Open-circuit fault diagnosis method for five-phase motor

Abstract

The invention discloses a five-phase motor open-circuit fault diagnosis method, and belongs to the field of motor online diagnosis. According to the method, in a model training stage, time sequence motor parameters of a motor under different working conditions are collected through a down-sampling technology, and a training database is constructed; and analyzing the relationship between the motor parameters and the winding current by using a Pearson correlation coefficient, and performing regression prediction on the motor state by using a BP neural network model. In the on-line diagnosis stage, on-line training and current residual comparison are carried out on the basis of the digital twin virtual motor obtained through training in combination with electrical angle, torque and phase voltage data of a physical motor collected in real time, and whether the motor breaks down or not is detected. And accurately positioning the fault position by analyzing the differential value of the current residual error and the diagnosis variable, and judging whether the fault type is a winding open-circuit fault or a power tube open-circuit fault. According to the method, accurate design parameters and fault data of the motor are not needed, the open-circuit fault of the motor can be quickly diagnosed in real time only through normal operation data, and the robustness is very high.

Description

Technical Field

[0001] The present invention relates to the field of online diagnosis of motors, and in particular to a method for diagnosing open-circuit faults of five-phase motors. Background Art

[0002] In recent years, with the rapid rise of the low-altitude economy, the application of drone technology in logistics and transportation, agricultural monitoring and environmental protection has become increasingly widespread. As a core component of the low-altitude economy, the performance and reliability of drones directly affect the efficiency and safety of related industries. In the motor drive system of drones, online fault diagnosis of motors is particularly important, especially the open circuit fault of the winding may cause the drone to lose power or control, and then cause serious accidents. At present, most of the existing online diagnosis methods rely on regular inspections and manual maintenance, which cannot achieve real-time monitoring of the motor status and lack timely emergency response mechanisms. Therefore, there is an urgent need for an innovative online diagnosis technology for open circuit faults in motor drive systems to improve the safety and reliability of drones, thereby providing guarantees for the sustainable development of the low-altitude economy.

[0003] Motor drive system faults usually include winding faults and power tube faults. Existing motor online diagnosis methods can be divided into signal processing methods, analytical model methods and data-driven methods according to the selected characteristic values. However, these methods have their own shortcomings in practical applications. Signal processing methods rely on extracting fault features from complex signals, but their diagnostic accuracy is low in high-noise environments, and their real-time performance is poor due to the large amount of calculation. The analytical model method requires an accurate motor mathematical model, but in practical applications, due to the inevitable differences between the motor mathematical model and the actual working conditions, its adaptability is poor, especially in multi-parameter fault diagnosis, it is often difficult to deal with complex fault situations, and the modeling process is complex and requires high professional knowledge. Although the data-driven method can establish a diagnostic database based on a large amount of fault data, it also requires high computing resources and is often limited by the existing data acquisition and processing capabilities. Therefore, the existing online diagnosis methods still have room for improvement in accuracy, adaptability and real-time performance, and need to be further optimized to meet the diversified and high-real-time online diagnosis needs of motor systems. Summary of the invention

[0004] The purpose of the present invention is to overcome the above-mentioned defects existing in the background technology and provide a five-phase motor open circuit fault diagnosis method, which does not require the motor parameters at the time of the fault, does not require the addition of additional hardware circuits, has a simple diagnostic algorithm, a fast diagnostic speed, and a strong ability to prevent misdiagnosis, and can quickly and accurately diagnose the motor's winding open circuit fault and power tube open circuit fault.

