阅读说明：本技术 用于电气系统中的故障检测的负序电压的分段估计 (segmented estimation of negative sequence voltage for fault detection in electrical systems ) 是由 A·帕尼 S·K·沙玛 于 2018-05-03 设计创作，主要内容包括：本发明提供了一种被配置为检测包括多个定子绕组的电机中的定子绕组故障的诊断系统。该诊断系统包括处理器,处理器被编程为：接收提供给电机的三相电压和电流的测量值；根据三相电压和电流计算电压和电流的正序分量、负序分量和零序分量；以及通过执行包括修改的递归最小二乘(RLS)方法的两步初始化算法来识别对于负序电压的噪声因子贡献和定子故障贡献,噪声因子贡献包括由正序电流、负序电流和正序电压中的一者或多者引起的电机中的不平衡。处理器仍被进一步编程为基于对于负序电压的定子故障贡献来检测电机中的定子故障。(The present invention provides a diagnostic system configured to detect stator winding faults in an electric machine including a plurality of stator windings. The diagnostic system includes a processor programmed to: receiving measurements of three-phase voltages and currents supplied to the motor; calculating a positive sequence component, a negative sequence component and a zero sequence component of the voltage and the current according to the three-phase voltage and the current; and identifying a noise factor contribution to the negative sequence voltage and a stator fault contribution by performing a two-step initialization algorithm comprising a modified Recursive Least Squares (RLS) method, the noise factor contribution comprising an imbalance in the electric machine caused by one or more of the positive sequence current, the negative sequence current, and the positive sequence voltage. The processor is still further programmed to detect a stator fault in the electric machine based on the stator fault contribution to the negative sequence voltage.)

1. A diagnostic system configured to detect stator winding faults in an electric machine including a plurality of stator windings, the diagnostic system comprising:

A processor programmed to:

Receiving measurements of three-phase voltages and currents provided to the motor, the measurements received from voltage and current sensors associated with the motor;

Calculating a positive sequence component, a negative sequence component and a zero sequence component of the voltage and the current according to the three-phase voltage and the current;

Identifying a noise factor contribution to the negative sequence voltage and a stator fault contribution, the noise factor contribution comprising an imbalance in the electric machine caused by one or more of a positive sequence current, a negative sequence current, and a positive sequence voltage; and

detecting a stator fault in the electric machine based on the stator fault contribution to the negative sequence voltage;

Wherein in identifying the noise factor contribution and stator fault contribution to the negative sequence voltage, the processor is further programmed to execute a two-step initialization algorithm comprising a modified Recursive Least Squares (RLS) method to identify the noise factor contribution.

2. The diagnostic system of claim 1, wherein in detecting the stator fault in the electric machine, the processor is further programmed to:

Determining an amount of voltage gain in the stator windings based on the stator fault contribution to the negative sequence voltage;

Comparing the voltage gain to a voltage threshold; and

Identifying a stator fault in the AC machine if the voltage gain is greater than the voltage threshold;

Otherwise, it is determined that a stator fault is not present in the AC electric machine.

3. The diagnostic system of claim 2, wherein the processor is further programmed to identify the stator winding fault in the electric machine and localize the stator fault to one or more phases of the AC electric machine according to the formula:

V+aV+aV＝-3V，

where VSFa, VSFb, VSFc are the voltage gains in phase a, phase B, phase C indicative of stator fault, and V2SF is the stator fault contribution to the negative sequence voltage.

4. The diagnostic system of claim 1, wherein the processor is further programmed to calculate the negative sequence voltage according to the following equation:

V＝K*I+K*V+K*I，

Wherein I1 is the positive sequence current, I2 is the negative sequence current, V1 is the positive sequence voltage, and K1, K2, and K3 are coefficients of a function of the positive sequence current I1.

5. the diagnostic system of claim 4, wherein in executing the two-step initialization algorithm, the processor is further programmed to:

Estimating the coefficients K1, K2 for constant load and setting the coefficient K3 to zero;

setting the constant load to a base load;

calculating an error between the actual negative sequence voltage and the estimated negative sequence voltage at a plurality of different loads by the estimated coefficients K1, K2 at the base load and the coefficient K3 at zero; and

Estimating the coefficient K3 by optimizing the error between the actual negative sequence voltage and the estimated negative sequence voltage.

6. the diagnostic system of claim 5, wherein the processor is further programmed to estimate the coefficient K3 by optimizing the error according to the formula:

Error＝K*(I–I)，

wherein I1 is the positive sequence current under current load conditions and I1_ base is the positive sequence current under the base load conditions.

7. The diagnostic system of claim 5, wherein the processor is further programmed to estimate the coefficients K1, K2 for a constant load during an initialization period, wherein there is no stator fault contribution to the negative sequence voltage during the initialization period.

8. the diagnostic system of claim 1, wherein the processor is further programmed to determine fundamental components of the three-phase voltages and currents provided to the AC electric machine, and wherein positive, negative, and zero sequence components of the voltages and currents are determined from the fundamental components.

