Voltage source inverter open-circuit fault diagnosis method

Abstract

The invention provides a voltage source inverter open-circuit fault diagnosis method. The method comprises the following steps: in each PWM interruption period, obtaining a three-phase stator current and a PWM modulated reference voltage signal, and determining a stator voltage vector required by a voltage model; utilizing the deviation between the current model flux linkage and the voltage model flux linkage to obtain a compensation voltage vector; converting the compensation voltage vector into a three-phase compensation voltage through inverse Clark transformation; the periodic integral average feature of a certain phase is converted from time domain integral to phase angle domain integral, and periodic features of the three-phase compensation voltage are extracted in a phase angle domain; constructing a three-phase absolute periodic characteristic according to the periodic characteristic, and comparing the three-phase absolute periodic characteristic with a first fault detection threshold value to obtain a fault mark; and constructing a fault variable vector, mapping the fault variable vector to 1, 0 or-1 according to a second fault detection threshold value, and positioning a fault switch tube according to a table look-up rule. Under the conditions of variable speed, variable load and parameter mismatch, rapid and low-calculation-amount online diagnosis and positioning can be realized.

Description

Technical Field

[0001] This invention relates to the technical field of inverter open-circuit fault diagnosis, and more specifically, to a method for diagnosing open-circuit faults in voltage source inverters. Background Technology

[0002] Voltage source inverters are widely used in motor drives, new energy power generation, and other fields, and their reliability is of paramount importance. Power switching transistors are among the most vulnerable components in inverters. While open-circuit faults do not immediately cause system collapse, they can lead to current distortion and torque fluctuations, and prolonged operation may result in secondary system faults.

[0003] Existing methods for diagnosing open-circuit faults in inverters can be broadly categorized into signal feature-based methods, model-based methods, and data-driven methods. Signal feature-based methods typically extract features from the waveform distortion, DC-side ripple, or voltage-current imbalance of the three-phase current. While simple to implement, these features are often strongly correlated with operating frequency and load variations, making it difficult to uniformly set thresholds and prone to false alarms or missed alarms. Model-based methods establish a model of the motor-inverter system or an observer, extracting features using residuals or estimation errors. While offering some interpretability, they are highly dependent on the accuracy of model parameters, changes in operating conditions before and after the fault, and the internal computational links of the controller. Data-driven methods require a large amount of training data covering all operating conditions and suffer from generalization and interpretability issues in engineering applications. In summary, in variable-frequency operation control scenarios without speed vector sensors, the key challenge remains how to obtain fault features that are insensitive to frequency changes and easily implemented in an embedded-in-the-loop manner without adding extra hardware, and further achieve rapid location of single open-circuit faults.

[0004] Furthermore, many feature extraction methods based on periodic integrals or moving averages typically use a fixed period in the time domain as the window. However, in variable frequency speed control scenarios, the fundamental period varies with frequency, resulting in an indefinite window length. This leads to incomparability or instability of feature quantities at different frequencies. To address this issue, we point out that although the period varies with frequency, the phase angle sweeping from a specified starting angle corresponds to a complete period. This allows us to convert time-domain integral features into phase-domain integral features to adapt to variable frequency operating conditions. We further propose a discrete implementation to reduce storage and computational burden, making it more suitable for online execution within the controller's PWM interrupt cycle.

[0005] For the reasons mentioned above, it is necessary to propose an online diagnosis and location method for single-power switch open-circuit faults in voltage source inverters. Summary of the Invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art or related art.

[0007] In long-term operation of a voltage source inverter in a motor drive system, an open-circuit fault may occur in a single power switch. This fault leads to phase current distortion and torque fluctuations, affecting system stability. In sensorless vector control systems operating at variable frequencies, the fundamental period varies with frequency, and feature extraction methods based on fixed time windows are prone to instability or incomparability at different frequencies. Therefore, the purpose of this invention is to propose an open-circuit fault diagnosis method for voltage source inverters. Without adding external voltage measurement hardware, this method utilizes the compensation voltage available within the controller to construct fault features and transforms the period integral from the time domain to the phase angle domain to adapt to frequency variations, achieving rapid diagnosis and location of open-circuit faults.

[0008] To achieve the above objectives, the present invention provides a method for diagnosing open-circuit faults in voltage source inverters. This fault diagnosis method is used in a sensorless vector control system, which includes a current model and a voltage model; the voltage source inverter is a two-level voltage source inverter; the two-level voltage source inverter is a three-phase bridge inverter circuit, and the power supply for the three-phase bridge inverter circuit is a DC power supply V. d The output of the three-phase bridge inverter circuit is connected to a three-phase asynchronous motor; the three-phase bridge inverter circuit consists of three independent A-phase, B-phase, and C-phase bridge arms, and the A-phase bridge arm includes a No. 1 switching transistor and an anti-parallel diode. D Switches 1 and 2 and anti-parallel diodes D 2; Phase B bridge arm includes switch number 3 and anti-parallel diode. D Switches 3 and 4 and anti-parallel diodes D 4; The C-phase bridge arm includes switch number 5 and an anti-parallel diode. D Switches 5 and 6 and anti-parallel diodes D 6; Midpoint of each bridge arm a , b , c As AC output terminals, they are respectively connected to the corresponding phases of the three-phase asynchronous motor; the midpoint a , b , c The output currents are respectively i a , i b and i c ; i a , i b and i cThis refers to the three-phase stator current; the fault diagnosis method includes: Step S1: Obtaining the three-phase stator current during each PWM interrupt cycle. i a , i b , i c And the PWM-modulated reference voltage signal, and determine the stator voltage vector required for the voltage model. Step S2: Establish a hybrid model flux linkage observer to calculate the current model flux linkage and the voltage model flux linkage separately, and use the deviation between the current model flux linkage and the voltage model flux linkage through a proportional-integral circuit to obtain the compensated voltage vector. To further obtain the high-pass and low-pass weighted relationship of the voltage model stator flux linkage after compensation by the hybrid model flux linkage observer; wherein, the hybrid model flux linkage observer is a combination of the current model flux linkage observer and the voltage model flux linkage observer; the current model flux linkage is the current model stator flux linkage, and the voltage model flux linkage includes the voltage model stator flux linkage and the voltage model rotor flux linkage; Step S3: The compensated voltage vector is obtained by inverse Clarke transform. Convert to three-phase compensation voltage Step S4: Calculate the flux linkage phase angle from the rotor flux linkage components of the voltage model. And define the phase angle of the magnetic flux linkage. The output range is 0 to Step S5: Utilize the magnetic flux phase angle Sweep starting from the designated starting angle The property corresponding to a complete fundamental period T is the characteristic of the integral average of a certain phase period. The time-domain integral is transformed into the phase-domain integral, and the periodic characteristics of the three-phase compensation voltage are extracted in the phase-domain. Step S6: Based on periodic characteristics Constructing three-phase absolute periodic characteristics and the first fault detection threshold The fault flag is obtained by comparison. Step S7: Construct the fault variable vector And based on the second fault detection threshold fault variable vector The mapping is 1, 0, or -1, and the faulty switching transistor is located according to the lookup table rules; wherein, the fault variable vector Three-phase fault variables composition.

