in windings can cause bearing failures. The current unbalance is computed in percentage as in (2).
Neural Comput & Applic
Imax IN Imax [ IN
min %I ¼ ImaxII 100 N Imax Imin %I ¼ I max 100
ð2Þ
Imin, minimum phase current; Imax, maximum phase current; IN, full-load current Under voltage could occur when a motor is suddenly connected to a power supply and voltage falls to a level too low for safe operation of motor. The under voltage increases the motor current, the rated copper losses, and temperature of windings. Phase interruption is the situation in which the interruption of one of three phases takes place. 3.1 Hardware The block diagram of system is given in Fig. 1. System control is based on PIC18F4620 microcontroller that includes UART communication module and analog–digital converter (ADC). Thus, any additional hardware is not need for communication with computer and processing of voltage, current, and temperature values. Graphical LCD having KS108 chip was used for displaying measured values. The user-interface card was constituted with six button added on graphical LCD card to input some settings. Figure 2 illustrates the experimental setup for fault detection. In this system, the induction motor is connected to a three-phase supply of 400 V, 50 Hz through Variacs. The DC generator is used to load the motor. The characteristics of induction motor chosen for testing are given in Table 1. A current transducer measuring up to 200 A and working with hall-effect principle was used to measure current. The measurement range can be adjusted as 50, 100, or 200 with dip switch on the transducer. Since
maximum current value in the system was less than 50 A, the range was set to 50 A. The voltage transducer produces analog voltage between 0 and 10 V in proportion to the measured voltage values. Additionally, temperature transducer can measure up to 1,200C that produces linear voltage in proportional to the measured temperature. This transducer’s measurement range was set to 0–300C. If current and voltage transducers are supplied by a regulated power source as in this study, the calibration set should be done. Thus, in the case where the input value is zero, Zero/ Span trimpots are set until the output voltages reach to zero. The signal communicated through optocoupler is sent to SSR (Solid State Relay) for controlling motor. The optocoupler provides isolation between the control side and power side that protects the control system from electrical noises and arcs on the main power lines. There are various input–output modules in block of control unit such as SSR drive circuit with optocoupler and buzzer driver with transistor. The system control panel is shown in Fig. 3. This enables you to view and change a wide range of parameters. The Menu button (M) is connected to RB0 external interrupt of microcontroller. Thus, microcontroller discontinues the current operations for achieving setting mode. The Control buttons are used efficiently to accomplish its tasks. Consequently, each protection function is set the suitable parameters using the adjustment mode. In addition to the faults of induction, motor can simulate using the adjustment mode without loading the motor and thus a time delay can generate. The following steps should be followed for setting parameters. •
Press Menu (M) button.
Fig. 1 Structure of experimental setup
123
Neural Comput & Applic Fig. 2 Experimental setup
Table 1 Characteristics of induction motor
3.2 Software
Power
2.2 kW
Speed
1,500 d/d
Voltage
220/380 V
Torque
16 Nm
Current
5.3 A
Efficiency
81%
Fig. 3 LCD display in adjustment mode
•
• • •
Select the cell for setting by the Up and Down buttons (between 1 and 7, Menu number can be seen above the thermometer symbol). Press the Menu button again to complete selection. Set the value of the selected area with the Plus and Minus buttons. Press the Enter button to accept the changes.
123
Assembly language programing was used because of its close relationship between the hardware languages that use less hardware source and work faster. However, the writing program is relatively harder than high-level languages. Because programers have to deal with the hardware devices such as ADC, UART, or any internal module. The troubleshooting of program is so hard because of longer program codes. Many companies considering these disadvantages develop higher-level programing languages. The most widely used compilers are CCS C, Hi-Tech C, Proton Basic, Pic Basic Plus, Micro Basic, Micro C etc. The CCS C programing language was chosen for the system software. Unlike others, the compiler almost completely supports the ANSI C language, PIC12XX, PIC14XX, PIC16XX, PIC18XX products. There are different versions for 24 bit PIC and dsPIC microcontrollers. This compiler also has many library files including a lot of functions for easy use of peripheral devices. The flowchart for the software algorithm can be seen in Fig. 4. The phase voltages, the phase currents and the winding temperature are measured and displayed on the LCD. They are compared with their predefined limit values. If any fault is detected, the program jumps to the fuzzy
Neural Comput & Applic Fig. 4 The flowchart for software algorithm
decision mechanism and the time delay is calculated. After the time delay has elapsed, the motor is disconnected from the supply immediately whether there has been any improvement in motor faults. The error description message is displayed on the LCD. When commissioning any induction motor, some problems can arise such as over current and overloading. Therefore, while the motor runs in a transient mode, the protection is not available until a certain time. The motor can carry the starting current safely and the system protection starts after the transient time. Hereafter, the program continues to measure and calculate the parameters by returning to the beginning.
4 Fuzzy logic system for motor protection Fuzzy logic is a type of mathematics and programing that helps the making a model by using learning and experience for representation of vague concepts in mathematical expressions. Therefore, fuzzy systems are very useful in situations involving highly complex systems whose behaviors are not well understood and in situations where
an approximate, but fast, solution is warranted. A fuzzy system attempts to understand a system for which no model exists, and it does so with information that can be uncertain in a sense of being vague, or fuzzy, or imprecise, or altogether lacking. Behaviors of systems, which are both understood and controllable, display certain robustness to spurious changes. Fuzzy systems are robust because the system has been designed to control within some frame of uncertain conditions. Outputs of the system are used in formulating the system structure itself. Conventional systems analyze requires a model based on a collective set of assumptions needed to formulate a mathematical form. Since most industrial applications currently are more complex, fuzzy control should be developed about the control process exists by using a number of fuzzy rules [24]. In this study, fuzzy logic-based induction motor protection systems was developed. In this system, the timedelay calculation used to stop motor was derived with fuzzy rules. These values were determined according to characteristics of the chosen motor. The values in Table 2 were used for preparing fuzzy rule base. The limit values
123
Neural Comput & Applic Table 2 Input–output limit values Min
Table 3 1st rule base samples Max
Unit
Rule no.