9. a method for identifying a turn-to-turn stator fault in an electric machine including a plurality of stator windings, the method comprising:

measuring three-phase voltages and currents supplied to terminals of the motor by voltage and current sensors;

cause a diagnostic system to identify an inter-turn stator fault in the stator winding of the electric machine, wherein causing the diagnostic system to identify the inter-turn stator fault comprises:

Receiving the measured three-phase voltages and currents provided to the terminals of the motor;

calculating positive sequence voltage and current, negative sequence voltage and current, and zero sequence voltage and current from the measured three-phase voltages and currents;

compensating for noise factors in the negative sequence voltage to isolate a stator fault negative sequence voltage; and

identifying a turn-to-turn stator fault in the electric machine based on the stator fault negative sequence voltage;

Wherein compensating for the noise factor in the negative sequence voltage comprises performing a modified Recursive Least Squares (RLS) estimation to estimate a noise factor contribution to the negative sequence voltage.

10. The method of claim 9, wherein the negative sequence voltage is calculated according to the following equation:

V＝K*I+K*V+K*I，

Wherein I1 is the positive sequence current, I2 is the negative sequence current, V1 is the positive sequence voltage, and K1, K2, and K3 are first, second, and third coefficients, respectively, as a function of the positive sequence current I1.

11. the method of claim 10, further comprising:

Estimating the coefficients K1, K2 for constant load and setting the coefficient K3 to zero;

Setting the constant load to the base load;

Calculating an error between the actual negative sequence voltage and the estimated negative sequence voltage at a plurality of different loads by the estimated coefficients K1, K2 at the base load and the coefficient K3 at zero; and

estimating the coefficient K3 by optimizing the error between the actual negative sequence voltage and the estimated negative sequence voltage.

12. the method of claim 11, wherein the third coefficient K3 is estimated by optimizing the error according to the following equation:

Error＝K*(I–I)，

wherein I1 is the positive sequence current under current load conditions and I1_ base is the positive sequence current under the base load conditions.

13. The method of claim 9, wherein identifying the inter-turn stator fault comprises calculating a Fault Severity Index (FSI), the FSI comprising a voltage phasor having a magnitude indicative of an amount of voltage gain or voltage drop at the terminals of the electric machine and an angle indicative of the one or more phases for which the voltage gain or voltage drop is present.

14. the method of claim 13, further comprising localizing the stator fault to one or more phases of the electric machine according to the formula:

V+aV+aV＝-3V，

where VSFa, VSFb, VSFc are the voltage gains in phase a, phase B, phase C indicative of the stator fault, and V2SF is the stator fault negative sequence voltage.

15. the method of claim 9, wherein upon identifying the inter-turn stator fault in the electric machine, the method further comprises:

Determining an amount of voltage gain in the stator windings based on the stator fault negative sequence voltage;

Comparing the voltage gain to a voltage threshold; and

identifying an inter-turn stator fault in the electric machine if the voltage gain is greater than the voltage threshold;

Otherwise, determining that no inter-turn stator fault exists in the motor.

background

The present invention relates generally to electrical systems, electrical machines, and electric motors, and more particularly to systems and methods for performing segment estimation of negative sequence voltage to detect stator winding faults in such electrical machines.

In industrial facilities, electrical machines, such as generators, AC motors and/or transformers, are used in a variety of applications. As one example, induction motors are used in applications such as pumping, cooling, material movement, and other applications requiring cost effective and robust motors. In such applications, power distribution systems are used in conjunction with electrical machines, where the power distribution system includes protection and control components (such as circuit breakers, contactors, starters, etc.).

In three-phase systems having a power distribution system and an AC motor, it has been recognized that various factors can cause three-phase voltage asymmetries/imbalances in the system. That is, both high resistance connections and stator winding faults (stator winding faults) change the resistance in the system, thereby resulting in three phase voltage asymmetries/imbalances. With respect to stator winding faults in a system, these faults may be due to gradual deterioration of the winding insulation due to a combination of: vibration caused by electromechanical forces, high dv/dt voltage surges, thermal overloads and/or contamination. If a stator winding fault occurs and the windings of the stator are shorted, large circulating fault currents are induced in the shorted turns, resulting in a local thermal overload. If not detected, such localized thermal overloads may eventually cause the motor to fail for a short period of time due to a ground fault/phase-to-phase insulation or open circuit fault. Accordingly, it is desirable to detect stator winding faults in an efficient and cost effective manner.

