Sensor-based frequency converter operation fault diagnosis method and system

Abstract

The application relates to the technical field of fault diagnosis, in particular to a sensor-based frequency converter operation fault diagnosis method and system, which comprises the following steps: acquiring each-phase current sequences, temperature sequences and rotating speed sequences of each collection cycle of a frequency converter; acquiring current precursor degrees of each-phase current sequences of a current collection cycle according to fluctuation characteristics of the each-phase current sequences; acquiring load asynchronous coefficients of the each-phase current sequences of the current collection cycle in combination with short-time high-frequency disturbance characteristics in the each-phase current sequences of the current collection cycle and the correlation degree with rotating speed data, acquiring temperature precursor degrees of the current collection cycle in combination with fluctuation characteristics of temperature sequences of the current collection cycle, acquiring fault precursor coefficients of the current collection cycle, and then judging whether the frequency converter has faults. The application improves the accuracy of frequency converter fault diagnosis by analyzing the differential data characteristics when the motor load changes and when the IGBT breaks down.

Description

Technical Field

[0001] This application relates to the field of fault diagnosis technology, specifically to a sensor-based method and system for diagnosing faults in frequency converter operation. Background Technology

[0002] A frequency converter (FD) is an electronic device used to control the speed of an AC motor. It controls the motor's speed and operating mode by changing the frequency and voltage of the power supply, thus achieving frequency regulation of the motor. FDs offer advantages such as energy saving, precise control, and reduced mechanical wear, and are widely used in the power and industrial sectors due to their superior performance. However, with prolonged use and the aging of its components, FDs can malfunction, posing risks to industrial production and personal safety. Therefore, fault detection of FDs during real-time operation is necessary to improve the reliability and safety of industrial production.

[0003] IGBT (Insulated Gate Bipolar Transistor) modules are the core power components in frequency converters, and their performance and stability directly affect the converter's efficiency and lifespan. With frequent switching operations and heat accumulation, IGBT modules are prone to breakdown faults, thus affecting the normal operation of the frequency converter. Current technology for detecting IGBT module breakdown faults in frequency converters typically uses a single data point compared to a threshold to determine whether a fault has occurred. This approach does not fully consider the difference between current fluctuations caused by changes in motor load and the precursors to IGBT breakdown faults. Consequently, current fluctuations caused by changes in motor load are easily misdetected as precursors to IGBT breakdown faults, resulting in low accuracy in frequency converter fault diagnosis. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this application is to provide a sensor-based method and system for diagnosing inverter operation faults. The specific technical solution adopted is as follows:

[0005] In a first aspect, embodiments of this application provide a sensor-based method for diagnosing inverter operation faults, the method comprising the following steps:

[0006] Obtain the phase current sequence, temperature sequence, and speed sequence of the frequency converter in each acquisition cycle;

[0007] The data with a value of 0 in each phase current sequence are recorded as zero-point elements. Based on the slope of the fitted line corresponding to the peak value in each phase current sequence of the current acquisition period and the correlation of the data in the neighborhood window corresponding to all adjacent zero-point elements in each phase current sequence, the current precursor of each phase current sequence in the current acquisition period is obtained.

[0008] The current sequence of each phase is divided into multiple current subsequences, and the rotational speed subsequence with the same position sequence as each current subsequence is obtained. Based on the position sequence range corresponding to the neighborhood window of the zero-point element with high-frequency energy in each phase current sequence of the current period, the high-frequency subsequence of each phase current sequence of the current period is obtained. Based on the proportion of high-frequency subsequences with a smaller dispersion than the rotational speed sequence in all high-frequency subsequences of each phase current sequence of the current period, the correlation between the average level of each phase current subsequence of the current period and the average level of its corresponding rotational speed subsequence, and the current precursor of each phase current sequence, the load asynchrony coefficient of each phase current sequence of the current period is obtained.

[0009] Based on the proportion of negative numbers in the first-order difference sequence of the temperature sequence in the current acquisition period, the dispersion of the temperature sequence, and the difference between the average level of the temperature sequence in the current acquisition period and the previous acquisition period, the temperature precursor of the current acquisition period is obtained. Combined with the average level of the load asynchronous coefficient of all phase current sequences in the current acquisition period, the fault precursor coefficient of the current acquisition period is obtained, thereby determining whether the frequency converter has a fault.

[0010] Preferably, the process of obtaining the current precursor of each phase current sequence in the current acquisition period is as follows:

[0011] Based on the correlation of the data within the neighborhood window corresponding to all adjacent zero-point elements in each phase current sequence, the first correlation of each phase current sequence in the current acquisition period is obtained.