Over voltage
Over current
Temperature
Out
Over voltage
240
270 and more
V
1
OVL
OCL
TL
VL
Over current
5.5
8.4 and more
A
2
OVL
OCL
TM
VL
Temperature
135
155 and more
C
3
OVL
OCM
TH
N
Voltage unbalance
20
50 and more
V
4
OVL
OCH
TL
LN
Current unbalance
0.5
2 and more
A
5
OVM
OCH
TH
VS
Under voltage
200
160 and less
V
6
OVH
OCL
TL
LN
Time (output)
0
4.5
s
7
OVL
OCM
NA
LN
8
OVL
NA
TL
VL
9
NA
OCL
TL
VL
10
OVL
NA
NA
LN
11
NA
OCM
NA
N
NA not available
Table 4 2nd rule base samples
Fig. 5 Structure of fuzzy control system
were selected in accordance with NEMA standards and TS3205 EN 60034-1 numbered standard of TSI (Turkish Standards Institution). Figure 5 shows the structure of fuzzy control system. Two separate rule bases were constituted by dividing six different input values to reduce processing time and size of the constituted fuzzy system. The fuzzy expressions consist of three membership levels such as low–medium–high. While 729 rules (36) must be generated before dividing inputs into two parts, the rules are decreased to 126 by separating into two groups. The abbreviations used in rule base are listed in ‘‘Appendix 1’’. Some part of the rule bases used in protection system are listed in Tables 3 and 4. The blank cells in the tables show that this particular error did not occur in the induction motor. The membership functions for over voltage and time delay are listed in Table 5. The membership functions are shown in Figs. 6 and 7. In this study, the Simulink model was developed to test the performance of the control strategy. Simulink model was given in ‘‘Appendix 2’’. The function blocks in ‘‘Appendix 2’’ can be explained as follows:
123
Rule no.
Voltage unbalance
Current unbalance
Under voltage
Out
1
VUL
CUL
LVL
VL
2
VUL
CUL
LVM
VL
3
VUL
CUL
LVH
LN
4
VUL
CUM
LVL
VL
5
VUL
CUM
LVM
LN
6
VUL
CUM
LVH
N
7
VUL
CUH
LVL
LN
8
VUM
CUM
NA
N
9
VUL
NA
LVL
VL
10
NA
CUL
LVL
VL
11
VUL
NA
NA
LN
NA not available
GR-1, GR-2, and GR-3 are inputs for phase voltages, phase currents, and winding temperature, respectively. Min–max compares the minimum and maximum of input values. Rule base 1–2: is block of fuzzy logic, which calculates the time delay. Display shows the minimum one of the calculated delay times. Voltage–current unbalance compares the input voltages and currents. To do this, three inputs are subtracted from one another and the biggest absolute value is considered a result. The internal structure of this block was given in ‘‘Appendix 3’’.
5 Experimental results In this study, a flexible protection system based on fuzzy logic method was implemented for induction motors. Three-phase four-pole 2.2-kW induction motor was run
Neural Comput & Applic Table 5 Membership functions
Table 6 Experimental results for rule base 1
Over voltage OVL OVM OVH
lOVL ðxÞ ¼ 1 ¼ 260 x=10
240 B x \ 250
Test no.
Over voltage (V)
Over current (A)
Temp. (C)
Time delay (s)
250 B x \ 260
1
256.5
6.4
146.7
2.5
lOVM ðxÞ ¼ 260 x=10 ¼ 270 x=10
250 B x \ 260
2
260.9
8.6
155.2
1.5
260 B x \ 270
3
256.9
8.8
NA
2
lOVH ðxÞ ¼ 260 x=10 ¼1
260 B x \ 270
4
257.2
5.7
NA
2.3
270 B x
5
246.6
NA
138.2
3.5
6
NA
8.6
155.5
1.2
278.2 NA
NA NA
NA 153.8
2.3 1.5
Time delay lVS ðxÞ ¼ 1 ¼ 1:5 x=0:75
0 B x \ 0.75 0.75 B x \ 1.5
7 8
lS ðxÞ ¼ 1:5 x=0:75 ¼ 2:25 x=0:75
0.75 B x \ 1.5
NA not available
lN ðxÞ ¼ 2:25 x=0:75 ¼ 3 x=0:75
1.5 B x \ 2.25 2.25 B x \ 3
Table 7 Experimental results for rule base 2
L
lL ðxÞ ¼ 3 x=0:75 ¼ 3:75 x=0:75
2.25 B x \ 3
Test no.
Voltage unbalance (V)
Current unbalance (A)
Under voltage (V)
Delay time (s)
VL
lVL ðxÞ ¼ 3:75 x=0:75 ¼1
3 B x \ 3.75
1 2
23.8 30.9
0.9 1.6
192.5 186.2
3.4 2.8
3
54
1.5
187.4
2.1
4
37.6
1.5
NA
2.5
5
59.8
NA
168.3
1.1
6
NA
2.6
168.8
1.1
7
278.2
NA
NA
2.3
8
NA
NA
182.1
2.4
VS S N
1.5 B x \ 2.25
3 B x \ 3.75 3.75 B x
NA not available
Fig. 6 Membership function for over voltage
Fig. 7 Membership function for time delay
directly at 50 Hz. The DC generator was used to load the induction motor for over current test. Over voltage and under voltage were produced by using variable transformer with ratings from 50 to 440 V. The temperature of motor winding was measured by using the thermocouple.