It has been recognized that some existing sensorless techniques developed for detecting stator winding faults rely on a primary method/concept of monitoring the effects of asymmetric "variations" of a three-phase system, as faults occur in one of the phases. Parameters (such as three-phase voltages and currents supplied to the motor in the circuit, positive, negative and zero sequence components of the voltages and currents, voltage imbalances at the motor terminals, and current imbalances at the motor terminals) can be acquired/analyzed to identify stator winding faults in a three-phase motor circuit where turn-to-turn faults result in identifiable negative sequence voltage variations.

however, with existing sensorless techniques for detecting stator winding faults that utilize the negative sequence parameters (i.e., the negative sequence voltage) of the motor, it has been recognized that various factors, commonly referred to as "noise factors," can cause changes in the negative sequence voltage in addition to the inter-turn fault itself. Thus, the accuracy and reliability of turn-to-turn fault detection will depend at least in part on determining the noise factor contribution to the negative sequence voltage, where the estimated noise factor contribution is subtracted from the total negative sequence voltage to identify the negative sequence voltage due to the turn-to-turn fault.

in existing stator winding fault detection techniques, this noise factor contribution has been performed using linear optimization techniques, such as Least Mean Square (LMS) and Recursive Least Squares (RLS) techniques. However, it has been recognized that negative sequence parameters have a non-linear dependence on load and supply voltage, and therefore linear estimation techniques have poor accuracy in the presence of load variations. This poor accuracy of the noise factor contribution estimation limits the minimum severity of the detectable inter-turn faults and may therefore lead to incorrect and/or missed identification of the occurrence of a fault. While attempts have been made to develop techniques for more accurately estimating noise factor contribution-including the loadbox method, the neural network method, and the use of complex non-linear equations to estimate noise factor contribution-such methods are computationally intensive, difficult to implement, and costly.

it is therefore desirable to provide systems and methods that can accurately estimate the noise factor contribution to negative sequence voltage variations when detecting stator winding faults. It is also desirable to implement such systems and methods in a simple and cost effective manner.

Brief description of the invention

according to one aspect of the invention, a diagnostic system is configured to detect a stator winding fault in an electric machine including a plurality of stator windings. The diagnostic system includes a processor programmed to receive measurements of three-phase voltages and currents provided to the motor, the measurements being received from voltage and current sensors associated with the motor. The processor is further programmed to: the method includes calculating positive, negative, and zero sequence components of the voltages and currents from the three-phase voltages and currents, and identifying a noise factor contribution to the negative sequence voltage that includes an imbalance in the electric machine caused by one or more of the positive, negative, and positive sequence currents and stator fault contributions. The processor is still further programmed to detect a stator fault in the electric machine based on the stator fault contribution to the negative sequence voltage. In identifying the noise factor contribution to the negative sequence voltage and the stator fault contribution, the processor is further programmed to execute a two-step initialization algorithm that includes a modified Recursive Least Squares (RLS) method to identify the noise factor contribution.

According to another aspect of the invention, an electrical system includes an input connectable to an AC power source and an output connectable to terminals of an electric machine to provide three-phase power thereto, the electric machine including a plurality of stator windings. The electrical system also includes a diagnostic system configured to identify stator faults in the stator windings of the electric machine, the diagnostic system including a processor programmed to receive three-phase supply voltage and current measurements provided to the electric machine by voltage and current sensors connected to the power distribution circuit between the input and the output. The processor is further programmed to calculate positive, negative, and zero sequence components of the supply voltage and current, compensate for noise factors in the negative sequence voltage to isolate the stator fault negative sequence voltage, the noise factors including an imbalance in the electric machine caused by one or more of the positive, negative, and positive sequence currents. The processor is still further programmed to identify a stator fault in the electrical distribution circuit based on the stator fault negative sequence voltage. To compensate for noise factors in the negative sequence voltage, the processor is further programmed to estimate first and second coefficients at a base load value as a function of the positive sequence current, and when a third coefficient as a function of the positive sequence current is at a zero value, calculate an error between the actual negative sequence voltage and the estimated negative sequence voltage at a plurality of load values using the estimated first and second coefficients and the zero value of the third coefficient, and estimate the third coefficient by optimizing the error.

According to yet another aspect of the present invention, a method for identifying a turn-to-turn stator fault in an electric machine including a plurality of stator windings is provided. The method comprises the following steps: measuring, by voltage and current sensors, three-phase voltages and currents provided to terminals of the electric machine, and causing a diagnostic system to identify an inter-turn stator fault in a stator winding of the electric machine, wherein causing the diagnostic system to identify the inter-turn stator fault further comprises: the method includes receiving measured three-phase voltages and currents provided to terminals of the electric machine, calculating positive sequence voltages and currents, negative sequence voltages and currents, and zero sequence voltages and currents from the measured three-phase voltages and currents, compensating for noise factors in the negative sequence voltages to isolate stator fault negative sequence voltages, and identifying inter-turn stator faults in the electric machine based on the stator fault negative sequence voltages. To compensate for noise factors in the negative sequence voltage, the method further includes performing a modified Recursive Least Squares (RLS) estimation to estimate a noise factor contribution to the negative sequence voltage.

various other features and advantages of the present invention will become apparent from the following detailed description and the accompanying drawings.