[0009] Preferably, in step S2, the current model flux linkage is a current model stator flux linkage, and the voltage model flux linkage includes a voltage model stator flux linkage and a voltage model rotor flux linkage.

[0010] The expression corresponding to the stator flux linkage in the current model is:

[0011] (1)

[0012] In equation (1), This provides data for estimating stator flux linkage in a current model. For current model rotor flux estimation data; L m This refers to the excitation inductance parameter data; L r This refers to rotor inductance parameter data; L s These are the stator inductance parameter data; This is the stator current vector data;

[0013] The expression corresponding to the stator flux linkage in the voltage model is:

[0014] (2)

[0015] In equation (2), Input data for the stator voltage vector; R s This refers to the stator resistance parameter data; This is the stator flux linkage estimation data for the voltage model;

[0016] The expression corresponding to the rotor flux linkage in the voltage model is:

[0017] (3)

[0018] In equation (3), For voltage model rotor flux estimation data;

[0019] The expression for the stator flux linkage in the compensated voltage model is:

[0020] (4)

[0021] In equation (4), The stator flux linkage data for the compensated voltage model; To compensate for voltage vector data;

[0022] The compensation voltage vector The corresponding expression is:

[0023] (5)

[0024] In equation (5), These are proportional coefficient data; This is integral coefficient data; and The difference is the flux linkage error data;

[0025] The expression for the high-pass and low-pass weighted relationship of the stator flux linkage in the voltage model after compensation by the hybrid model flux linkage observer is as follows:

[0026] (6)

[0027] In equation (6), This is the transfer function data for a second-order high-pass filter; This is the transfer function data for a second-order low-pass filter.

[0028] Preferably, in step S3, the three-phase compensation voltage The corresponding expression is:

[0029] (7)

[0030] In equation (7), Compensation voltage Component data; These are the three-phase compensation voltage data.

[0031] Preferably, in step S4, the flux linkage phase angle The corresponding expression is:

[0032] (8)

[0033] In equation (8), This is the flux linkage phase angle data; For the rotor flux linkage component data of the voltage model; when the two input components of atan2 When both are 0, Set to 0.

[0034] Preferably, in step S5, the periodic integral average characteristic of a certain phase is... The corresponding expression is:

[0035] (9)

[0036] In equation (9), T represents the periodic integral average characteristic data of a certain phase; T represents a complete fundamental period data. t 1 represents the data at the start of the integration process; p For the phase sequence, p takes the values ​​a, b, and c, where a, b, and c correspond to phase A, phase B, and phase C, respectively.

[0037] The periodic integral average characteristic of a certain phase The expression corresponding to converting the time-domain integral to the phase-domain integral is:

[0038] (10)

[0039] In equation (10), This is the flux linkage phase angle data; This is the angular span data of the magnetic flux phase angle sweeping around one revolution.

[0040] Preferably, in step S5, periodic characteristics of the three-phase compensation voltage are extracted in the phase angle domain. Specifically, it includes:

[0041] According to the sampling period T s Calculate the number of sampling points per cycle with the complete fundamental period T. Constructing the three-phase compensation voltage and flux phase angle Discrete sampling sequence; within each PWM interrupt cycle, the three-phase compensation voltage and flux linkage phase angle Each sample value is obtained and entered into the compensation voltage sampling set and phase angle sampling set respectively; 0 to Divide into n equal parts to obtain the phase angle step of equal flux linkage and with Construct a discrete phase angle sequence with equal flux linkage phase angles for the initial phase angle; when the flux linkage phase angle... Scan At that time, the moving average of the equal phase angle within each magnetic flux phase angle interval is calculated; the mean characteristics of the n phase angle intervals are summed by second-order equal phase angle shift to obtain the periodic characteristics. ;

[0042] Among them, the number of sampling points per cycle The corresponding expression is:

[0043] (11)

[0044] In equation (11), T s Data for the sampling period; This is data on the number of sampling points per cycle;

[0045] Three-phase compensation voltage and flux phase angle The expression corresponding to the discrete sampling sequence is:

[0046] (12)

[0047] (13)

[0048] In equations (12) and (13), k is the sampling time; For the sampling point index and Take (0, ); This is the compensation voltage for phase A; This is the compensation voltage for phase B; This is the C-phase compensation voltage; Phase angle of magnetic flux At time ( The sampled value of )

[0049] The expressions for the compensation voltage sampling set and the phase angle sampling set are as follows:

[0050] (14)

[0051] (15)

[0052] In equations (14) and (15), To compensate for voltage sampling data; This is the phase angle sampling set data;

[0053] Equal flux phase angle step The calculation expression is:

[0054] (16)

[0055] In equation (16), n is the number of equal divisions of the phase angle domain; For each segment of flux linkage phase angle, a fixed increment is used, i.e., from 0 to... The phase angle of each part after dividing it into n equal parts;

[0056] The expression for the discrete phase angle sequence with equal flux linkage is:

[0057] (17)

[0058] In equation (17), This is the initial flux linkage phase angle data; The sequence number of the flux linkage phase angle step segment and Take (0, n-1); Indicates from Start by step size Phase angle points obtained by segment-by-segment discretization;

[0059] The expression for the moving average of the phase angle within each flux linkage phase angle interval is:

[0060] (18)

[0061] In equation (18), This represents the number of sampling points falling within this phase angle interval; This represents the mean characteristic data for this phase angle interval; This is for compensating voltage sampling data.

[0062] Periodic characteristics The calculation expression is:

[0063] (19)

[0064] In equation (19), The periodic feature is obtained by summing the mean features of each small segment within the phase angle domain.

[0065] Preferably, n is less than To ensure that within each phase angle interval Not equal to 0, and n is in The range is 10 to 15 percent.

[0066] Preferably, in step S6, the three-phase absolute periodicity characteristics The expression is:

[0067] (20)

[0068] In equation (20), The absolute value data represents the three-phase periodic characteristics.