[0005] The present invention adopts the following technical solutions to solve the above technical problems:

[0006] A five-phase motor open circuit fault diagnosis method, characterized in that the diagnosis method includes a model training stage and an online diagnosis stage, and the specific steps are as follows:

[0007] Step S1: In the model training stage, sequential motor parameters of the five-phase motor when it is operating normally under different working conditions are collected by downsampling technology to build a training database;

[0008] Step S2: In the model training stage, the Pearson correlation coefficient is used to sort the importance of the collected five-phase motor parameters, and the electrical angle θ(t), torque T at the same time are selected according to the importance. e (t) and phase voltage u dc (t) as input data of BP neural network regression model;

[0009] Step S3: In the model training stage, an online training method is adopted, using the normal operating electrical angle, torque and phase voltage data in the training database, combined with the corresponding winding current data as output, and a preliminary BP neural network model is obtained through training;

[0010] Step S4: In the online diagnosis stage, the BP neural network model and its parameters obtained by training in step S3 are saved and used as the digital twin virtual motor; the electrical angle, torque and phase voltage of the physical motor at the same time are collected in real time through downsampling technology, and these data are input into the digital twin virtual motor. After online training, the winding current i of the virtual motor is updated. pn , n represents the five-phase motor A, B, C, D, E phases, and updates the parameters of the digital twin virtual motor in real time;

[0011] Step S5: In the online diagnosis stage, the actual winding current i of the physical motor is rn and the digital twin virtual motor winding current i pn Compare and calculate the current residual i error_n ;

[0012] Step S6: In the online diagnosis stage, by analyzing the differential value Δi of the current residual, it is determined whether the motor has a winding open circuit fault or a power tube open circuit fault;

[0013] Step S7: In the online diagnosis stage, record the sampled current value i of 0.5 current fundamental wave period after the fault moment. 1 and the sampling current value i of 0.75 current fundamental wave cycle 2 , add the two current values ​​and calculate the diagnostic variable K n Perform fault location.

[0014] Furthermore, when the step S1 uses downsampling technology to collect motor parameter data to build a database, the collected data is preprocessed, null values ​​in the database are deleted, and the Z-score method is used to remove obvious abnormal values ​​in the database.

[0015] Furthermore, in step S2, the correlation between various motor parameters and winding current is evaluated by calculating the Pearson correlation coefficient, which specifically includes the following steps:

[0016] Step S2-1, randomly selecting a subset from the database as a sample set, wherein various motor parameters in the subset are used as an input set, and the corresponding winding current is used as an output set;

[0017] Step S2-2: Calculate the correlation coefficient between each motor parameter and the winding current using the Pearson correlation coefficient formula, the formula is as follows:

[0018]

[0019] Among them, Q i is the motor parameter sample, P i is the winding current sample, and are the mean values ​​of the corresponding samples, r Q,P The value range of is [-1,1];

[0020] Step S2-3, judging the influence of each motor parameter on the winding current by the absolute value of the correlation coefficient. The higher the correlation, the more significant the influence of the parameter on the winding current.

[0021] Step S2-4, sorting by the calculated correlation coefficients to screen out key motor parameters that have a significant impact on the winding current.

[0022] Furthermore, the BP neural network model in step S2 is a neural network with a multi-hidden layer structure, wherein the three layers are respectively an input layer, a hidden layer and an output layer, wherein the input layer has three input ports, a is the number of data of each motor parameter in the database, and the input layer ports are respectively electrical angle θ(t), torque T e (t) and phase voltage u dc (t); The output layer has one output port, and the output is the digital twin virtual motor winding current i pn ; The number of neurons in its hidden layer is M, and the number of hidden layers is N. The specific values ​​of N and M are adjusted according to actual working conditions.

[0023] Furthermore, in step S3, the BP neural network model is trained by online training, and the feature quantity x input at time t n for:

[0024] xn =[θ(t),T e (t),u dc (t)]

[0025] where x n The electrical angle θ and torque T in the database e , phase voltage u dc and the data set of the motor winding current at the previous moment; θ(t) represents the electrical angle of the motor at time t in the database; T e (t) represents the electrical angle of the motor at time t, u dc (t) represents the phase voltage of the motor at time t; the characteristic quantities are input into the BP neural network model in chronological order for model training.