Drawings

the drawings illustrate preferred embodiments presently contemplated for carrying out the invention.

In the drawings:

Figure 1 is a schematic diagram of a three-phase power distribution circuit for use with embodiments of the present invention.

Fig. 2 is a schematic diagram of a motor circuit in which there is a stator winding fault according to an embodiment of the present invention.

Fig. 3 is a flow diagram illustrating a technique for detecting stator winding faults in an electric machine according to an embodiment of the invention.

FIG. 4 is a flow diagram illustrating a technique for performing a piece-wise estimation of a negative sequence voltage according to an embodiment of the invention.

FIG. 5 is a graph illustrating a comparison of the performance of the segment estimation technique of FIG. 4 compared to a prior art estimation technique for estimating negative sequence voltage during a low severity turn-to-turn stator fault.

FIG. 6 is a flow chart illustrating the error of the estimated negative sequence voltage using each of the piecewise estimation techniques of FIG. 4.

Detailed Description

Embodiments of the invention set forth herein relate to systems and methods for performing segment estimation of negative sequence voltage to detect stator winding faults in electric machines. Systems and methods estimate a non-linear contribution of a noise factor to an overall change in a negative sequence parameter of an electric machine to accurately identify a contribution of a stator fault to a negative sequence voltage change.

referring to fig. 1, a three-phase power distribution circuit 10 is shown according to one embodiment of the present invention. The distribution circuit 10 is connected between three-phase AC inputs 12a-12c and a load 14, such as an AC motor, to provide protection to the motor and condition power from the three-phase AC inputs 12a-12c for delivery to the motor. According to one embodiment of the invention, the electric machine 14 is in the form of an induction motor 14, and is therefore referred to hereinafter in fig. 1 as induction motor 14. However, it is recognized that the electric machine 14 may also be, for example, a generator or a transformer, or any other load that may be driven by three-phase power and is useful in an industrial environment. Thus, embodiments of the invention should not be limited to a particular type of motor or motor.

The power distribution circuit 10 includes an input 16 connectable to the three-phase AC inputs 12a-12c to receive power therefrom. The power distribution circuit 10 also includes a three-phase output 18 connectable to motor terminals 20 of the induction motor to provide three-phase voltages and currents to the induction motor 14. According to one embodiment, output 18 may be connected to motor terminal 20, for example, at a junction box 22 of power distribution circuit 10.

As further shown in fig. 1, the power distribution circuit 10 includes therein a plurality of circuit components positioned between the input 16 and the output 18, wherein the circuit components provide protection from and control of the voltage and current provided from the three-phase AC inputs 12a-12c for delivery to the induction motor 14. Many such protection and control components are shown in fig. 1, but it has been recognized that other components may also/alternatively be included in the power distribution circuit 10 in accordance with embodiments of the present invention. In the embodiment of the power distribution circuit 10 shown in fig. 1, a Motor Control Center (MCC)24 is shown as being included in the circuit. The motor control center 24 may include an assembly of one or more enclosed sections having a common power bus and containing a number of motor control units, such as a number of motor starters. The motor control center 24 may also include variable frequency drives, programmable controllers, and meters. Associated with the operation of the motor control center 24 are a number of protection components/devices for protecting the motor 14, providing short circuit protection, and/or isolating the motor circuit. For example, fuses 26 and contactors 28 are provided in the power distribution circuit 10 (such as in a local distribution panel 30) to provide short circuit protection and control of the induction motors 14. A circuit breaker 32 and disconnect switch 34 are also provided to provide short circuit protection and isolation of the power distribution circuit 10.

it has been recognized that the power distribution circuit 10 shown in fig. 1 illustrates only a motor power distribution circuit that may be associated with embodiments of the present invention, and that various configurations and arrangements of three-phase AC circuits may alternatively be provided in accordance with embodiments of the present invention.

With respect to the power distribution circuit 10 shown in fig. 1, it is desirable to be able to detect stator winding faults in the induction motor 14. Stator winding faults may be due to gradual deterioration of stator winding insulation due to a combination of electromechanical force induced vibration, high dv/dt voltage surges, thermal overloads, and/or contamination. If a stator winding fault occurs and the windings of the stator are short circuited, a large circulating fault current is induced in the short-circuited windings, resulting in a local thermal overload. If not detected, such localized thermal overloads may eventually cause the motor to fail for a short period of time due to a ground fault/phase-to-phase insulation or open circuit fault.

To provide for detection of such stator winding faults, a diagnostic system 40 is included in the electrical distribution circuit 10 in accordance with an embodiment of the present invention. The diagnostic system 40 receives inputs regarding the three-phase supply voltage and current provided to the induction motor 14. According to an exemplary embodiment, the diagnostic system 40 receives voltage and current measurements taken from voltage and current sensors (generally indicated at 41) integrated into the motor starter in the MCC 24; however, it has been recognized that separate dedicated voltage and current sensors may be included in power distribution circuit 10 to acquire voltage and current data from a location between input 16 and output 18 and provide it to diagnostic system 40. As shown in fig. 1, a processor 42 in the diagnostic system 40 receives the measured three-phase voltages and currents and is programmed to analyze the data to identify stator winding faults in the induction motor.