[0069] Fault signs The expression is:

[0070] (twenty one)

[0071] In equation (21), This is fault flag data; The first fault detection threshold data is used, and the G value in any phase exceeds... Time determination The indicator shows an open circuit fault in the switching transistor; The absolute periodic characteristics of the A-phase compensation voltage; The absolute periodic characteristics of the B-phase compensation voltage; The absolute periodic characteristics of the C-phase compensation voltage;

[0072] First fault detection threshold By loading the maximum value of the transient mean feature Perform calibration. The range of values ​​is greater than and not less than To reduce the risk of false alarms.

[0073] Preferably, in step S7, the fault variable vector The expression is:

[0074] (twenty two)

[0075] In equation (22), This is data for three-phase fault variables;

[0076] Based on the second fault detection threshold fault variable vector The expressions corresponding to mappings to 1, 0, or -1 are:

[0077] (twenty three)

[0078] In equation (23), This is the second fault detection threshold data; Periodic feature data; second fault detection threshold Determined through experimental tuning, and based on the fault variable vector. The rules for determining the values ​​of the fault variable vector allow it to take the values ​​1, 0, or -1.

[0079] Preferably, in step S7, the mapping relationship corresponding to the lookup rule is Table 1; in Table 1, the output... Each switch corresponds one-to-one with the fault switch number, including the normal state and the fault state of switch numbers 1 to 6; 0 indicates no open circuit fault, and 1 or -1 indicates the presence of an open circuit fault.

[0080] Table 1. Mapping Relationships Corresponding to Table Lookup Rules

[0081] .

[0082] The beneficial effects of this invention are:

[0083] Compared with the prior art, the open-circuit fault diagnosis method for voltage source inverters provided by the present invention has the following advantages:

[0084] (1) Different sources of diagnostic information: Existing technologies are mostly based on external quantities such as three-phase current distortion, negative sequence component, mean offset or voltage imbalance for judgment. This patent changes to using the compensation voltage generated by the deviation of the two models through proportional-integral in the hybrid model flux linkage observer as the diagnostic quantity, and uses its periodic characteristics as the fault feature.

[0085] (2) Different feature extraction scales: Existing technologies often use a fixed time window for moving integrals or averaging. When the frequency is changed, the fundamental frequency period changes, causing the features to be unstable. This patent converts the moving integral from the time domain to the phase domain and sweeps the phase angle. It corresponds to a complete cycle, so that feature extraction is not affected by cycle changes caused by frequency changes;

[0086] (3) Different engineering implementation methods: Existing technology often requires a large number of sampling points to be stored and calculated for the whole cycle for periodic feature extraction. This patent divides 0 to n equally into equal phase angle intervals, performs equal phase angle interval mean and second-order shift summation, and controls the calculation burden with n, which significantly reduces storage and operation, and is suitable for PWM interrupt online implementation;

[0087] (4) Different output granularity: Existing technologies often only realize fault detection or phase location. This patent first obtains fault flags through absolute features and thresholds to realize detection, then maps the three-phase periodic features to discrete fault variables of 1, 0 or -1 and constructs a fault variable vector, and outputs the specific fault switch number through table lookup rules to realize switch-level location.

[0088] (5) Different hardware dependencies: Some existing solutions rely on external phase voltage sensors or additional sampling circuits. In this patent, the voltage model input can be the reference voltage or the internal signal reconstructed voltage, which is not obtained through external hardware circuit measurement, thus reducing hardware modification costs and improving feasibility.

[0089] Additional aspects and advantages of the invention will become apparent from the description which follows, or may be learned by practice of the invention. Attached Figure Description

[0090] Figure 1 A structural diagram of an induction motor drive system powered by a three-phase two-level voltage source inverter according to an embodiment of the present invention is shown.

[0091] Figure 2 A schematic diagram of a sensorless vector control system for an induction motor according to an embodiment of the present invention is shown.

[0092] Figure 3 A schematic diagram illustrating the conversion principle from time-domain integration to phase-domain integration according to an embodiment of the present invention is shown.

[0093] Figure 4 A schematic diagram of the distribution of sampling points in the phase angle domain in a second-order equal-phase-angle moving average method according to an embodiment of the present invention is shown;

[0094] Figure 5 A flowchart illustrating the in-loop implementation of a voltage source inverter open-circuit fault diagnosis method according to an embodiment of the present invention in the controller is shown.

[0095] Figure 6 This diagram illustrates the experimental results of a switch (No. 1) experiencing an open-circuit fault during acceleration, according to an embodiment of the present invention.

[0096] Figure 7 This diagram illustrates the experimental results of a No. 2 switch in an embodiment of the present invention when an open-circuit fault occurs during a transient motor parameter mismatch.

[0097] Figure 8 A schematic diagram of experimental results showing an open-circuit fault in switch transistor No. 2 when the rotational speed changes according to an embodiment of the present invention is shown.

[0098] Figure 9 The diagram shows the experimental results of an open-circuit fault of switch No. 1 in an embodiment of the present invention when the motor parameters are mismatched. Detailed Implementation

[0099] To better understand the above-mentioned objects, features, and advantages of the present invention, such as Figures 1 to 9 As shown in the accompanying drawings and specific embodiments, the present invention will be further described in detail below. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0100] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0101] The technical solution of the present invention will be illustrated below with a specific embodiment. This specific embodiment describes a voltage source inverter open-circuit fault diagnosis method, used in a sensorless vector control system, such as... Figure 2 As shown, the sensorless vector control system includes a current model and a voltage model; as Figure 1 As shown, the voltage source inverter is a two-level voltage source inverter; the two-level voltage source inverter is a three-phase bridge inverter circuit, and the power supply for the three-phase bridge inverter circuit is a DC power supply V. d The output of the three-phase bridge inverter circuit is connected to the three-phase asynchronous motor. The three-phase bridge inverter circuit consists of three independent A-phase, B-phase, and C-phase bridge arms. The A-phase bridge arm includes switch 1 and anti-parallel diode D1, switch 2 and anti-parallel diode D2; the B-phase bridge arm includes switch 3 and anti-parallel diode D3, switch 4 and anti-parallel diode D4; and the C-phase bridge arm includes switch 5 and anti-parallel diode D5, switch 6 and anti-parallel diode D6. The midpoints a, b, and c of each bridge arm serve as AC output terminals, respectively connected to the corresponding phases of the three-phase asynchronous motor. The currents output from midpoints a, b, and c are i... a i b and i c i a i b and i c This refers to the three-phase stator current; the implementation of this fault diagnosis method relies on a typical sensorless vector control system, the structure of which is as follows: Figure 2As shown, this typical sensorless vector control system includes a three-phase two-level voltage source inverter, an induction motor, and a sensorless vector controller based on a hybrid model flux linkage observer.