[0026] Furthermore, in step S4, the trained BP neural network model is used as the digital twin virtual motor, and the downsampling technology is used to collect the electrical angle, torque and phase voltage data of the physical entity motor at the same time when it is running in real time. The collected real-time data is input into the digital twin virtual motor model in the form of streaming data to obtain the digital twin virtual motor winding current i pn And process historical data through the sliding window method. The model only retains the data points of the most recent fundamental cycle at each time step as the real-time training set for online training.

[0027] Furthermore, the current residual i of the motor winding current at time t in step S5 is error_n for:

[0028] i error_n (t) = i rn (t)-i pn (t)

[0029] where i rn (t) is the actual current of the physical motor winding at time t, i pn (t) is the digital twin virtual motor winding current at time t, and the differential value Δi of the motor winding current residual in step S6 is:

[0030]

[0031] where t n t n-1 the latter moment;

[0032] The judgment basis for online diagnosis is:

[0033]

[0034] where d 1 is the set online diagnosis threshold.

[0035] Furthermore, the diagnostic variable K in step S7 n The formula is:

[0036] K n =i 1 +i 2

[0037] where i 1 is the sampling current value of 0.5 current fundamental wave cycle after the fault, i 2 It is the sampling current value of 0.75 current fundamental wave cycle after the fault;

[0038] The basis for locating the fault is:

[0039]

[0040] Furthermore, the five-phase permanent magnet synchronous motor drive system is composed of a five-phase permanent magnet fault-tolerant motor and a set of five-phase half-bridge inverters. The power switch tubes T1, T2, T3, T4, T5, T6, T7, T8, T9, and T10 constitute the upper and lower bridge arms of the five-phase half-bridge inverter. The inverter bus terminal is U dc DC power supply, the five-phase permanent magnet fault-tolerant motor adopts a 10-slot 8-pole fractional slot double-layer concentrated winding structure, the electrical angle between the two-phase slots is 144°, the windings have no overlap, the rotor adopts a surface-mounted permanent magnet structure, and the centrifugal height is 5mm.

[0041] Compared with the prior art, the present invention adopts the above technical solution and has the following beneficial effects:

[0042] (1) The present invention provides a five-phase motor open circuit fault diagnosis method, which does not need to collect data on various motor faults. It only needs the normal operation data of the motor to diagnose the fault, which greatly reduces the difficulty of diagnostic data collection.

[0043] (2) The present invention provides a five-phase motor open circuit fault diagnosis method, which can obtain accurate motor winding current without knowing the precise parameters of the motor in advance.

[0044] (3) The present invention provides a five-phase motor open circuit fault diagnosis method, and the proposed downsampling technology can very accurately collect different parameter data of the motor at the same time, ensuring the time consistency of each parameter data and the validity of the training data.

[0045] (4) In a five-phase motor open circuit fault diagnosis method provided by the present invention, the BP neural network model is trained using health data during actual operation, so it can replicate the state of the target motion more accurately than the observer-based mathematical model.

[0046] (5) The present invention provides a five-phase motor open circuit fault diagnosis method, which uses a random forest model to extract the feature quantities required for training. It does not need to use all feature parameters as input, thereby reducing the model training time and ensuring the accuracy of the model.

[0047] (6) The present invention provides a five-phase motor open circuit fault diagnosis method, which has an extremely fast diagnosis speed and a simple diagnosis algorithm, is superior to general online diagnosis methods, and does not require the formulation of complex thresholds.

[0048] (7) The present invention provides a five-phase motor open circuit fault diagnosis method, which uses the differential value of the current residual as a standard for determining whether a fault has occurred. It has strong robustness and can effectively prevent misdiagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A flowchart of a five-phase motor open circuit fault diagnosis method according to the present invention;

[0050] Figure 2 A five-phase motor control block diagram of a five-phase motor open circuit fault diagnosis method according to the present invention;

[0051] Figure 3 This is a flow chart of a digital twin virtual motor and online diagnosis strategy for a five-phase motor open circuit fault diagnosis method of the present invention. DETAILED DESCRIPTION

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings of the specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work are within the scope of protection of the present invention.