[0012] The formula for calculating the current precursor of each phase current sequence in the current acquisition period is: In the formula, The current precursor of the A-phase current sequence in the current acquisition period; This represents the slope of the fitted straight line corresponding to all peak values ​​in the A-phase current sequence during the current acquisition period. It is an exponential function with the natural constant e as its base. The first correlation of the A-phase current sequence in the current acquisition period. This is a preset constant.

[0013] Preferably, the process of obtaining the first correlation of each phase current sequence in the current acquisition period is as follows: construct a window centered on each zero element in each phase current sequence, the width of the window being an odd number closest to T / 2, where T is the average of the intervals between all adjacent peaks in each phase current sequence; calculate the correlation coefficient between all adjacent window data in each phase current sequence in the current acquisition period, and record the average of the absolute values ​​of all correlation coefficients as the first correlation of each phase current sequence in the current acquisition period.

[0014] Preferably, the process of obtaining the high-frequency subsequences of each phase current sequence in the current acquisition period is as follows: performing a Fourier transform on each phase current sequence in the current acquisition period to obtain the fundamental frequency of each phase current sequence; obtaining the spectrum sequence of the window centered on each zero element in each phase current sequence, and recording the sum of all energy components with frequencies greater than the corresponding fundamental frequency in the spectrum sequence of each window as the high-frequency energy value of each window; recording the zero elements corresponding to all windows with high-frequency energy values ​​greater than a preset segmentation threshold in each phase current sequence as local high-frequency zeros in each phase current sequence; and extracting the corresponding subsequence from the rotational speed sequence according to the position range of the window data of all local high-frequency zeros in each phase current sequence in the current acquisition period, and recording it as the high-frequency subsequence of each phase current sequence in the current acquisition period.

[0015] Preferably, the process for obtaining the load asynchronous coefficient of each phase current sequence in the current acquisition period is as follows:

[0016] The first ratio of each phase current sequence in the current acquisition period is obtained based on the proportion of high-frequency subsequences with a dispersion degree less than that of the rotational speed sequence among all high-frequency subsequences of each phase current sequence in the current acquisition period.

[0017] Calculate the mean of each current subsequence and the mean of each speed subsequence of each phase current sequence respectively, and construct the current mean sequence and speed mean sequence of each phase current according to the position order of each subsequence.

[0018] Calculate the load asynchrony coefficient for each phase current sequence in the current acquisition cycle: In the formula, The load asynchrony coefficient is the current sequence of phase A in the current acquisition period. This represents the current precursor of the A-phase current sequence in the current acquisition period. This is the absolute value of the correlation coefficient between the mean current sequence of phase A current and its mean rotational speed sequence during the current acquisition period. The first ratio of the A-phase current sequence in the current acquisition period.

[0019] Preferably, the process of obtaining the first ratio of each phase current sequence in the current acquisition period is as follows: calculate the variance of the rotational speed sequence in the current acquisition period, denoted as z; calculate the variance of each high-frequency subsequence of each phase current sequence in the current acquisition period, and record the proportion of data with variances less than z as the first ratio of each phase current sequence in the current acquisition period.

[0020] Preferably, the formula for calculating the temperature precursor of the current acquisition cycle is: In the formula, W is the temperature precursor of the current acquisition period; H is the ratio of the number of negative numbers in the first-order difference sequence of the temperature sequence in the current acquisition period to the total number of data in the first-order difference sequence; L is the variance of the temperature sequence in the current acquisition period; and J is the ratio of the mean of the temperature sequence in the previous acquisition period to the mean of the temperature sequence in the current acquisition period.

[0021] Preferably, the formula for calculating the fault precursor coefficient of the current acquisition cycle is: In the formula, R is the fault precursor coefficient for the current acquisition period. W represents the average load asynchronous coefficient of the current sequences of phases A, B, and C in the current acquisition period, and W represents the temperature precursor in the current acquisition period.

[0022] Preferably, the specific process for determining whether the frequency converter has malfunctioned is as follows: if the normalized fault precursor coefficient is greater than or equal to a preset threshold, then the frequency converter is determined to have malfunctioned; otherwise, the frequency converter is determined not to have malfunctioned.

[0023] Secondly, embodiments of this application also provide a sensor-based inverter operation fault diagnosis system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the sensor-based inverter operation fault diagnosis methods described above.