[0102] This fault diagnosis method is implemented through the following technical solution:

[0103] 1) Signal acquisition and voltage vector determination:

[0104] During each PWM interrupt cycle, the three-phase stator current i is sampled. a i b i c The reference voltage signal modulated by PWM is used to simultaneously determine the stator voltage vector required for the voltage model. The aforementioned The selected voltage is either the reference voltage output by the controller or the voltage reconstructed from the internal signals of the controller, and the... Obtained without measurement through external hardware circuitry.

[0105] 2) Establish a hybrid model flux linkage observer and generate compensation voltage:

[0106] The flux linkages in the current model and the voltage model are calculated separately, and the compensation voltage vector is obtained by using the deviation between the two through a proportional-integral circuit. And used for magnetic flux compensation.

[0107] like Figure 2 As shown, the workflow of a sensorless vector control system includes: speed setting and regulation, current sampling and current control, and coordinate transformation (abc / The module includes modules such as / dq), PWM modulation and inverter drive, flux linkage observation and phase angle calculation; and below is a description of the composition of the hybrid model flux linkage observer: the current model and voltage model estimate the flux linkage in parallel, and the deviation between the two is used to generate a compensation voltage through a PI (proportional-integral) stage to correct the flux linkage of the voltage model.

[0108] Depend on Figure 2 As can be seen, the compensation voltage and its characteristics in this specific embodiment come from the internal signal link of the controller and can be obtained online without the need for additional hardware sensors; the compensation voltage is obtained by the deviation between the current model and the voltage model through a proportional-integral circuit.

[0109] The current model flux linkage is a current model stator flux linkage, and the voltage model flux linkage includes a voltage model stator flux linkage and a voltage model rotor flux linkage;

[0110] The current model, voltage model, and compensation relationship are calculated using the following expressions:

[0111] (1)

[0112] In equation (1), This provides data for estimating stator flux linkage in a current model. For rotor flux linkage estimation data in the current model; L m For excitation inductance parameter data; L r For rotor inductance parameter data; L s These are the stator inductance parameter data; This is the stator current vector data;

[0113] (2)

[0114] In equation (2), For stator voltage vector input data; R s This refers to the stator resistance parameter data; This is the stator flux linkage estimation data for the voltage model;

[0115] (3)

[0116] In equation (3), For voltage model rotor flux estimation data;

[0117] (4)

[0118] In equation (4), The stator flux linkage data for the compensated voltage model; To compensate for voltage vector data;

[0119] (5)

[0120] In equation (5), These are proportional coefficient data; This is integral coefficient data; and The difference is the flux linkage error data;

[0121] (6)

[0122] In equation (6), This is the transfer function data for a second-order high-pass filter; This is the transfer function data for a second-order low-pass filter.

[0123] 3) Obtain the three-phase compensation voltage:

[0124] Compensation voltage vector Transformed into three-phase compensated voltage Implemented according to the following expression:

[0125] (7)

[0126] In equation (7), Compensation voltage Component data; These are the three-phase compensation voltage data.

[0127] 4) Phase angle calculation and phase angle domain sliding integral:

[0128] Calculate the flux linkage phase angle from the rotor flux linkage components of the voltage model The flux linkage phase angle is calculated and defined by the following expression. The output range is 0 to :

[0129] (8)

[0130] In equation (8), This is the flux linkage phase angle data; For the rotor flux linkage component data of the voltage model; when the two input components of atan2 When both are 0, Set to 0.

[0131] Using magnetic flux phase angle Sweep starting from the designated starting angle For a property corresponding to a complete fundamental period T, the time-domain integral form of a certain phase is expressed by the following formula: The integral form in the time domain is converted to the integral form in the phase angle domain, thereby avoiding the impact of period changes during frequency conversion operation.

[0132] Figure 3 This is a schematic diagram illustrating the conversion principle from time-domain integration to phase-domain integration. (For example...) Figure 3 As shown, flux linkage phase angle Sweeping across 2 μm from the starting point always corresponds to a complete electrical angle cycle; therefore, the periodic integral / average characteristic can be converted from a fixed window in the time domain to a fixed window in the phase angle domain. Figure 3 It can be seen that the phase angle domain integral is based on " Using "scanning a week" as a window allows feature extraction to be decoupled from frequency changes, making features more stable and comparable under variable frequency / acceleration / deceleration conditions, thereby reducing false alarms and missed alarms.

[0133] Periodic integral average characteristics of a certain phase The corresponding expression is:

[0134] (9)

[0135] In equation (9), T represents the periodic integral average characteristic data of a certain phase; T represents a complete fundamental period data; t1 represents the data at the start of integration; p represents the phase sequence, where p takes the values ​​a, b, and c, and a, b, and c correspond to phases A, B, and C, respectively.

[0136] The periodic integral average characteristic of a certain phase The expression corresponding to converting the time-domain integral to the phase-domain integral is:

[0137] (10)

[0138] In equation (10), This is the flux linkage phase angle data; Angular span data for the phase angle of the magnetic flux linkage sweeping around one revolution

[0139] 5) Discrete implementation is used to extract periodic features and complete detection and localization:

[0140] From 0 to Divide the data into n equal segments, calculate the moving average of equal phase angles, and further perform second-order equal phase angle shifting and summation to obtain the periodic characteristics. .

[0141] Specifically, periodic characteristics are extracted from the three-phase compensation voltage in the phase angle domain. Specifically, it includes:

[0142] According to the sampling period T s Calculate the number of sampling points per cycle with the complete fundamental period T. ;

[0143] Constructing the three-phase compensation voltage and flux phase angle Discrete sampling sequence;

[0144] During each PWM interrupt cycle, the three-phase compensation voltage and flux phase angle Each sample value is obtained and entered into the compensation voltage sampling set and the phase angle sampling set respectively;

[0145] From 0 to Divide into n equal parts to obtain the phase angle step of equal flux linkage and with Construct a discrete phase angle sequence with equal flux linkage phase angles for the initial phase angle;

[0146] When the magnetic flux phase angle Scan At that time, the moving average of the phase angle within each flux linkage phase angle interval is calculated;

[0147] The periodic characteristics are obtained by performing a second-order equal-phase-angle shift and summation on the mean characteristics of n phase angle intervals. .