[0053] Example:

[0054] like Figure 1 FIG. 1 is a flowchart of a five-phase motor open circuit fault diagnosis method according to the present invention, which specifically includes the following steps:

[0055] Step S1, in the model training stage, when the motor parameter data is collected by using the downsampling technology to build a database, the collected data is preprocessed, the null values ​​in the database are deleted, and the Z-score method is used to remove obvious outliers in the database;

[0056] Step S2: In the model training phase, the Pearson correlation coefficient is used to sort the importance of the motor operation data in the database, and the most important parameters are selected to train the BP neural network model to reduce the training cost and training time of the model. The random forest model consists of T decision trees, and the number of T is determined according to the actual working conditions, including the following steps:

[0057] Step S2-1, randomly selecting a subset from the database as a sample set, wherein various motor parameters in the subset are used as an input set, and the corresponding winding current is used as an output set;

[0058] Step S2-2: Calculate the correlation coefficient between each motor parameter and the winding current using the Pearson correlation coefficient formula, the formula is as follows:

[0059]

[0060] Among them, Q i is the motor parameter sample, P i is the winding current sample, Q and are the mean values ​​of the corresponding samples, r Q,P The value range is [-1,1].

[0061] Step S2-3, judging the influence of each motor parameter on the winding current by the absolute value of the correlation coefficient. The higher the correlation, the more significant the influence of the parameter on the winding current.

[0062] Step S2-4, sorting by the calculated correlation coefficients to screen out key motor parameters that have a significant impact on the winding current.

[0063] Step S3: In the model training phase, the electrical angle θ(t) and torque T at the same time in the database are used. e (t) and phase voltage u dc (t) and the motor winding current i(t-1) at the previous moment are used as training set inputs. The BP neural network model has multiple hidden layers, which are input layer, hidden layer and output layer. The input layer has 4 input ports, a is the number of data for each motor parameter in the database, and the input layer ports are electrical angle θ, torque T e and phase voltage u dc and the motor winding current i(t-1) at the previous moment; the output layer has 1 output port, and the output is the digital twin virtual motor winding current i p ; The number of neurons in its hidden layer is M, and the specific value is adjusted according to the actual working conditions.

[0064] The BP neural network model is trained by online training. The feature quantity x input at time t n for:

[0065] xn =[θ(t),T e (t),u dc (t),i(t-1)]

[0066] where x n The electrical angle θ and torque T in the database e and phase voltage u dc and the data set of the winding current of the physical entity motor at the previous moment; θ(t) represents the electrical angle of the motor at time t in the database; T e (t) represents the electrical angle of the motor at time t, u dc (t) represents the phase voltage of the motor at time t, and i(t-1) represents the motor winding current at time t-1; the feature quantities are input into the BP neural network model in chronological order for model training.

[0067] Step S4: In the online diagnosis stage, the trained BP neural network model is used as the digital twin virtual motor, and the downsampling technology is used to collect the electrical angle, torque and phase voltage data of the physical entity motor at the same time when it is running in real time. The collected real-time data is input into the digital twin virtual motor model in the form of streaming data to obtain the digital twin virtual motor winding current i p And the historical data is processed by the sliding window method. The model only retains the data points of the most recent fundamental cycle as the real-time training set at each time step.

[0068] Step S5: the current residual i of the motor winding current at time t in the online diagnosis phase error for:

[0069] i error (t) = i r (t)-i p (t)

[0070] where i r (t) is the actual current of the physical motor winding at time t, i p (t) is the digital twin virtual motor winding current at time t.

[0071] Step S6: In the online diagnosis phase, the current residual i error The difference value Δi is used as the criterion for judging whether the motor has a winding open circuit fault or a power tube open circuit fault. The difference value Δi of the current residual of the motor winding is:

[0072]

[0073] where t n t n-1 the latter moment;

[0074] The judgment basis for online diagnosis is:

[0075]

[0076] Where d is the set online diagnosis threshold.