The time delays produced according to two separate rule base are given in Tables 6 and 7. The blank cells in the tables show that there is no fault in this parameter. In the experimental setup, the nominal phase voltage is 220 V and the nominal phase current is 5.3 A. Max winding temperature in nominal operating mode is 140C. As seen from Table 6, any value greater than these nominal values is generally considered a fault condition. As seen from Table 7, the maximum permissible voltage unbalance and current unbalance are ±10%. The value greater than this limit is considered a fault condition. The permissible under voltage is 200 V. The value smaller than this value is considered the under voltage. The authors developed Simulink model to simulate the time delay and compare with the experimental results. The model can produce appropriate motor data for different operating and loading conditions [25]. The input clusters and fuzzy rule bases have been entered to FIS GUI, fuzzy logic toolbox of Matlab. The calculated time delays using Simulink model are given in Tables 8 and 9. The test numbers are the same the test numbers in Tables 6 and 7. For example, inputs and results of rule 1 in Table 7 are as follows:
123
Neural Comput & Applic Table 8 The delay times for rule base 1 (unit: second) Test no.
1
2
3
Experimental results
2.5
1.5
Simulink results
2.5
1.5
Difference
0
0
4 2
5 2.3
2.1
2.4
-0.1
-0.1
6
3.5
7 1.2
3.5 0
8 2.3
1.5
1.1
2.2
1.5
?0.1
?0.1
0
7
8
Table 9 The delay times for rule base 2 (unit: second) Fig. 9 Simulation and experimental results for rule base 2 Test no.
1
Experimental results
3.4
Simulink results
3.4
Difference
0
2
3
4
5
6
2.1
2.5
1.1
1.1
2.9
2.2
2.5
1.1
1.1
-0.1
?0.1
0
0
0
2.8
2.3
2.4
6 Conclusion
2.4
2.4
This paper presents a fuzzy logic-based protection system detecting faulty operations for a three-phase induction motor. The system provides a protection against over voltage, over current, temperature rise, voltage unbalance, current unbalance, and under voltage situations. In many protection systems, when an error is detected, the system waits for a certain time and then stops induction motor unless fault recovered. The time delay is adjusted manually according to the error type. The time delay needed for stopping the motor is critical, and so it must be set by experts. The proposed system produces the time delay using fuzzy logic method for different levels of various error combinations. So the flexible and optimal time delay is determined. The experimental results are compared with the Simulink results to verify the effectiveness of the proposed method. It is seen that this methodology is both accurate and easy-to-implement. The developed system can enhances induction motors with different power ratings.
-0.1
0
Fig. 8 Simulation and experimental results for rule base 1
Over voltage : 256:5 V Over current : 6:4 A
Temperature : 146:7 C : Output ðcalculated time-delay) : 2:5 s
This output (2.5 s) is used in Table 8 with the Test Number 1. As can be seen from Tables 8 and 9, there is a little difference (-0.1 and ?0.1 s.) between experimental results and Matlab results. This difference may be caused by the following. Although Matlab is capable of evaluating floating point operations precisely, the calculations in microcontroller software have carried out with numbers having single digit after comma. Also, the centroid method used in fuzzy logic is evaluated by using linear approximation instead of integral method. Figures 8 and 9 show the differences between the Matlab-Simulink and the experimental results. The experimental results are quite similar to the Simulink results. This demonstrates the accuracy and the robustness of the developed protection system. It is clear from these results that the fuzzy logic-based protection system has successfully diagnosed the faults for a three-phase induction motor.
123
Acknowledgment This work was supported by Scientific Research Project Coordination of Selcuk University under Project with 09101047 project number.
Appendix 1: Abbreviations
Abbr.
Description
Abbr.
Description
OC
Over current
VS
Very short
OV
Over voltage
S
Short
T
Temperature
N
Normal
CU
Current unbalance
LN
Long
VU
Voltage unbalance
VL
Very long
LV
Under voltage
OCL
Over current low
L
Low
OVM
Over voltage medium
M
Medium
TH
Temperature high
H
High
Neural Comput & Applic
Appendix 2
Appendix 3
References 1. http://library.abb.com/global/scot/scot209.nsf/veritydisplay/ad42 3934ae0700f2c1256f48003a6fe1/File/2CDC131013B0201.pdf. Accessed 30 July 2011 2. Arslan M, C¸unkas¸ M, Sag˘ T (2011) Determination of induction motor parameters with differential evolution algorithm. Neural Comput Appl. doi:10.1007/s00521-011-0612-8 3. Siddique A, Yadava GD, Singh B (2005) A review of stator fault monitoring techniques of induction motors. IEEE Trans Energy Convers 20(1):106–114 4. Bellini A, Fillippetti F, Tassoni C, Capolino G (2008) Advances in diagnostic techniques for induction machines. IEEE Trans Energy Convers 55(12):4109–4126 ¨ zgo¨nenel O, Yalc¸ın T (2011) A complete motor protection 5. O algorithm based on PCA and ANN: a real time study. Turk J Elect Eng Comput Sci 19(3):317–334 6. Zarei J, Poshtan J (2007) Bearing fault detection using wavelet packet transform of induction motor stator current. Tribol Int 40(5):763–769 7. Benbouzid M, Vieira M, Theys C (1999) Induction motors’ faults detection and localization using stator current advanced signal processing techniques. IEEE Trans Power Electron 14(1):14–22 8. Blo¨dt M, Chabert M, Regnier J, Faucher J (2006) Mechanical load fault detection in induction motors by stator current time– frequency analysis. IEEE Trans Ind Appl 42(6):1454–1463
123
Neural Comput & Applic 9. Ayhan B, Chow M, Song M (2005) Multiple signature processing-based fault detection schemes for broken rotor bar in induction motors. IEEE Trans Energy Convers 20(2):336–343 10. Rajagopalan S, Restrepo J, Aller J, Habetler T, Harley R (2008) Nonstationary motor fault detection using recent quadratic time– frequency representations. IEEE Trans Ind Appl 44(3):735–744 11. Kowalski CT, Orlowska-Kowalska T (2003) Neural networks application for induction motor faults diagnosis. Math Comput Simul 63(3–5):435–448 12. Su H, Chong KT, Kumar RR (2011) Vibration signal analysis for electrical fault detection of induction machine using neural networks. Neural Comput Appl 20(2):183–194 13. Kim K, Parlos AG (2002) Model-based fault diagnosis of induction motors using non-stationary signal segmentation. Mech Syst Signal Process 16(2–3):223–253 14. Martins JF, Pires VF, Pires AJ (2007) Unsupervised neural-network-based algorithm for an on-line diagnosis of three-phase induction motor stator fault. IEEE Trans Ind Electron 54(1): 259–264 15. Gao XZ, Ovaska SJ (2001) Soft computing methods in motor fault diagnosis. Appl Soft Comput 1(1):73–81 16. Arabacı H, Bilgin O (2010) Automatic detection and classification of rotor cage faults in squirrel cage induction motor. Neural Comput Appl 19:713–723 17. C¸olak ˙I, C¸elik H, Sefa ˙I, Demirbas¸ S¸ (2005) On line protection systems for induction motors. Energy Convers Manag 46(17): 2773–2786 18. C¸olak I˙, Bayindir R, Bektas¸ A, Sefa ˙I, Bal G (2007) Protection of induction motor using PLC. International Conference on Power
123
19.