Although diagnostic system 40 is shown in fig. 1 as being in the form of a stand-alone product/device, it has been recognized that such a system may be incorporated into protection and control components included in power distribution circuit 10. I.e., the processor 42 having the program/algorithm thereon capable of detecting stator winding faults in the induction motor 14, may be located in an existing starter, relay, drive, circuit breaker, motor control center, and/or other motor control or protection product in the power distribution circuit 10. The diagnostic system 40 may thus provide online monitoring of the power distribution circuit 10 from the location of the power distribution circuit or at a location remote from the power distribution circuit.

Further, while the present inventors' embodiments are described below with respect to a processor 42 of the diagnostic system 40 being programmed to execute techniques for identifying stator winding faults in the induction motor 14, it should be appreciated that the term "processor" as used herein need not be a programmable device. That is, it should be understood that the processor 42 (and steps performed thereby) as described below also includes equivalent hardware and computing devices that perform the same tasks.

according to an embodiment of the present invention, in order to detect whether there is a voltage drop or a voltage gain in a three-phase motor circuit, a method of symmetrical components is employed in order to simplify the analysis of the motor circuit when the motor circuit becomes unbalanced. The asymmetric/unbalanced phasors (voltage and current) are represented as three symmetrical sets of balanced phasors — the first set has the same phase sequence as the system under study (positive sequence, e.g., ABC), the second set has the opposite phase sequence (negative sequence, e.g., ACB), and in the third set, the phasors A, B and C are in phase with each other (zero sequence). In essence, this method converts the three unbalanced phases into three independent sources, which makes asymmetric fault analysis easier to handle. Using the sequence phasors of the voltage and current, a Fault Severity Index (FSI) is calculated, the magnitude of which is an indicator of the amount of voltage gain or voltage drop in the circuit, and the angle of which indicates the phase or phases having voltage gain/voltage drop.

embodiments of the present invention are provided for detecting the presence of a resistance-based fault in an AC electric machine (included in a delta-connection based motor circuit or a star-connection based motor circuit). A schematic representation of a generic induction motor circuit 44 is provided in fig. 2. In fig. 2, the supply line voltages Va, Vb, Vc are shown, i.e. as phase voltages VaM, VbM, VcM at the motor terminals present at the motor. The motor terminal voltages present are determined in part by any stator winding faults present in the stator windings 46 (on one or more phases) of the motor, which are indicated as VSFa, VSFb, VSFc.

Referring now to fig. 3, and with continued reference to fig. 1 and 2, a technique 50 implemented by the processor 42 of the diagnostic system 40 to identify, locate and quantify stator winding faults in the delta or wye connected power distribution (motor) circuit 10 is shown, in accordance with an embodiment of the present invention. It has been recognized that the technique 50 for identifying, locating and quantifying stator winding faults (and the steps included therein) is the same for delta-connected motors and for star-connected motors.

In the first step of the technique 50, the processor 42 receives three-phase current and voltage measurements at step 52. According to an exemplary embodiment, the processor 42 receives three-phase current and voltage data as measured in the MCC (i.e., at the sensed location between the input 16 and the output 18), wherein the voltage and current from the MCC 24 is then supplied to the terminals 20 of the electric machine 14 (e.g., an induction motor). Upon receiving the three-phase current and voltage measurements, the processor 42 then extracts the fundamental components of the three-phase current and voltage at step 54 according to known techniques/methods. Step 54 is shown in dashed lines in fig. 3 because it has been recognized that the determination of the fundamental component is optional for performing the technique 50-because the detection, localization, and quantification of stator winding faults in the distribution circuit 10 can be performed without the fundamental component. However, it should be appreciated that more accurate stator winding fault analysis may be achieved by determining and using the fundamental component, as described below.

as shown in fig. 3, the technique 50 continues at step 56 with calculating the sequence components of the fundamental supply voltage and current, where the sequence components of the voltage are identified as V1,2,0 and the sequence components of the current are identified as I1,2,0, where the positive, negative and zero sequence components are identified by 1,2 and 0, respectively. With respect to determining the voltage sequence components, it has been recognized that Kirchhoff's Voltage Law (KVL) may be applied to a connected power distribution circuit to describe the relationship between the supply line voltage, the phase voltages at the motor terminals, and the stator fault in the form according to the following equations:

Where VaM, VbM, VcM are the three-phase voltages on the motor windings, Va, Vb, Vc are the three-phase line voltages (measured at the inductive locations), and VSFa, VSFb, VSFc are the voltage gains due to stator faults in phase a, phase B, phase C, respectively.

the positive, negative and zero sequence component voltages may be obtained by applying the transformation T to equation 1. According to one embodiment, a sequence transformation T is utilized, defined according to:

Where (i.e., the unit vector at 120 degrees).