[0024] This application has at least the following beneficial effects:

[0025] This application addresses the problem of low detection accuracy in existing technologies due to insufficient consideration of the difference between current fluctuations caused by load and current fluctuations caused by IGBT breakdown precursors. It constructs a current precursor coefficient to reflect the current growth trend and the degree of zero-point waveform distortion, accurately identifying the current characteristics of IGBT breakdown precursors. It also constructs a load asynchrony coefficient to reflect the synchronicity difference between the inverter output current and the motor load, eliminating interference from normal load fluctuations in fault diagnosis. Furthermore, it constructs a temperature precursor coefficient to reflect whether the IGBT module temperature change conforms to the implicit temperature rise in the fault precursor stage, and uses this to construct a fault precursor coefficient. This allows for a comprehensive assessment of the inverter's operating status, achieving accurate diagnosis and early warning of IGBT breakdown precursors. Attached Figure Description

[0026] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 A flowchart illustrating the steps of a sensor-based inverter operation fault diagnosis method provided in one embodiment of this application;

[0028] Figure 2 A flowchart illustrating the acquisition of fault precursor coefficients for the current acquisition period, provided as an embodiment of this application. Detailed Implementation

[0029] To further illustrate the technical means and effects adopted by this application to achieve the intended inventive objective, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the sensor-based inverter operation fault diagnosis method and system proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0031] The following description, in conjunction with the accompanying drawings, details the specific scheme of the sensor-based inverter operation fault diagnosis method and system provided in this application.

[0032] Please see Figure 1 The diagram illustrates a flowchart of a sensor-based inverter operation fault diagnosis method according to an embodiment of this application. The method includes the following steps:

[0033] Step 1: Obtain the phase current sequence, temperature sequence, and speed sequence of each acquisition cycle of the frequency converter.

[0034] This application collects the output current data of the frequency converter through a three-phase current sensor installed at the output terminal of the frequency converter; acquires the temperature of the IGBT module in real time through a temperature sensor; and collects the speed data of the motor drive shaft connected to the frequency converter through a speed sensor to reflect the load changes of the motor. All data are collected synchronously and in real time, with a collection time interval of 1 second and a collection cycle of Y minutes, which is 30 in this embodiment. Based on the time sequence of data collection, the current sequence, temperature sequence, and speed sequence of phase A, phase B, and phase C for each collection cycle are constructed respectively.

[0035] To eliminate the influence of different dimensions among the data, all data are normalized. Normalization methods include Z-score, sigmoid normalization, and maximum / minimum value normalization. This embodiment uses Z-score normalization.

[0036] Step 2: Record the data with a value of 0 in each phase current sequence as zero-point elements. Based on the slope of the fitted line corresponding to the peak value in each phase current sequence of the current acquisition period and the correlation of the data in the neighborhood window corresponding to all adjacent zero-point elements in each phase current sequence, obtain the current precursor of each phase current sequence in the current acquisition period.

[0037] When the frequency converter is operating well, the output current waveform collected by the sensor should be a stable sine wave, and the temperature of the IGBT module should remain within a safe range due to the stable cooling system. However, if there are early signs of IGBT breakdown, the current waveform will show significant distortion and falsification due to the degradation of IGBT switching characteristics; at the same time, abnormal current conduction, partial discharge, or deterioration of insulation performance inside the module will cause the temperature of the IGBT module to rise.

[0038] However, considering that changes in motor load under normal circumstances can also cause changes in the output current and temperature of the frequency converter, it is difficult to distinguish between normal fluctuations or early signs of breakdown faults in the frequency converter by relying solely on fixed thresholds. Therefore, further in-depth analysis is required.

[0039] Due to the degradation of the switching characteristics of the damaged IGBT, the dynamic imbalance of the current vector will cause the faulty phase to conduct asymmetrically, which will increase the current and cause distortion. Therefore, the change of the three-phase current can be used to make a preliminary analysis of whether an abnormality has occurred.

[0040] The analysis will be based on the current sequence of phase A in the current acquisition period.

[0041] The peak values ​​in the phase A current sequence are obtained using a peak detection algorithm, and then all peak values ​​are linearly fitted using the least squares method. The peak detection algorithm is not limited to the AMPD peak detection algorithm or the maximum detection algorithm; this embodiment uses the AMPD peak detection algorithm.

[0042] Furthermore, the degradation of IGBT switching characteristics can lead to conduction delay or turn-off residue near the zero-crossing point of the current, manifesting as high-frequency glitches and waveform distortion.

[0043] Before normalization, the data with a value of 0 in the A-phase current sequence are recorded as zero-point elements. After normalization, a window is constructed with each zero-point element as the center. The width of the window is the odd number closest to T / 2, where T is the average of the intervals between all adjacent peaks in the A-phase current sequence.