20.
21.
22.
23. 24. 25.
26.
27.
Engineering, Energy and Electrical Drives (POWERENG), pp 96–99, Portugal Bayindir R, Sefa ˙I (2007) Novel approach based on microcontroller to online protection of induction motors. Energy Convers Manag 48(3):850–856 Bayindir R, Sefa ˙I, C¸olak ˙I, Bektas A (2008) Fault detection and protection of induction motors using sensors. IEEE Trans Energy Convers 23(3):734–741 ¨ ztu¨rk A (2000) Protection of AC motors C¸unkas¸ M, Akkaya R, O by means of microcontrollers. 10th Mediterranean Electrotechnical Conference (MELECON), vol 3, pp 1093–1096 Uyar O (2010) Design and implementation of fuzzy logic based protection system for induction motors. MS thesis, Selcuk University, Turkey Pillay P, Manyage M (2001) Definitions of voltage unbalance. IEEE Power Eng Rev 21(5):10–12 Ross TJ (2004) Fuzzy logic with engineering applications. Wiley, London Uyar O, C¸unkas¸ M (2011) Design of fuzzy logic based motor protection system. 6th International Advanced Technologies Symposium (IATS’11), 16–18 May 2011, Turkey Zhang P, Du Y, Habetler TG, Lu B (2011) A survey of condition monitoring and protection methods for medium-voltage induction motors. IEEE Trans Ind Appl 47(1):34–46 Sin ML, Erugrul N (2003) Induction machine on-line condition monitoring and fault diagnosis—a survey. AUPEC2003, Australasian Universities Power Engineering Conference, Christchurch, New Zealand
ORIGINAL ARTICLE
Fuzzy logic-based induction motor protection system Okan Uyar • Mehmet C ¸ unkas¸
Received: 15 August 2011 / Accepted: 23 January 2012 Springer-Verlag London Limited 2012
Abstract The protection is very important to detect abnormal motor running conditions such as over current, over voltage, overload, over temperature, and so on. When a failure is sensed by the protection system, a time delay should be specified to trip the motor. In the classical systems, motors are stopped with the time delay, which is adjusted constantly without considering the fault level. This paper presents a fuzzy logic-based protection system covering six different fault parameters for induction motors. This paper focuses on a new time-delay calculation for stopping induction motor and improves the overall detection performance. The time delay is computed by fuzzy logic method according to various fault parameters when one of the failures occurs on the motor. This system is successfully tested in real-time faults on the motor, and it shows that it provides sensitive protection by fuzzy rules. Keywords Fault diagnosis Fuzzy logic Induction motor protection
1 Introduction Induction motors are widely used in many industrial applications because of their simple construction, high reliability, relatively low cost, and low maintenance O. Uyar Techical Education Faculty, Electronic and Computer Education, Selc¸uk University, 42003 Konya, Turkey e-mail: [email protected] M. C¸unkas¸ (&) Technology Faculty, Electric and Electronic Engineering, Selc¸uk University, 42003 Konya, Turkey e-mail: [email protected]
requirements. The long-term working without interruption is very crucial for production systems. Many electrical and mechanical errors such as a cable cut, phase failure, short circuit, or overload can lead to break down of induction motors or pose danger for the whole production line and for the people who work there. Motor operating life is reduced if these faults left undetected. In addition, it causes delay of production process and financial losses due to stopping motor. This is the reason why a motor protection is very import and in cases of faults or damage, every motor needs a reliable protection [1, 2]. The major faults of induction motors can generally be categorized as follows: [26] 1. 2. 3. 4.
stator faults resulting in the opening or shorting of the phase winding; rotor cage failures (broken rotor bars/end-rings); bearing and gearbox failures; air-gap irregularities (static/dynamic).
The surveys of faults of induction motors focus on monitoring of thermal, vibration, electrical, noise, torque, and flux so far. The electricity-related faults are one of the important problems that demands special attention. Various sensors are used to detect the distinctive signals resulting from these faults. To extract the particular features from the faults, different types of signal processing techniques are applied to these sensor signals [27]. Nowadays, microcontroller-based methods can monitor the operating induction motor continuously that does not require human inspection to detect faults in motor. There has been extensive research on detecting abnormal conditions of induction machine using microcontroller-based systems [17–21]. In the previous studies, the time delays are adjusted constantly without considering fault level. When an error
123
Neural Comput & Applic
was detected, the motor is disconnected from supply without waiting for recovering errors or after the constant time delay. This situation can lead disruption of the production line. Similarly, if protection system waits too long to stop motor in a crucial fault, it causes serious defects. This paper focuses on a new time-delay calculation for stopping induction motor and improves the overall detection performance. The protection system is proposed to trip the motor in which the time delay is calculated between 0 and 4.5 s according to the fault level using the fuzzy logic method. If the failure is not recovered within that time, the system sends the trip signal to the solid-state relay to stop motor. So the flexible control was performed for especially industrial applications in which the time delay plays an important role. In addition to the performance of the control, strategy has been verified by MATLAB/Simulink. The developed system offers comprehensive motor protection as well as automation, control, communication, and monitoring functions. The protection method can be adapted to more than one induction motor.