Applying the transformation T to equation 1 yields:

considering that the line voltage drops in each phase are equal (i.e., there is no additional imbalance in the stator windings), equation 2 can be rewritten as:

where V0M, V1M, and V2M are the zero, positive, and negative sequence voltage components at the motor terminals.

rearranging the negative sequence equation to isolate the negative sequence voltage V2M at the motor terminals yields:

equation 4 is then rewritten as:

Where V2 is the negative sequence voltage observed at the sensing location on the power supply side, and VSFa, VSFb, VSFc are the voltage gains observed when there is a stator fault in phase a, phase B, phase C, respectively.

Rearrangement equation 5 yields:

VSFa + a2VSFb + aVSFc ═ -3(V2-V2M) [ formula 6 ].

It has been recognized that the negative sequence voltage V2M at the motor terminals is a function of the negative sequence motor current, supply voltage, positive sequence current (load), and stator fault. Therefore, in order to detect stator faults, the negative sequence voltage must be compensated for negative sequence motor current, supply voltage and load (i.e., "noise factor"), with the recognition that the non-linearity of these parameters makes its estimation difficult, especially when attempting to perform such estimation using linear optimization techniques. Thus, according to an exemplary embodiment of the present invention, a segment estimation algorithm is executed at step 58 to determine the noise factor contribution and the stator fault contribution to the negative sequence parameters. As indicated in fig. 4, a number of sub-steps are performed at step 58 in order to accurately estimate the contribution of noise factors to the negative sequence voltage, thereby enabling these noise factors to be removed and the stator fault negative sequence voltage to be isolated. In one embodiment, and as explained in more detail below, the segment estimation algorithm includes a modified Recursive Least Squares (RLS) estimation that accounts for the non-linearity of the noise factor described above.

To estimate the negative sequence voltage at the motor terminals, the negative sequence voltage may be defined according to:

V2M ═ K1 ═ I2+ K2 ═ V1+ K3 ═ I1 [ formula 7],

Where I1 is the positive sequence current, I2 is the negative sequence current, V1 is the positive sequence voltage, and K1, K2, and K3 are values that are a function of the positive sequence current I1. It should also be understood that the negative sequence current I2 is a function of the positive sequence current I1, such as according to the following function:

I2(I1) ═ I2_ base + f (I1) [ formula 8 ].

Since equation 7 is a non-linear equation, estimating K1, K2, and K3 together using a linear estimation technique (e.g., RLS) does not give a correct estimate of the negative sequence voltage. Thus, when the segment estimation algorithm of step 58 is employed to estimate V2M, it is assumed that the motor is operating in a healthy condition (i.e., no stator winding faults in the motor windings) -i.e., an "initialization" phase. Initially assuming no stator winding fault, the negative sequence voltage (at the inductive location between input 16 and output 18) is equal to the negative sequence voltage at the motor terminals (i.e., V2-V2M). Thus, during initialization (assuming no stator winding faults are present), the segment estimation begins by performing a "first step" of a two-step optimization at steps 60 and 62 of the algorithm. At step 60, the coefficients K1 and K2 are estimated for constant load on the induction motor 14, and K3 is set to zero, so that equation 7 can be optimized according to:

V2M ═ K1 ═ I2+ K2 ═ V1+ K2 ═ I1 [ formula 9]

For example, the induction motor 14 may operate at a constant load of 20%. The load value is defined as a base value, as indicated by step 62.

the segment estimation algorithm 58 continues to perform the "second step" of the two-step optimization algorithm at steps 64 and 66. At step 64, the induction motor 14 is operated at its normal duty cycle and the error between the actual negative sequence voltage and the estimated negative sequence voltage is calculated by taking the values of K1 and K2 derived from the first step and setting K3 to zero. At step 66, K3 is then estimated by optimizing the error according to the following equation:

Error is K3 (I1-I1 _ base) [ formula 10],

where I1 is the positive sequence current under the current load condition and I1_ base is the positive sequence current under the base load condition.