[0077] Step S7: In the online diagnosis stage, record the sampled current value i of 0.5 current fundamental wave period after the fault moment. 1 and the sampling current value i of 0.75 current fundamental wave cycle 2 , add the two current values ​​and calculate the diagnostic variable K n Perform fault location.

[0078] Diagnostic variable K n The formula is:

[0079] K n =i 1 +i 2

[0080] where i 1 is the sampling current value of 0.5 current fundamental wave cycle after the fault, i 2 It is the sampling current value of 0.75 current fundamental wave cycle after the fault;

[0081] The basis for locating the fault is:

[0082]

[0083] The five-phase permanent magnet synchronous motor drive system is composed of a five-phase permanent magnet fault-tolerant motor and a set of five-phase half-bridge inverters. The power switch tubes T1, T2, T3, T4, T5, T6, T7, T8, T9, and T10 constitute the upper and lower bridge arms of the five-phase half-bridge inverter. The inverter bus terminal is U dc DC power supply, the five-phase permanent magnet fault-tolerant motor adopts a 10-slot 8-pole fractional slot double-layer concentrated winding structure, the electrical angle between the two-phase slots is 144°, the windings have no overlap, the rotor adopts a surface-mounted permanent magnet structure, and the centrifugal height is 5mm.

[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A five-phase motor open circuit fault diagnosis method, characterized in that: The diagnostic method includes a model training phase and an online diagnostic phase, and the specific steps are as follows: Step S1: In the model training stage, sequential motor parameters of the five-phase motor when it is operating normally under different working conditions are collected by downsampling technology to build a training database; Step S2: In the model training stage, the Pearson correlation coefficient is used to sort the importance of the collected five-phase motor parameters, and the electrical angle θ(t), torque T at the same time are selected according to the importance. e (t) and phase voltage u dc (t) as input data of BP neural network regression model; Step S3: In the model training stage, an online training method is adopted, using the normal operating electrical angle, torque and phase voltage data in the training database, combined with the corresponding winding current data as output, and a preliminary BP neural network model is obtained through training; Step S4: In the online diagnosis stage, the BP neural network model and its parameters obtained by training in step S3 are saved and used as the digital twin virtual motor; the electrical angle, torque and phase voltage of the physical motor at the same time are collected in real time through downsampling technology, and these data are input into the digital twin virtual motor. After online training, the winding current i of the virtual motor is updated. pn , n represents the five-phase motor A, B, C, D, E phases, and updates the parameters of the digital twin virtual motor in real time; Step S5: In the online diagnosis stage, the actual winding current i of the physical motor is rn and the digital twin virtual motor winding current i pn Compare and calculate the current residual i error_n ; Step S6: In the online diagnosis stage, by analyzing the differential value Δi of the current residual, it is determined whether the motor has a winding open circuit fault or a power tube open circuit fault; Step S7: In the online diagnosis stage, record the sampled current value i1 of 0.5 fundamental wave cycle of the current and the sampled current value i2 of 0.75 fundamental wave cycle of the current after the fault moment, add the two current values, and calculate the diagnostic variable K n Perform fault location.

2. A five-phase motor open circuit fault diagnosis method according to claim 1, characterized in that: When the step S1 uses downsampling technology to collect motor parameter data to build a database, the collected data is preprocessed, null values ​​in the database are deleted, and the Z-score method is used to remove obvious abnormal values ​​in the database.