[0148] Figure 4 This is a schematic diagram illustrating the distribution of sampling points in the phase angle domain in the second-order equal-phase-angle moving average method. For example... Figure 4 As shown, from 0 to After the phase angle is divided into n equal segments, the sampling points follow The idea is to discretize the data by sampling within each phase angle interval, taking a moving average within each interval, and then summing the interval averages using a second-order shift to form the periodic characteristics. This is based on the discretization concept of sampling within each phase angle interval and taking a moving average within each interval. Figure 4 It can be seen that this implementation does not rely on time-based sampling alignment of the entire cycle, but instead uses segmented statistics based on phase angle. The computational / storage load can be directly controlled by n, making it more suitable for online implementation of embedded PWM interrupts while maintaining the stability of the cycle characteristics.

[0149] Specifically, it can be implemented using the following expression:

[0150] The expression for the moving average of the phase angle within each flux linkage phase angle interval is:

[0151] (18)

[0152] In equation (18), This represents the number of sampling points falling within this phase angle interval; This represents the mean characteristic data for this phase angle interval; To compensate for voltage sampling data;

[0153] Periodic characteristics The calculation expression is:

[0154] (19)

[0155] In equation (19), The periodic feature is obtained by summing the mean features of each small segment within the phase angle domain;

[0156] Among them, the number of sampling points per cycle Determine by the following expression, and set n to be less than :

[0157] (11)

[0158] In equation (11), T s Data for the sampling period; This is data on the number of sampling points per cycle;

[0159] Depend on Obtain absolute features And generate fault flags Implement z according to the following expression, and set the threshold to . :

[0160] (20)

[0161] In equation (20), The absolute value data represents the three-phase periodic characteristics.

[0162] Fault signs The expression is:

[0163] (twenty one)

[0164] In equation (21), This is fault flag data; The first fault detection threshold data is used, and the G value in any phase exceeds... Time determination The indicator shows an open circuit fault in the switching transistor; The absolute periodic characteristics of the A-phase compensation voltage; The absolute periodic characteristics of the B-phase compensation voltage; The absolute periodic characteristics of the C-phase compensation voltage;

[0165] when When a fault occurs, a fault variable vector is constructed, and the three-phase results are mapped to discrete values ​​according to the following expression:

[0166] (twenty two)

[0167] In equation (22), This is data for three-phase fault variables;

[0168] (twenty three)

[0169] In equation (23), This is the second fault detection threshold data; Periodic feature data; second fault detection threshold Determined through experimental tuning, and based on the fault variable vector. The rules for determining the values ​​of the fault variable vector allow it to take the values ​​1, 0, or -1.

[0170] Then, based on the lookup table rules, the fault switch numbers are output to locate open-circuit faults in switches 1 through 6. The mapping relationship corresponding to the lookup table rules is shown in Table 1; in Table 1, the output... Each switch corresponds one-to-one with the fault switch number, including the normal state and the fault state of switch numbers 1 to 6; 0 indicates no open circuit fault, and 1 or -1 indicates the presence of an open circuit fault.

[0171] Table 1. Mapping Relationships Corresponding to Table Lookup Rules

[0172] .

[0173] In this specific embodiment, Figure 6 This is a schematic diagram showing the experimental results when switch #1 experiences an open-circuit fault during acceleration. (Example:) Figure 6 As shown, the experimental waveforms of switch 1 being open during acceleration are displayed, including: (a) three-phase current; (b) three-phase compensation voltage; (c) time-domain average characteristics of compensation voltage; (d) phase angle average characteristics of compensation voltage; and (e) detection and positioning output results (including threshold line and positioning conclusion of switch 1). Figure 6 As can be seen, before and after the fault occurs, the characteristic quantity abruptly changes and exceeds the threshold. Figure 6 It can be seen that, under the condition of frequency change, the phase angle domain feature can still stably trigger the fault flag and correctly locate "switch tube No. 1". The verification method still has the ability to quickly and accurately detect and locate under dynamic frequency conversion conditions.

[0174] In this specific embodiment, Figure 7 This is a schematic diagram showing the experimental results of switch #2 when an open-circuit fault occurs due to transient motor parameter mismatch. Figure 7 The diagram shows the similar waveforms and characteristic outputs when switch #2 is open, under conditions of motor parameter mismatch and transient disturbances: three-phase current, three-phase compensation voltage, time-domain / phase-domain characteristics, threshold values, and positioning results. Figure 7 It can be seen that even if there is transient disturbance caused by parameter mismatch, the fault characteristics can still be clearly distinguished from the threshold criterion and the location of "switch No. 2" can be given, indicating that the method is robust to parameter uncertainty and is not easily misled by transients.

[0175] In this specific embodiment, Figure 8 This is a schematic diagram showing the experimental results of an open-circuit fault in switch transistor #2 when the rotational speed changes. Figure 8 As shown, in addition to the three-phase current, the compensation voltage and its periodic characteristics, threshold line and positioning output are also given; Figure 8 The fault location is marked, and it is evident that the feature values ​​changed significantly after the fault and remained stable within the criterion interval. Figure 8 It can be seen that when the speed change causes the fundamental frequency period to change, the feature remains stable and can continuously output the correct positioning result, proving that the phase angle domain feature extraction has good adaptability to variable speed and frequency conversion and is suitable for online diagnosis in actual speed regulation operation.

[0176] In this specific embodiment, Figure 9 This is a schematic diagram showing the experimental results of an open-circuit fault in switch #1 when the motor parameters are mismatched. (Example:) Figure 9 The diagram shows the experimental waveform and diagnostic output for an open-circuit switch (switch #1): including compensation voltage, periodic characteristics, threshold judgment, and the final result locating the switch as "switch #1". Figure 9It can be seen that the method can still correctly trigger the threshold and stably locate the fault under the combined conditions of "parameter mismatch + open circuit fault", which shows that the method has a stronger fault tolerance to model parameter deviation and higher engineering usability.

[0177] In this specific embodiment, the fault diagnosis algorithm is executed in the microprocessor unit of the controller. The fault diagnosis algorithm is specifically implemented through three main aspects: parameter setting and storage preparation, online calculation process within each PWM interrupt cycle, and threshold calibration and experimental tuning.

[0178] I. Parameter Setting and Storage Preparation

[0179] (1) Sampling period T s :

[0180] Sampling period T s The PWM interrupt cycle is set to ensure that a three-phase stator current sample value is acquired and updated once within each PWM interrupt cycle. s Used to calculate the number of sampling points in a single period , This represents the number of sampling points per cycle.