Advantageously, using the segment estimation algorithm of fig. 4 provides a more accurate estimate of the non-linear negative sequence parameter for identifying turn-to-turn stator faults in the electrical system. Fig. 5 shows a performance comparison of a segment estimation technique compared to a prior art estimation technique (i.e., an RLS technique-used to estimate the negative sequence voltage during a low severity turn-to-turn stator fault). As can be seen therein, during the initialization period (indicated at 68), a constant load will be maintained — there are constant K1 and K2 values in equation 9. After the initialization period, the load is then changed (such as at every four plotted points) and the stator fault negative sequence voltage V2SF is monitored to identify the presence of a turn-to-turn stator fault. The stator fault negative sequence voltage data point falling below the specified fault threshold 70 and thus indicating a healthy motor is indicated at 72, while the stator fault negative sequence voltage data point above the specified fault threshold 70 and thus indicating a faulty motor is indicated at 74. By comparing the stator fault negative sequence voltage data points calculated using the segment estimation technique with the points obtained using the prior art RLS technique, it can be seen that the data points showing a healthy motor 72 and showing a faulty motor 74 vary greatly between the two techniques — thus demonstrating that the higher accuracy of the estimated stator fault negative sequence voltage data points provided by the segment estimation technique can greatly improve the ability to accurately and reliably detect inter-turn stator faults in an electrical system. This accuracy is further reflected in the graph of fig. 6, where it can be seen that the error in the calculated stator fault negative sequence voltage (indicated at 76) when calculated by the segment estimation algorithm is greatly reduced compared to the error in the calculated stator fault negative sequence voltage (indicated at 78) when calculated by known RLS techniques.

Referring now back to the technique 50 of fig. 3, after completing step 58 and the segment estimation algorithm performed therein to identify the noise factor and the contribution of the stator fault to the negative sequence parameters, equation 6 may be implemented to determine the negative sequence voltage V2M at the motor terminals. The negative sequence voltage V2SF due to stator winding fault only, i.e., the stator fault negative sequence voltage V2SF, may then be determined at step 79 according to the following equation:

v2SF ═ V2-V2M [ formula 11 ].

By substituting equation 11 into equation 6, equation 6 can be rewritten according to the following equation:

VSFa + a2VSFb + avscfc ═ -3V2SF [ equation 12 ].

Equation 12 can be used to derive conditions and relationships to quantify and localize the voltage gain due to stator winding faults in single or multiple phases of a delta connected motor.

the technique 50 then continues at step 80, where it is determined whether the magnitude of the stator fault negative sequence voltage V2SF is greater than a threshold voltage level, thereby enabling a determination of whether the stator fault negative sequence voltage is indicative of a stator fault in the electrical distribution circuit 10. It will be appreciated that the threshold voltage level may be set according to the severity of the alarm that will occur, and thus the threshold may be a predefined value in the program, or may be determined according to user settings or using user input. According to an exemplary embodiment, the voltage threshold is set to 100mV — such that a voltage drop greater than 100mV is classified as a stator fault-although it is recognized that the threshold may be a higher or lower value. If it is determined at step 80 that the stator fault negative sequence voltage V2SF is less than the threshold voltage (as indicated at 82), then it is determined that no stator fault exists in the system, as shown at step 84. The technique then loops back to step 52 where the processor 42 receives additional three-phase current and voltage measurements to continue monitoring for stator faults.

Conversely, if it is determined at step 80 that the magnitude of the stator fault negative sequence voltage V2SF is greater than the threshold voltage (as indicated at 86), the technique 50 continues with step 88, where a positioning reference phase angle is calculated for each phase in the distribution circuit 10/motor 14. According to an exemplary embodiment, the positioning reference phase angle calculated at step 88 is derived in part by using the phase angle of the fundamental current flowing through each respective phase. Thus, for phase a, the reference phase angle can be located by the following formula:

Where is the angle of the fundamental component of the phase current flowing through phase a.

For phase B, the reference phase angle may be located as described by the following equation:

where is the angle of the fundamental component of the phase current flowing through phase B.

for phase C, the positioning reference phase angle can be described by the following equation:

where is the angle of the fundamental component of the phase current flowing through phase C.

After the positioning reference phase angle is determined at step 88, calculation of the voltage gain due to stator winding faults is next performed at step 90. At step 90, the calculation of voltage gain due to stator winding faults may alternatively be described as the calculation of FSI — where FSI is a phasor whose magnitude is an indicator of the amount of voltage gain due to a stator winding fault and whose angle indicates the presence of one or more phases of voltage gain due to a stator winding fault. The amount of voltage gain relative to the magnitude of the FSI phasor is derived from the stator fault negative sequence voltage V2SF, as described in equation 11-where the stator fault negative sequence voltage is phase separated, as described in equation 12. The phase or phases to which the stator winding fault (and accompanying voltage gain) is to be attributed (i.e., the location of the stator winding fault) are determined by comparing the phase angle of the stator fault negative sequence voltage to the location reference phase angle for each phase, relative to the angle of the FSI phasor.

after quantifying and locating the voltage gain due to stator winding faults at step 90, a condition check for stator turn faults is performed at step 92. When performing the check, the voltage at the motor terminals of each of phase a, phase B and phase C due to stator failure alone is described as:

V＝V+V

V＝a*V+a*V

VcSF a V1+ a2V 2SF [ equation 16],

Where V1 is the positive sequence voltage observed at the sensing location on the power supply side and V2SF is the stator fault negative sequence voltage.

It is then checked at step 92 whether, for the phase identified as having a stator winding fault thereon, the magnitude of the voltage identified as being due to stator fault on the corresponding motor terminal of that phase only is greater than the voltages on the other motor terminals due to stator fault only.