3. A five-phase motor open circuit fault diagnosis method according to claim 1, characterized in that: In step S2, the correlation between various motor parameters and winding current is evaluated by calculating the Pearson correlation coefficient, which specifically includes the following steps: Step S2-1, randomly selecting a subset from the database as a sample set, wherein various motor parameters in the subset are used as an input set, and the corresponding winding current is used as an output set; Step S2-2: Calculate the correlation coefficient between each motor parameter and the winding current using the Pearson correlation coefficient formula, the formula is as follows: Among them, Q i is the motor parameter sample, P i is the winding current sample, and are the mean values ​​of the corresponding samples, r Q,P The value range of is [-1,1]; Step S2-3, judging the influence of each motor parameter on the winding current by the absolute value of the correlation coefficient. The higher the correlation, the more significant the influence of the parameter on the winding current. Step S2-4, sorting by the calculated correlation coefficients to screen out key motor parameters that have a significant impact on the winding current.

4. A five-phase motor open circuit fault diagnosis method according to claim 1, characterized in that: The BP neural network model in step S2 is a neural network with multiple hidden layers, wherein the three layers are respectively an input layer, a hidden layer and an output layer, wherein the input layer has three input ports, a is the number of data of each motor parameter in the database, and the input layer ports are respectively electrical angle θ(t), torque T e (t) and phase voltage u dc (t); The output layer has one output port, and the output is the digital twin virtual motor winding current i pn ; The number of neurons in its hidden layer is M, and the number of hidden layers is N. The specific values ​​of N and M are adjusted according to actual working conditions.

5. A five-phase motor open circuit fault diagnosis method according to claim 1, characterized in that: In step S3, the BP neural network model is trained by online training. The feature value x input at time t n for: x n =[θ(t),T e (t),u dc (t)] where x n The electrical angle θ and torque T in the database e , phase voltage u dc and the data set of the motor winding current at the previous moment; θ(t) represents the electrical angle of the motor at time t in the database; T e (t) represents the electrical angle of the motor at time t, u dc (t) represents the phase voltage of the motor at time t; the characteristic quantities are input into the BP neural network model in chronological order for model training.

6. A five-phase motor open circuit fault diagnosis method according to claim 1, characterized in that: In step S4, the trained BP neural network model is used as the digital twin virtual motor, and the downsampling technology is used to collect the electrical angle, torque and phase voltage data of the physical entity motor at the same time when it is running in real time. The collected real-time data is input into the digital twin virtual motor model in the form of streaming data to obtain the digital twin virtual motor winding current i pn And process historical data through the sliding window method. The model only retains the data points of the most recent fundamental cycle at each time step as the real-time training set for online training.

7. A five-phase motor open circuit fault diagnosis method according to claim 1, characterized in that: The current residual i of the motor winding current at time t in step S5 is error_n for: i error_n (t)=i rn (t)-i pn (t) where i rn (t) is the actual current of the physical motor winding at time t, i pn (t) is the digital twin virtual motor winding current at time t, and the differential value Δi of the motor winding current residual in step S6 is: where t n t n-1 the latter moment; The judgment basis for online diagnosis is: Where d1 is the set online diagnosis threshold.

8. A five-phase motor open circuit fault diagnosis method according to claim 1, characterized in that: In step S7, the diagnostic variable K n The formula is: K n =i1+i2 Where i1 is the sampling current value of 0.5 fundamental wave cycle of the current after the fault, and i2 is the sampling current value of 0.75 fundamental wave cycle of the current after the fault; The basis for locating the fault is:

9. A five-phase motor open circuit fault diagnosis method according to claim 1, characterized in that: The five-phase permanent magnet synchronous motor drive system is composed of a five-phase permanent magnet fault-tolerant motor and a set of five-phase half-bridge inverters. The power switch tubes T1, T2, T3, T4, T5, T6, T7, T8, T9, and T10 constitute the upper and lower bridge arms of the five-phase half-bridge inverter. The inverter bus terminal is U dc DC power supply, the five-phase permanent magnet fault-tolerant motor adopts a 10-slot 8-pole fractional slot double-layer concentrated winding structure, the electrical angle between the two-phase slots is 144°, the windings have no overlap, the rotor adopts a surface-mounted permanent magnet structure, and the centrifugal height is 5mm.

Images