[0181] (2) Phase angle division parameter n and step size :

[0182] From 0 to Divide into n equal parts to obtain the equal phase angle step size. n represents the number of equal divisions of the phase angle domain. To ensure the number of sampling points within each phase angle interval... Not 0, n satisfies less than and take The range is 10 to 15 percent. This represents the number of sampling points falling within the phase angle interval.

[0183] (3) Threshold and :

[0184] Set fault detection thresholds With fault location threshold . Used to obtain fault flags from absolute feature G . Used to transfer fault variable vectors The mapping is 1, 0, or -1 to facilitate table lookup for locating the switching transistor.

[0185] (4) Sample set and circular storage:

[0186] Constructing a compensation voltage sampling set With phase angle sampling set . To compensate for voltage sampling data, This is the phase angle sampling set data. Simultaneously, it establishes the cumulative values ​​and counts for n phase angle intervals of the three phases, storing them in a circular storage manner with a circular storage size of 3n. This circular storage is used to perform segmented accumulation and statistics of the three-phase compensation voltage within the phase angle domain, avoiding the storage of the complete sequence of each sampling point.

[0187] II. Online Calculation Process within Each PWM Interrupt Cycle

[0188] (1) Obtain the three-phase stator current and stator voltage vectors :

[0189] Acquire three-phase stator current in each PWM interrupt cycle . It contains current sampling values ​​for phases A, B, and C, which are used for current model flux estimation and subsequent compensation voltage calculation.

[0190] Acquire the reference voltage signal and determine the stator voltage vector within the same PWM interrupt cycle. . The selection method is to take the reference voltage output by the controller or the voltage reconstructed from the internal signals of the controller, and This is determined without measurement using external hardware circuitry. The goal is to enable voltage model inputs to be obtained within the controller, reducing reliance on external measurement circuits.

[0191] (2) Hybrid model flux linkage observer calculation and compensation voltage vector generate:

[0192] A hybrid model flux linkage observer is established to calculate the flux linkage under both the current model and voltage model. The current model flux linkage is calculated using the current model flux linkage expression, yielding estimated stator and rotor flux linkage data under the current model. This calculation uses the magnetizing inductance L. m Rotor inductance L r Stator inductance L s and stator current vector data Wait for the input data.

[0193] The stator flux linkage in the voltage model is calculated using the expression for the stator flux linkage in the voltage model. The input includes the stator voltage vector. Stator resistance R s Stator current vector Data such as...

[0194] The rotor flux linkage of the voltage model is calculated according to the expression of the rotor flux linkage of the voltage model, and the estimated data of the rotor flux linkage of the voltage model are obtained.

[0195] Based on the deviation between the flux linkage in the current model and the flux linkage in the voltage model, the compensation voltage vector is obtained through a proportional-integral circuit. The proportional-integral process uses a proportionality coefficient. With integral coefficient The input is the difference between the compensated stator flux linkage and the stator flux linkage of the current model, which is used as flux linkage error data.

[0196] During normal operation, the compensation voltage vector The average period tends to be shorter. After a single-switch open-circuit fault occurs, the three-phase current waveform becomes missing or distorted, disrupting the variation pattern of the flux linkage error within one cycle, thus... Significant deviations occur on the periodic scale, which facilitates the subsequent extraction of periodic features in the phase angle domain.

[0197] The hybrid model uses a weighted relationship between high-pass and low-pass filters. , Perform weighted summation and satisfy the following conditions: and The sum of is 1, which keeps the observer output stable within the operating range.

[0198] (3) The inverse Clarke transformation yields the three-phase compensation voltage:

[0199] Compensation voltage vector Converted into a three-phase compensation voltage via inverse Clarke transform . These are the three-phase compensation voltage data. The three-phase compensation voltage is used for subsequent phase angle domain piecewise statistics and periodic characteristics. Input data for the calculation.

[0200] (4) Calculate the flux linkage phase angle And limit the output range:

[0201] Calculate the flux linkage phase angle from the rotor flux linkage components of the voltage model , The input is the rotor flux linkage component data from the voltage model, which represents the flux linkage phase angle data. Limit the output range to make Keep at 0 to .

[0202] When both input components of atan2 are 0, Set to the default value of 0 to avoid calculation errors caused by undefined phase angles.

[0203] flux linkage phase angle Used to determine whether the phase angle originates from the starting angle. Start scanning Sweeping through the phase angle Determine a complete fundamental period T to avoid the instability of the period integration window caused by T not being constant due to changes in the reference frequency.

[0204] (5) Phase angle sweep Determine the period and extract periodic features in the phase angle domain :

[0205] Time-domain periodicity Calculated based on the expression defined by the time-domain periodic characteristics. Let p be the periodic integral average characteristic data of a certain phase, where a, b, and c correspond to phases A, B, and C, respectively.

[0206] Since T is not a constant value as a function of the reference frequency, the integration is transformed from the time domain to the phase domain. The phase angle is used to integrate from the starting angle. Start scanning It was determined that a complete cycle T had passed, and the cycle characteristics of the three-phase compensation voltage were extracted in the phase angle domain. .

[0207] In the digital implementation, first, based on the sampling period T... s Calculate the number of sampling points per period with period T And construct a discrete sampling sequence of compensation voltage and phase angle. k is the sampling time. As the sampling point index, the compensation voltage and phase angle obtain a sample value in each PWM interrupt cycle and enter it into the sampling set. and .

[0208] Dividing 0 to 2π into n equal parts yields the equifluence phase angle step. and with Construct a discrete phase angle sequence with equal flux linkage for the initial phase angle. Within each PWM interrupt cycle, based on the current... Determine the phase angle interval it falls into, and then set the current u a ,u b ,u c Accumulate them separately into the corresponding intervals, and then set the intervals accordingly. Add 1.

[0209] When the phase angle sweeps through 2π, it is determined that a cycle has ended. For each phase angle interval, the moving average of the equal phase angles is calculated. . This represents the mean characteristic data for this phase angle interval. Then, a second-order equal-phase-angle shift summation is performed on the mean characteristics of the n phase angle intervals to obtain the periodic characteristics. . This is to obtain periodic feature data by accumulating the mean features of each small segment within the phase angle domain. To reduce storage usage, the accumulated values ​​of n intervals across the three phases are used for cyclic storage, with a cyclic storage size of 3n. The complete waveform of each sampling point is not saved; only the interval accumulated value and the number of sampling points in each interval are stored. .

[0210] (6) Constructing absolute features And generate fault flags :

[0211] Due to periodic characteristics Constructing absolute periodic features , This represents the absolute value data of the three-phase periodic characteristics. With fault detection threshold The fault flag is obtained by comparison. . This is fault flag data. When any phase... Value exceeds Time determination This indicates an open circuit fault.