[0044] The correlation coefficients between all adjacent window data in the A-phase current sequence of the current acquisition period are calculated, and the mean of the absolute values ​​of all correlation coefficients is recorded as the first correlation degree of the A-phase current sequence of the current acquisition period. The first correlation degree reflects whether waveform distortion occurs at zero points. The smaller the value, the greater the waveform difference at all zero points, and the less it conforms to the characteristics of a good current waveform. The algorithm for calculating the correlation coefficient is not limited to Pearson correlation coefficient, Spearman correlation coefficient, or cosine similarity; this embodiment uses the Pearson correlation coefficient.

[0045] As a preferred implementation, the current precursor degree of each phase current sequence in the current acquisition period is obtained based on the slope of the fitted line corresponding to the peak value in each phase current sequence of the current acquisition period and the correlation degree of the data in the neighborhood window corresponding to all adjacent zero-point elements in each phase current sequence. This is used to characterize the degree to which the current data of each phase in the current acquisition period conforms to the IGBT breakdown fault precursor.

[0046] In this embodiment, the current precursor of the A-phase current sequence in the current acquisition period is denoted as... Its specific expression is: In the formula, The current precursor of the A-phase current sequence in the current acquisition period; This represents the slope of the fitted straight line corresponding to all peak values ​​in the A-phase current sequence during the current acquisition period. It is an exponential function with the natural constant e as its base. The first correlation of the A-phase current sequence in the current acquisition period. As a preset constant, to avoid the denominator being 0, a value is taken from the empirical range (0.005, 0.01). The value has little impact on the calculation and can be ignored. The implementer can choose the value as needed.

[0047] The peak value of the A-phase current sequence in the current acquisition period can reflect whether the peak value has increased over time. The larger the value, the greater the increase in the A-phase current, and the greater the possibility of an IGBT breakdown fault. It can reflect whether there is an increase in current and distortion in the output current of the frequency converter during the current acquisition cycle. The larger the value, the more the current trend reflected by the A-phase current sequence is consistent with the precursor characteristics of IGBT breakdown fault, and the greater the possibility that there is an abnormality in the operating status of the frequency converter.

[0048] Step 3: Divide each phase current sequence into multiple current subsequences and obtain the speed subsequences with the same position sequence as each current subsequence; according to the position sequence range corresponding to the neighborhood window of the zero-point element with high-frequency energy in each phase current sequence of the current period, obtain the high-frequency subsequences of each phase current sequence of the current period; according to the proportion of high-frequency subsequences with a smaller dispersion than the speed sequence in all high-frequency subsequences of each phase current sequence of the current period, the correlation between the average level of each phase current subsequence of the current period and the average level of its corresponding speed subsequences, and the current precursor of each phase current sequence, obtain the load asynchrony coefficient of each phase current sequence of the current period.

[0049] Furthermore, considering that if the load of the motor connected to the frequency converter changes frequently, the output current of the frequency converter will also change frequently in order to make the motor run more stably. Therefore, analyzing the operating status of the frequency converter by only collecting output current changes from sensors may produce errors. It is necessary to combine the motor's operating data to achieve further analysis.

[0050] If no fault occurs, the inverter output current and motor speed will have synchronous variation characteristics; however, if the inverter shows signs of breakdown, the synchronous characteristics between the output current and motor speed will be disrupted.

[0051] Based on the peak value in the current phase A current sequence of the current acquisition cycle, the phase A current sequence is divided into multiple current subsequences, and based on the position range of each current subsequence, the rotational speed sequence is divided into multiple rotational speed subsequences.

[0052] The mean values ​​of each current subsequence and each speed subsequence are calculated separately, and the mean current sequence and mean speed sequence are constructed according to the sorting order of the subsequences. Since the current waveform is constantly changing, constructing the mean sequence can effectively remove the influence of instantaneous fluctuations, thereby extracting the steady changing trend of current and speed.

[0053] Furthermore, current fluctuations caused by load changes are usually low-frequency and smooth; however, as the IGBT gradually deteriorates, the current data will exhibit jitter, which manifests as short-term high-frequency disturbances, and these abnormal disturbances will not immediately cause a speed synchronization response.

[0054] The fundamental frequency of the A-phase current sequence in the current acquisition period is obtained by performing a Fourier transform. Then, within the A-phase current sequence, windows centered on each zero-point element are subjected to Fourier transforms to obtain the spectral sequence of each window. The sum of all energy components with frequencies greater than the fundamental frequency in the spectral sequence of each window is recorded as the high-frequency energy value of each window. The high-frequency energy values ​​of all windows in the A-phase current sequence are used as input to the Otsu thresholding method, which outputs a segmentation threshold. This segmentation threshold is recorded as the preset segmentation threshold. Zero-point elements corresponding to all windows with high-frequency energy values ​​greater than the preset segmentation threshold are recorded as local high-frequency zeros. The Otsu thresholding method can extract window data that generate high-frequency disturbances in the A-phase current sequence.