For stator winding failure in phase a:

| VaSF | > | VbSF |, | VcSF | [ equation 17 ].

for stator winding failure in phase B:

| VbSF | > | VaSF |, | VcSF | [ equation 18 ].

For stator winding failure in phase C:

| VcSF | > | VaSF |, | VbSF | [ equation 19 ].

a condition check at step 92 may be performed to verify the presence of a stator winding fault for a particular phase.

Advantageously, embodiments of the present invention thus provide systems and methods for detecting turn-to-turn stator winding faults in a three-phase power distribution circuit by using three-phase voltages and currents provided to an electric machine (e.g., an AC motor). Since inter-turn stator winding faults in a three-phase motor circuit can result in voltage imbalances (i.e., voltage gains at one or more terminals) and consequent current imbalances at the motor terminals, stator winding faults can be detected by analyzing and processing the measured three-phase voltages and currents. To identify turn-to-turn stator faults, a piecewise estimation algorithm is employed that provides a more accurate estimate of the non-linear negative sequence parameters (as compared to linear optimization techniques such as LMS and RLS). The piecewise estimation algorithm is less computationally intensive than other non-linear estimation techniques that have been employed previously, making it possible to implement in a simple and cost-effective manner.

a technical contribution to the disclosed method and apparatus is that it provides a processor-implemented technique for detecting inter-turn stator faults in a three-phase AC motor circuit, in which a segment estimation technique is employed to calculate negative sequence parameters in the circuit that are necessary to accurately identify such stator faults.

thus, according to one embodiment of the invention, a diagnostic system is configured to detect a stator winding fault in an electric machine comprising a plurality of stator windings. The diagnostic system includes a processor programmed to receive measurements of three-phase voltages and currents provided to the motor, the measurements being received from voltage and current sensors associated with the motor. The processor is further programmed to: the method includes calculating positive, negative, and zero sequence components of the voltages and currents from the three-phase voltages and currents, and identifying a noise factor contribution to the negative sequence voltage that includes an imbalance in the electric machine caused by one or more of the positive, negative, and positive sequence currents and stator fault contributions. The processor is still further programmed to detect a stator fault in the electric machine based on the stator fault contribution to the negative sequence voltage. In identifying the noise factor contribution to the negative sequence voltage and the stator fault contribution, the processor is further programmed to execute a two-step initialization algorithm that includes a modified Recursive Least Squares (RLS) method to identify the noise factor contribution.

According to another embodiment of the invention, an electrical system includes an input connectable to an AC power source and an output connectable to terminals of an electric machine to provide three-phase power thereto, the electric machine including a plurality of stator windings. The electrical system also includes a diagnostic system configured to identify stator faults in the stator windings of the electric machine, the diagnostic system including a processor programmed to receive three-phase supply voltage and current measurements provided to the electric machine by voltage and current sensors connected to the power distribution circuit between the input and the output. The processor is further programmed to calculate positive, negative, and zero sequence components of the supply voltage and current, compensate for noise factors in the negative sequence voltage to isolate the stator fault negative sequence voltage, the noise factors including an imbalance in the electric machine caused by one or more of the positive, negative, and positive sequence currents. The processor is still further programmed to identify a stator fault in the electrical distribution circuit based on the stator fault negative sequence voltage. To compensate for noise factors in the negative sequence voltage, the processor is further programmed to estimate first and second coefficients at a base load value as a function of the positive sequence current, and when a third coefficient as a function of the positive sequence current is at a zero value, calculate an error between the actual negative sequence voltage and the estimated negative sequence voltage at a plurality of load values using the estimated first and second coefficients and the zero value of the third coefficient, and estimate the third coefficient by optimizing the error.

according to yet another embodiment of the present invention, a method for identifying a turn-to-turn stator fault in an electric machine including a plurality of stator windings is provided. The method comprises the following steps: measuring, by voltage and current sensors, three-phase voltages and currents provided to terminals of the electric machine, and causing a diagnostic system to identify an inter-turn stator fault in a stator winding of the electric machine, wherein causing the diagnostic system to identify the inter-turn stator fault further comprises: the method includes receiving measured three-phase voltages and currents provided to terminals of the electric machine, calculating positive sequence voltages and currents, negative sequence voltages and currents, and zero sequence voltages and currents from the measured three-phase voltages and currents, compensating for noise factors in the negative sequence voltages to isolate stator fault negative sequence voltages, and identifying inter-turn stator faults in the electric machine based on the stator fault negative sequence voltages. To compensate for noise factors in the negative sequence voltage, the method further includes performing a modified Recursive Least Squares (RLS) estimation to estimate a noise factor contribution to the negative sequence voltage.

The present invention has been described in terms of the preferred embodiment, and it is recognized that equivalents, alternatives, and modifications, aside from those expressly stated, are possible and within the scope of the appending claims.