[0212] (7) Construct the fault variable vector and locate the faulty switch tube by looking up the table:

[0213] Construct a fault variable vector, the vector is formed by... The components consist of three phase fault variable data, based on threshold values. fault variable vector The mapping is 1, 0, or -1, where 0 indicates no open-circuit fault characteristics and 1 or -1 indicates the presence of open-circuit fault characteristics.

[0214] Then, by consulting the table and referring to the mapping relationship shown in Table 1, Each switch corresponds to a fault switch number and outputs either a normal state or a fault state for switches 1 through 6, thus enabling the location of single-switch open-circuit faults.

[0215] II. Threshold Calibration and Experimental Tuning

[0216] Fault detection threshold By loading the maximum value of the transient mean feature Perform calibration to make Greater than and Not less than To reduce the risk of false alarms. Fault location threshold. Through experimental tuning, it was determined that the output would be stable at 1, 0, or -1 under the fault variable value rules, and that the lookup table location results would be repeatable and consistent under normal and single open-circuit fault conditions.

[0217] Through simulation and experimental verification, the method proposed in this invention can quickly diagnose and locate single-switch open-circuit faults with a diagnosis time of less than T / 4 under different speeds, different loads, and motor parameter mismatches. Moreover, it does not require additional hardware circuits, has a low computational burden, and has good prospects for engineering applications.

[0218] like Figure 5 As shown in Table 1, the voltage source inverter open-circuit fault diagnosis method provided by this invention samples the three-phase current and reconstructs the reference voltage within each triangular carrier cycle; and calculates based on the voltage model and current model. and Calculate the compensation voltage; calculate the periodic characteristics of the compensation voltage and the fault flag quantity using the SoE-PAMM method. With fault variable vector Determine the fault flag quantity Is it greater than or equal to the first fault detection threshold? When the judgment result is yes, based on the fault variable vector... Locate the fault switch in Table 1; finally output the fault information; when the judgment result is yes, directly return to the steps of sampling the three-phase current and reconstructing the reference voltage in each triangular carrier cycle.

[0219] like Figure 5 As shown, the online steps within each PWM interrupt cycle are: sampling current and reference voltage → calculating flux linkage and compensation voltage → calculating flux linkage phase angle → extracting three-phase compensation voltage cycle characteristics in the phase angle domain → comparing with a threshold to obtain fault flags → constructing a fault variable vector and locating the faulty switch transistor by looking up a table → outputting fault information. Figure 5 It can be seen that the diagnostic algorithm runs in real time within the control closed loop, which is a combination of online detection and online positioning.

[0220] In summary, the open-circuit fault diagnosis method for voltage source inverters provided by this invention has unparalleled technical advantages.

[0221] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for diagnosing open-circuit faults in a voltage source inverter, used in a sensorless vector control system, wherein the sensorless vector control system includes a current model and a voltage model; the voltage source inverter is a two-level voltage source inverter; the two-level voltage source inverter is a three-phase bridge inverter circuit, and the power supply for the three-phase bridge inverter circuit is a DC power supply. V d The output of the three-phase bridge inverter circuit is connected to a three-phase asynchronous motor; the three-phase bridge inverter circuit consists of three independent A-phase, B-phase, and C-phase bridge arms, and the A-phase bridge arm includes a No. 1 switching transistor and an anti-parallel diode. D Switches 1 and 2 and anti-parallel diodes D 2; Phase B bridge arm includes switch number 3 and anti-parallel diode. D 3 No. 4 switch and anti-parallel diode D 4; The C-phase bridge arm includes switch number 5 and an anti-parallel diode. D Switches 5 and 6 and anti-parallel diodes D 6; Midpoint of each bridge arm a , b , c As AC output terminals, they are respectively connected to the corresponding phases of the three-phase asynchronous motor; the midpoint a , b , c The output currents are respectively i a , i b and i c ; i a , i b and i c That is, the three-phase stator current; characterized in that, The fault diagnosis method includes: Step S1: Obtain the three-phase stator current during each PWM interrupt cycle. i a , i b , i c And the PWM-modulated reference voltage signal, and determine the stator voltage vector required for the voltage model. ; Step S2: Establish a hybrid model flux linkage observer to calculate the current model flux linkage and the voltage model flux linkage separately, and use the deviation between the current model flux linkage and the voltage model flux linkage through a proportional-integral circuit to obtain the compensated voltage vector. To further obtain the high-pass and low-pass weighted relationship of the voltage model stator flux linkage after compensation by the hybrid model flux linkage observer; wherein, the hybrid model flux linkage observer is a combination of the current model flux linkage observer and the voltage model flux linkage observer; the current model flux linkage is the current model stator flux linkage, and the voltage model flux linkage includes the voltage model stator flux linkage and the voltage model rotor flux linkage; Step S3: Convert the compensation voltage vector using the inverse Clarke transform. Convert to three-phase compensation voltage ; Step S4: Calculate the flux linkage phase angle from the rotor flux linkage components of the voltage model. And define the phase angle of the magnetic flux linkage. The output range is 0 to ; Step S5: Utilize the magnetic flux phase angle Sweep starting from the designated starting angle The property corresponding to a complete fundamental period T is the characteristic of the integral average of a certain phase period. The time-domain integral is transformed into the phase-domain integral, and the periodic characteristics of the three-phase compensation voltage are extracted in the phase-domain. ; Step S6: Based on periodic characteristics Constructing three-phase absolute periodic characteristics and the first fault detection threshold The fault flag is obtained by comparison. ; Step S7: Construct the fault variable vector And based on the second fault detection threshold fault variable vector The mapping is 1, 0, or -1, and the faulty switching transistor is located according to the lookup table rules; wherein, the fault variable vector Three-phase fault variables composition.