2 Related works Many monitoring and protection systems are proposed for commonly encountered failures today. Approximately, 30–40% of these faults consist of stator failures [3]. The diagnostic methods for electrical machines with special reference to induction machines are summarized under four main titles. These are electrical and mechanical failures, signal processing, and artificial intelligence techniques [4]. Some studies using stator currents can diagnose the mechanical faults such as broken rotor bar, bearing, stator, and armature failures. The conventional methods detect the mechanical failures of induction motors by means of mathematical models [5–10]. However, these methods may be insufficient for complex and nonlinear error conditions that cannot be modeled mathematically. Soft computing and other computing methods such as signal processing and fast Fourier transform are used to increase accuracy and efficiency of detection methods. Soft computing techniques provide more advanced solutions than classical methods in error detection and diagnosis problems. Artificial intelligence methods such as neural networks, fuzzy logic, genetic algorithms, and their hybrids are useful in this type of problems [11–16]. Mechanic and electronic motor protection relays are still widely used in the motor protection applications. Recently, another class of motor protection consists of PLC and microprocessor-based systems [17–19]. One of most important features of these microprocessor-based systems is their ability to communicate data back to master computer. So, the system can provide motor protection, control,
123
metering, diagnostics, statistical functions, and communication. Current, voltage, and temperature values are measured in online fault detection systems. Measured data are transferred to computer by means of PLC, microcontroller, or data acquisition card; so many parameters are monitored through the designed software. When an error detected, the motor has stopped automatically or with the help of operator [20, 21]. A new time-delay calculation for tripping induction motor due to the fault condition that are the subject of this study are briefly introduced in the next section.
3 Proposed protection system The protection system is to detect the failures in a reliable and quick manner that occurs in the operating motor. The system is expected to perform the protection functions against the following conditions [22]. Over voltage: It is a condition in which voltage is higher than the rated level that is most often used to refer to voltage values in power lines. The problem can affect electric and electronics devices in all industries connected to the power line. Over current: It is a condition in which current is higher than the rated current capacity that can be caused by a short circuit, loose connection, and ground fault. Winding temperature is one of the most important parameters in induction motors since every rise in temperature greatly reduces the expected life span. The heat occurs in the winding of the motors if they operate for a long time. This protection can prevent a premature insulation failure. Voltage unbalance occurs in a three-phase system where the magnitudes of phase or line voltages are different and the phase angles differ from the balanced conditions, or voltage unbalance generally causes torque, speed variations, additional losses, current unbalance, and noise in three-phase motors [23]. This paper uses the maximum deviation from the rated values of three line voltages. Thus, the voltage unbalance is computed in percentage as in (1). Vmax VN Vmax [ VN
min %V ¼ VmaxVV 100 N Vmax Vmin %V ¼ Vmax 100
ð1Þ
Vmin, minimum phase voltage; Vmax, maximum phase voltage; VN, full-load voltage. Current unbalance leads to higher losses, circulating currents in windings, and increased heat in the motor. The current unbalance stems from voltage unbalance, large and unequal distribution of phase loads, and the circulating currents in windings can cause bearing failures. The current unbalance is computed in percentage as in (2).
Neural Comput & Applic
Imax IN Imax [ IN
min %I ¼ ImaxII 100 N Imax Imin %I ¼ I max 100
ð2Þ
Imin, minimum phase current; Imax, maximum phase current; IN, full-load current Under voltage could occur when a motor is suddenly connected to a power supply and voltage falls to a level too low for safe operation of motor. The under voltage increases the motor current, the rated copper losses, and temperature of windings. Phase interruption is the situation in which the interruption of one of three phases takes place. 3.1 Hardware The block diagram of system is given in Fig. 1. System control is based on PIC18F4620 microcontroller that includes UART communication module and analog–digital converter (ADC). Thus, any additional hardware is not need for communication with computer and processing of voltage, current, and temperature values. Graphical LCD having KS108 chip was used for displaying measured values. The user-interface card was constituted with six button added on graphical LCD card to input some settings. Figure 2 illustrates the experimental setup for fault detection. In this system, the induction motor is connected to a three-phase supply of 400 V, 50 Hz through Variacs. The DC generator is used to load the motor. The characteristics of induction motor chosen for testing are given in Table 1. A current transducer measuring up to 200 A and working with hall-effect principle was used to measure current. The measurement range can be adjusted as 50, 100, or 200 with dip switch on the transducer. Since
maximum current value in the system was less than 50 A, the range was set to 50 A. The voltage transducer produces analog voltage between 0 and 10 V in proportion to the measured voltage values. Additionally, temperature transducer can measure up to 1,200C that produces linear voltage in proportional to the measured temperature. This transducer’s measurement range was set to 0–300C. If current and voltage transducers are supplied by a regulated power source as in this study, the calibration set should be done. Thus, in the case where the input value is zero, Zero/ Span trimpots are set until the output voltages reach to zero. The signal communicated through optocoupler is sent to SSR (Solid State Relay) for controlling motor. The optocoupler provides isolation between the control side and power side that protects the control system from electrical noises and arcs on the main power lines. There are various input–output modules in block of control unit such as SSR drive circuit with optocoupler and buzzer driver with transistor. The system control panel is shown in Fig. 3. This enables you to view and change a wide range of parameters. The Menu button (M) is connected to RB0 external interrupt of microcontroller. Thus, microcontroller discontinues the current operations for achieving setting mode. The Control buttons are used efficiently to accomplish its tasks. Consequently, each protection function is set the suitable parameters using the adjustment mode. In addition to the faults of induction, motor can simulate using the adjustment mode without loading the motor and thus a time delay can generate. The following steps should be followed for setting parameters. •
Press Menu (M) button.
Fig. 1 Structure of experimental setup
123
Neural Comput & Applic Fig. 2 Experimental setup
Table 1 Characteristics of induction motor
3.2 Software
Power
2.2 kW
Speed
1,500 d/d
Voltage
220/380 V
Torque
16 Nm
Current
5.3 A
Efficiency
81%
Fig. 3 LCD display in adjustment mode
•
• • •
Select the cell for setting by the Up and Down buttons (between 1 and 7, Menu number can be seen above the thermometer symbol). Press the Menu button again to complete selection. Set the value of the selected area with the Plus and Minus buttons. Press the Enter button to accept the changes.
123
Assembly language programing was used because of its close relationship between the hardware languages that use less hardware source and work faster. However, the writing program is relatively harder than high-level languages. Because programers have to deal with the hardware devices such as ADC, UART, or any internal module. The troubleshooting of program is so hard because of longer program codes. Many companies considering these disadvantages develop higher-level programing languages. The most widely used compilers are CCS C, Hi-Tech C, Proton Basic, Pic Basic Plus, Micro Basic, Micro C etc. The CCS C programing language was chosen for the system software. Unlike others, the compiler almost completely supports the ANSI C language, PIC12XX, PIC14XX, PIC16XX, PIC18XX products. There are different versions for 24 bit PIC and dsPIC microcontrollers. This compiler also has many library files including a lot of functions for easy use of peripheral devices. The flowchart for the software algorithm can be seen in Fig. 4. The phase voltages, the phase currents and the winding temperature are measured and displayed on the LCD. They are compared with their predefined limit values. If any fault is detected, the program jumps to the fuzzy
Neural Comput & Applic Fig. 4 The flowchart for software algorithm
decision mechanism and the time delay is calculated. After the time delay has elapsed, the motor is disconnected from the supply immediately whether there has been any improvement in motor faults. The error description message is displayed on the LCD. When commissioning any induction motor, some problems can arise such as over current and overloading. Therefore, while the motor runs in a transient mode, the protection is not available until a certain time. The motor can carry the starting current safely and the system protection starts after the transient time. Hereafter, the program continues to measure and calculate the parameters by returning to the beginning.
4 Fuzzy logic system for motor protection Fuzzy logic is a type of mathematics and programing that helps the making a model by using learning and experience for representation of vague concepts in mathematical expressions. Therefore, fuzzy systems are very useful in situations involving highly complex systems whose behaviors are not well understood and in situations where
an approximate, but fast, solution is warranted. A fuzzy system attempts to understand a system for which no model exists, and it does so with information that can be uncertain in a sense of being vague, or fuzzy, or imprecise, or altogether lacking. Behaviors of systems, which are both understood and controllable, display certain robustness to spurious changes. Fuzzy systems are robust because the system has been designed to control within some frame of uncertain conditions. Outputs of the system are used in formulating the system structure itself. Conventional systems analyze requires a model based on a collective set of assumptions needed to formulate a mathematical form. Since most industrial applications currently are more complex, fuzzy control should be developed about the control process exists by using a number of fuzzy rules [24]. In this study, fuzzy logic-based induction motor protection systems was developed. In this system, the timedelay calculation used to stop motor was derived with fuzzy rules. These values were determined according to characteristics of the chosen motor. The values in Table 2 were used for preparing fuzzy rule base. The limit values
123
Neural Comput & Applic Table 2 Input–output limit values Min
Table 3 1st rule base samples Max
Unit
Rule no.
Over voltage
Over current
Temperature
Out
Over voltage
240
270 and more
V
1
OVL
OCL
TL
VL
Over current
5.5
8.4 and more
A
2
OVL
OCL
TM
VL
Temperature
135
155 and more
C
3
OVL
OCM
TH
N
Voltage unbalance
20
50 and more
V
4
OVL
OCH
TL
LN
Current unbalance
0.5
2 and more
A
5
OVM
OCH
TH
VS
Under voltage
200
160 and less
V
6
OVH
OCL
TL
LN
Time (output)
0
4.5
s
7
OVL
OCM
NA
LN
8
OVL
NA
TL
VL
9
NA
OCL
TL
VL
10
OVL
NA
NA
LN
11
NA
OCM
NA
N
NA not available
Table 4 2nd rule base samples
Fig. 5 Structure of fuzzy control system
were selected in accordance with NEMA standards and TS3205 EN 60034-1 numbered standard of TSI (Turkish Standards Institution). Figure 5 shows the structure of fuzzy control system. Two separate rule bases were constituted by dividing six different input values to reduce processing time and size of the constituted fuzzy system. The fuzzy expressions consist of three membership levels such as low–medium–high. While 729 rules (36) must be generated before dividing inputs into two parts, the rules are decreased to 126 by separating into two groups. The abbreviations used in rule base are listed in ‘‘Appendix 1’’. Some part of the rule bases used in protection system are listed in Tables 3 and 4. The blank cells in the tables show that this particular error did not occur in the induction motor. The membership functions for over voltage and time delay are listed in Table 5. The membership functions are shown in Figs. 6 and 7. In this study, the Simulink model was developed to test the performance of the control strategy. Simulink model was given in ‘‘Appendix 2’’. The function blocks in ‘‘Appendix 2’’ can be explained as follows:
123
Rule no.
Voltage unbalance
Current unbalance
Under voltage
Out
1
VUL
CUL
LVL
VL
2
VUL
CUL
LVM
VL
3
VUL
CUL
LVH
LN
4
VUL
CUM
LVL
VL
5
VUL
CUM
LVM
LN
6
VUL
CUM
LVH
N
7
VUL
CUH
LVL
LN
8
VUM
CUM
NA
N
9
VUL
NA
LVL
VL
10
NA
CUL
LVL
VL
11
VUL
NA
NA
LN
NA not available
GR-1, GR-2, and GR-3 are inputs for phase voltages, phase currents, and winding temperature, respectively. Min–max compares the minimum and maximum of input values. Rule base 1–2: is block of fuzzy logic, which calculates the time delay. Display shows the minimum one of the calculated delay times. Voltage–current unbalance compares the input voltages and currents. To do this, three inputs are subtracted from one another and the biggest absolute value is considered a result. The internal structure of this block was given in ‘‘Appendix 3’’.