2. The method for diagnosing open-circuit faults in voltage source inverters according to claim 1, characterized in that, In step S2, the current model flux linkage is a current model stator flux linkage, and the voltage model flux linkage includes a voltage model stator flux linkage and a voltage model rotor flux linkage. The expression corresponding to the stator flux linkage in the current model is: （1） In equation (1), This provides data for estimating stator flux linkage in a current model. For current model rotor flux estimation data; L m This refers to the excitation inductance parameter data; L r This refers to rotor inductance parameter data; L s These are the stator inductance parameter data; This is the stator current vector data; The expression corresponding to the stator flux linkage in the voltage model is: （2） In equation (2), Input data for the stator voltage vector; R s This refers to the stator resistance parameter data; This is the stator flux linkage estimation data for the voltage model; The expression corresponding to the rotor flux linkage in the voltage model is: （3） In equation (3), For voltage model rotor flux estimation data; The expression for the stator flux linkage in the compensated voltage model is: （4） In equation (4), The stator flux linkage data for the compensated voltage model; To compensate for voltage vector data; The compensation voltage vector The corresponding expression is: （5） In equation (5), These are proportional coefficient data; This is integral coefficient data; and The difference is the flux linkage error data; The expression for the high-pass and low-pass weighted relationship of the stator flux linkage in the voltage model after compensation by the hybrid model flux linkage observer is as follows: （6） In equation (6), This is the transfer function data for a second-order high-pass filter; This is the transfer function data for a second-order low-pass filter.

3. The method for diagnosing open-circuit faults in voltage source inverters according to claim 1, characterized in that, In step S3, the three-phase compensation voltage The corresponding expression is: （7） In equation (7), Compensation voltage Component data; These are the three-phase compensation voltage data.

4. The method for diagnosing open-circuit faults in voltage source inverters according to claim 1, characterized in that, In step S4, the flux linkage phase angle The corresponding expression is: （8） In equation (8), This is the flux linkage phase angle data; For the rotor flux linkage component data of the voltage model; when the two input components of atan2 When both are 0, Set to 0.

5. The method for diagnosing open-circuit faults in voltage source inverters according to claim 1, characterized in that, In step S5, the periodic integral average characteristic of a certain phase The corresponding expression is: （9） In equation (9), The periodic integral average characteristic data for a certain phase; T represents a complete fundamental frequency cycle; t1 represents the data at the start of integration; p represents the phase sequence, where p takes the values ​​a, b, and c, corresponding to phases A, B, and C, respectively. The periodic integral average characteristic of a certain phase The expression corresponding to converting the time-domain integral to the phase-domain integral is: （10） In equation (10), This is the flux linkage phase angle data; This is the angular span data of the magnetic flux phase angle sweeping around one revolution.

6. The method for diagnosing open-circuit faults in voltage source inverters according to claim 5, characterized in that, In step S5, periodic characteristics of the three-phase compensation voltage are extracted in the phase angle domain. Specifically, it includes: According to the sampling period T s Calculate the number of sampling points per cycle with the complete fundamental period T. ; Constructing the three-phase compensation voltage and flux phase angle Discrete sampling sequence; During each PWM interrupt cycle, the three-phase compensation voltage and flux phase angle Each sample value is obtained and entered into the compensation voltage sampling set and the phase angle sampling set respectively; Dividing 0 to 2π into n equal parts yields the equifluence phase angle step. and with Construct a discrete phase angle sequence with equal flux linkage phase angles for the initial phase angle; When the magnetic flux phase angle When sweeping through 2π, calculate the moving average of the phase angles within each flux linkage phase angle interval; The periodic characteristics are obtained by performing a second-order equal-phase-angle shift and summation on the mean characteristics of n phase angle intervals. ; Among them, the number of sampling points per cycle The corresponding expression is: （11） In equation (11), T s Data for the sampling period; This is data on the number of sampling points per cycle; Three-phase compensation voltage and flux phase angle The expression corresponding to the discrete sampling sequence is: （12） （13） In equations (12) and (13), k is the sampling time; For the sampling point index and Take (0, ); This is the compensation voltage for phase A; This is the compensation voltage for phase B; This is the C-phase compensation voltage; Phase angle of magnetic flux At time ( The sampled value of ) The expressions for the compensation voltage sampling set and the phase angle sampling set are as follows: （14） （15） In equations (14) and (15), To compensate for voltage sampling data; This is the phase angle sampling set data; Equal flux phase angle step The calculation expression is: （16） In equation (16), n is the number of equal divisions of the phase angle domain; For each segment of flux linkage phase angle, a fixed increment is used, i.e., from 0 to... The phase angle of each part after dividing it into n equal parts; The expression for the discrete phase angle sequence with equal flux linkage is: （17） In equation (17), This is the initial flux linkage phase angle data; The sequence number of the flux linkage phase angle step segment and Take (0, n-1); Indicates from Start by step size Phase angle points obtained by segment-by-segment discretization; The expression for the moving average of the phase angle within each flux linkage phase angle interval is: （18） In equation (18), This represents the number of sampling points falling within this phase angle interval; This represents the mean characteristic data for this phase angle interval; To compensate for voltage sampling data; Periodic characteristics The calculation expression is: （19） In equation (19), The periodic feature is obtained by summing the mean features of each small segment within the phase angle domain.

7. The method for diagnosing open-circuit faults in voltage source inverters according to claim 6, characterized in that, n is less than To ensure that within each phase angle interval Not equal to 0, and n is in The range is 10 to 15 percent.

8. The method for diagnosing open-circuit faults in voltage source inverters according to claim 1, characterized in that, In step S6, the three-phase absolute periodicity characteristics The expression is: （20） In equation (20), The absolute value data represents the three-phase periodic characteristics. Fault signs The expression is: （21） In equation (21), This is fault flag data; The first fault detection threshold data is used, and the G value in any phase exceeds... Time determination The indicator shows an open circuit fault in the switching transistor; The absolute periodic characteristics of the A-phase compensation voltage; The absolute periodic characteristics of the B-phase compensation voltage; The absolute periodic characteristics of the C-phase compensation voltage; First fault detection threshold By loading the maximum value of the transient mean feature Perform calibration. The range of values ​​is greater than and not less than To reduce the risk of false alarms.

9. The method for diagnosing open-circuit faults in voltage source inverters according to claim 1, characterized in that, In step S7, the fault variable vector The expression is: （22） In equation (22), This is data for three-phase fault variables; Based on the second fault detection threshold fault variable vector The expressions corresponding to mappings to 1, 0, or -1 are: （23） In equation (23), This is the second fault detection threshold data; This is periodic characteristic data; Second fault detection threshold Determined through experimental tuning, and based on the fault variable vector. The rules for determining the values ​​of the fault variable vector allow it to take the values ​​1, 0, or -1.

10. The method for diagnosing open-circuit faults in a voltage source inverter according to claim 1, characterized in that, In step S7, the mapping relationship corresponding to the lookup rule is Table 1; in Table 1, the output... Each switch corresponds one-to-one with the fault switch number, including the normal state and the fault state of switch numbers 1 to 6; 0 indicates no open circuit fault, and 1 or -1 indicates the presence of an open circuit fault. Table 1. Mapping Relationships Corresponding to Table Lookup Rules 。

Images