5 Experimental results In this study, a flexible protection system based on fuzzy logic method was implemented for induction motors. Three-phase four-pole 2.2-kW induction motor was run
Neural Comput & Applic Table 5 Membership functions
Table 6 Experimental results for rule base 1
Over voltage OVL OVM OVH
lOVL ðxÞ ¼ 1 ¼ 260 x=10
240 B x \ 250
Test no.
Over voltage (V)
Over current (A)
Temp. (C)
Time delay (s)
250 B x \ 260
1
256.5
6.4
146.7
2.5
lOVM ðxÞ ¼ 260 x=10 ¼ 270 x=10
250 B x \ 260
2
260.9
8.6
155.2
1.5
260 B x \ 270
3
256.9
8.8
NA
2
lOVH ðxÞ ¼ 260 x=10 ¼1
260 B x \ 270
4
257.2
5.7
NA
2.3
270 B x
5
246.6
NA
138.2
3.5
6
NA
8.6
155.5
1.2
278.2 NA
NA NA
NA 153.8
2.3 1.5
Time delay lVS ðxÞ ¼ 1 ¼ 1:5 x=0:75
0 B x \ 0.75 0.75 B x \ 1.5
7 8
lS ðxÞ ¼ 1:5 x=0:75 ¼ 2:25 x=0:75
0.75 B x \ 1.5
NA not available
lN ðxÞ ¼ 2:25 x=0:75 ¼ 3 x=0:75
1.5 B x \ 2.25 2.25 B x \ 3
Table 7 Experimental results for rule base 2
L
lL ðxÞ ¼ 3 x=0:75 ¼ 3:75 x=0:75
2.25 B x \ 3
Test no.
Voltage unbalance (V)
Current unbalance (A)
Under voltage (V)
Delay time (s)
VL
lVL ðxÞ ¼ 3:75 x=0:75 ¼1
3 B x \ 3.75
1 2
23.8 30.9
0.9 1.6
192.5 186.2
3.4 2.8
3
54
1.5
187.4
2.1
4
37.6
1.5
NA
2.5
5
59.8
NA
168.3
1.1
6
NA
2.6
168.8
1.1
7
278.2
NA
NA
2.3
8
NA
NA
182.1
2.4
VS S N
1.5 B x \ 2.25
3 B x \ 3.75 3.75 B x
NA not available
Fig. 6 Membership function for over voltage
Fig. 7 Membership function for time delay
directly at 50 Hz. The DC generator was used to load the induction motor for over current test. Over voltage and under voltage were produced by using variable transformer with ratings from 50 to 440 V. The temperature of motor winding was measured by using the thermocouple.
The time delays produced according to two separate rule base are given in Tables 6 and 7. The blank cells in the tables show that there is no fault in this parameter. In the experimental setup, the nominal phase voltage is 220 V and the nominal phase current is 5.3 A. Max winding temperature in nominal operating mode is 140C. As seen from Table 6, any value greater than these nominal values is generally considered a fault condition. As seen from Table 7, the maximum permissible voltage unbalance and current unbalance are ±10%. The value greater than this limit is considered a fault condition. The permissible under voltage is 200 V. The value smaller than this value is considered the under voltage. The authors developed Simulink model to simulate the time delay and compare with the experimental results. The model can produce appropriate motor data for different operating and loading conditions [25]. The input clusters and fuzzy rule bases have been entered to FIS GUI, fuzzy logic toolbox of Matlab. The calculated time delays using Simulink model are given in Tables 8 and 9. The test numbers are the same the test numbers in Tables 6 and 7. For example, inputs and results of rule 1 in Table 7 are as follows:
123
Neural Comput & Applic Table 8 The delay times for rule base 1 (unit: second) Test no.
1
2
3
Experimental results
2.5
1.5
Simulink results
2.5
1.5
Difference
0
0
4 2
5 2.3
2.1
2.4
-0.1
-0.1
6
3.5
7 1.2
3.5 0
8 2.3
1.5
1.1
2.2
1.5
?0.1
?0.1
0
7
8
Table 9 The delay times for rule base 2 (unit: second) Fig. 9 Simulation and experimental results for rule base 2 Test no.
1
Experimental results
3.4
Simulink results
3.4
Difference
0
2
3
4
5
6
2.1
2.5
1.1
1.1
2.9
2.2
2.5
1.1
1.1
-0.1
?0.1
0
0
0
2.8
2.3
2.4
6 Conclusion
2.4
2.4
This paper presents a fuzzy logic-based protection system detecting faulty operations for a three-phase induction motor. The system provides a protection against over voltage, over current, temperature rise, voltage unbalance, current unbalance, and under voltage situations. In many protection systems, when an error is detected, the system waits for a certain time and then stops induction motor unless fault recovered. The time delay is adjusted manually according to the error type. The time delay needed for stopping the motor is critical, and so it must be set by experts. The proposed system produces the time delay using fuzzy logic method for different levels of various error combinations. So the flexible and optimal time delay is determined. The experimental results are compared with the Simulink results to verify the effectiveness of the proposed method. It is seen that this methodology is both accurate and easy-to-implement. The developed system can enhances induction motors with different power ratings.
-0.1
0
Fig. 8 Simulation and experimental results for rule base 1
Over voltage : 256:5 V Over current : 6:4 A
Temperature : 146:7 C : Output ðcalculated time-delay) : 2:5 s
This output (2.5 s) is used in Table 8 with the Test Number 1. As can be seen from Tables 8 and 9, there is a little difference (-0.1 and ?0.1 s.) between experimental results and Matlab results. This difference may be caused by the following. Although Matlab is capable of evaluating floating point operations precisely, the calculations in microcontroller software have carried out with numbers having single digit after comma. Also, the centroid method used in fuzzy logic is evaluated by using linear approximation instead of integral method. Figures 8 and 9 show the differences between the Matlab-Simulink and the experimental results. The experimental results are quite similar to the Simulink results. This demonstrates the accuracy and the robustness of the developed protection system. It is clear from these results that the fuzzy logic-based protection system has successfully diagnosed the faults for a three-phase induction motor.
123
Acknowledgment This work was supported by Scientific Research Project Coordination of Selcuk University under Project with 09101047 project number.
Appendix 1: Abbreviations
Abbr.
Description
Abbr.
Description
OC
Over current
VS
Very short
OV
Over voltage
S
Short
T
Temperature
N
Normal
CU
Current unbalance
LN
Long
VU
Voltage unbalance
VL
Very long
LV
Under voltage
OCL
Over current low
L
Low
OVM
Over voltage medium
M
Medium
TH
Temperature high
H
High
Neural Comput & Applic
Appendix 2
Appendix 3
References 1. http://library.abb.com/global/scot/scot209.nsf/veritydisplay/ad42 3934ae0700f2c1256f48003a6fe1/File/2CDC131013B0201.pdf. Accessed 30 July 2011 2. Arslan M, C¸unkas¸ M, Sag˘ T (2011) Determination of induction motor parameters with differential evolution algorithm. Neural Comput Appl. doi:10.1007/s00521-011-0612-8 3. Siddique A, Yadava GD, Singh B (2005) A review of stator fault monitoring techniques of induction motors. IEEE Trans Energy Convers 20(1):106–114 4. Bellini A, Fillippetti F, Tassoni C, Capolino G (2008) Advances in diagnostic techniques for induction machines. IEEE Trans Energy Convers 55(12):4109–4126 ¨ zgo¨nenel O, Yalc¸ın T (2011) A complete motor protection 5. O algorithm based on PCA and ANN: a real time study. Turk J Elect Eng Comput Sci 19(3):317–334 6. Zarei J, Poshtan J (2007) Bearing fault detection using wavelet packet transform of induction motor stator current. Tribol Int 40(5):763–769 7. Benbouzid M, Vieira M, Theys C (1999) Induction motors’ faults detection and localization using stator current advanced signal processing techniques. IEEE Trans Power Electron 14(1):14–22 8. Blo¨dt M, Chabert M, Regnier J, Faucher J (2006) Mechanical load fault detection in induction motors by stator current time– frequency analysis. IEEE Trans Ind Appl 42(6):1454–1463
123
Neural Comput & Applic 9. Ayhan B, Chow M, Song M (2005) Multiple signature processing-based fault detection schemes for broken rotor bar in induction motors. IEEE Trans Energy Convers 20(2):336–343 10. Rajagopalan S, Restrepo J, Aller J, Habetler T, Harley R (2008) Nonstationary motor fault detection using recent quadratic time– frequency representations. IEEE Trans Ind Appl 44(3):735–744 11. Kowalski CT, Orlowska-Kowalska T (2003) Neural networks application for induction motor faults diagnosis. Math Comput Simul 63(3–5):435–448 12. Su H, Chong KT, Kumar RR (2011) Vibration signal analysis for electrical fault detection of induction machine using neural networks. Neural Comput Appl 20(2):183–194 13. Kim K, Parlos AG (2002) Model-based fault diagnosis of induction motors using non-stationary signal segmentation. Mech Syst Signal Process 16(2–3):223–253 14. Martins JF, Pires VF, Pires AJ (2007) Unsupervised neural-network-based algorithm for an on-line diagnosis of three-phase induction motor stator fault. IEEE Trans Ind Electron 54(1): 259–264 15. Gao XZ, Ovaska SJ (2001) Soft computing methods in motor fault diagnosis. Appl Soft Comput 1(1):73–81 16. Arabacı H, Bilgin O (2010) Automatic detection and classification of rotor cage faults in squirrel cage induction motor. Neural Comput Appl 19:713–723 17. C¸olak ˙I, C¸elik H, Sefa ˙I, Demirbas¸ S¸ (2005) On line protection systems for induction motors. Energy Convers Manag 46(17): 2773–2786 18. C¸olak I˙, Bayindir R, Bektas¸ A, Sefa ˙I, Bal G (2007) Protection of induction motor using PLC. International Conference on Power
123
19.
20.
21.
22.
23. 24. 25.
26.
27.
Engineering, Energy and Electrical Drives (POWERENG), pp 96–99, Portugal Bayindir R, Sefa ˙I (2007) Novel approach based on microcontroller to online protection of induction motors. Energy Convers Manag 48(3):850–856 Bayindir R, Sefa ˙I, C¸olak ˙I, Bektas A (2008) Fault detection and protection of induction motors using sensors. IEEE Trans Energy Convers 23(3):734–741 ¨ ztu¨rk A (2000) Protection of AC motors C¸unkas¸ M, Akkaya R, O by means of microcontrollers. 10th Mediterranean Electrotechnical Conference (MELECON), vol 3, pp 1093–1096 Uyar O (2010) Design and implementation of fuzzy logic based protection system for induction motors. MS thesis, Selcuk University, Turkey Pillay P, Manyage M (2001) Definitions of voltage unbalance. IEEE Power Eng Rev 21(5):10–12 Ross TJ (2004) Fuzzy logic with engineering applications. Wiley, London Uyar O, C¸unkas¸ M (2011) Design of fuzzy logic based motor protection system. 6th International Advanced Technologies Symposium (IATS’11), 16–18 May 2011, Turkey Zhang P, Du Y, Habetler TG, Lu B (2011) A survey of condition monitoring and protection methods for medium-voltage induction motors. IEEE Trans Ind Appl 47(1):34–46 Sin ML, Erugrul N (2003) Induction machine on-line condition monitoring and fault diagnosis—a survey. AUPEC2003, Australasian Universities Power Engineering Conference, Christchurch, New Zealand