[0055] Based on the positional range of the window data of all local high-frequency zero points in the A-phase current sequence of the current acquisition period, the corresponding subsequence is extracted from the speed sequence and denoted as the high-frequency subsequence of the A-phase current sequence of the current acquisition period. The variance of the speed sequence of the current acquisition period is calculated and denoted as z. Then, the variance of each high-frequency subsequence of the A-phase current sequence of the current acquisition period is calculated separately, and the proportion of all data with variances less than z is denoted as the first ratio of the A-phase current sequence of the current acquisition period. The first ratio reflects whether the motor speed has a synchronous response during high-frequency disturbances in the current data. The larger the value, the more stable the speed is in the high-frequency subsequence, which indicates that the drastic fluctuations in the output current are more likely not caused by the motor speed, and the greater the possibility of inverter failure.

[0056] In a preferred embodiment, the load asynchrony coefficient of each phase current sequence in the current acquisition period is obtained based on the first ratio of each phase current sequence in the current acquisition period, the correlation between the average level of each phase current subsequence in the current acquisition period and the average level of its corresponding speed subsequence, and the current precursor of each phase current sequence. This coefficient is used to characterize the degree of asynchrony between each phase current and the motor load in the current acquisition period.

[0057] In this embodiment, the load asynchronous coefficient of the A-phase current sequence in the current acquisition period is denoted as... Its expression is: In the formula, The load asynchrony coefficient is the current sequence of phase A in the current acquisition period. This represents the current precursor of the A-phase current sequence in the current acquisition period. This is the absolute value of the correlation coefficient between the mean current sequence of phase A current and its mean rotational speed sequence during the current acquisition period. The first ratio of the A-phase current sequence in the current acquisition period.

[0058] The variable frequency drive (VFD) output current and motor speed are synchronous. A smaller value indicates a weaker synchronous characteristic, suggesting a higher probability of abnormal VFD operation. The load asynchrony coefficient comprehensively reflects the asynchronous relationship between the VFD output current and the motor load. A larger value indicates greater abnormal fluctuations in the VFD output current, indicating a lack of synchronization with the motor load and a higher probability of abnormal VFD operation.

[0059] Similarly, the load asynchrony coefficients of the B-phase current sequence and the C-phase current sequence can be obtained.

[0060] Step 4: Based on the proportion of negative numbers in the first-order difference sequence of the temperature sequence in the current acquisition period, the dispersion of the temperature sequence, and the difference between the average level of the temperature sequence in the current acquisition period and the previous acquisition period, obtain the temperature precursor of the current acquisition period. Combined with the average level of the load asynchronous coefficient of each phase current sequence in the current acquisition period, obtain the fault precursor coefficient of the current acquisition period, and then determine whether the frequency converter has a fault.

[0061] Furthermore, before the IGBT module inside the inverter experiences a breakdown fault, not only will the output current fluctuate, but the temperature of the IGBT module will also rise. Therefore, a comprehensive analysis of the inverter can be performed by combining temperature data collected by temperature sensors. However, considering that no fault has occurred at this stage, it is still in the pre-fault stage, and the temperature change has a lag effect, not as significant as the change in current data. Therefore, the temperature may be difficult to identify due to the normal operation of the inverter's cooling system, meaning the inverter temperature has not exceeded the set temperature threshold. However, if it is in the pre-fault stage, even if the temperature has not reached the threshold, the characteristics of the temperature data change will still change. For example, the rate of temperature increase will be faster, and as time goes on, the average temperature of the inverter will be higher, and the cooling system will be activated more frequently, resulting in larger temperature fluctuations and no longer being relatively stable.

[0062] As a preferred implementation, the temperature precursor of the current acquisition period is obtained based on the proportion of negative numbers in the first-order difference sequence of the temperature sequence in the current acquisition period, the dispersion of the temperature sequence, and the difference between the average level of the temperature sequence in the current acquisition period and the previous acquisition period. This precursor is used to characterize the possibility of anomalies in the temperature data within the current acquisition period.

[0063] In this embodiment, the temperature precursor of the temperature sequence in the current acquisition period is denoted as... Its specific expression is: In the formula, W is the temperature precursor of the current acquisition period; H is the ratio of the number of negative numbers in the first-order difference sequence of the temperature sequence in the current acquisition period to the total number of data in the first-order difference sequence; L is the variance of the temperature sequence in the current acquisition period; and J is the ratio of the mean of the temperature sequence in the previous acquisition period to the mean of the temperature sequence in the current acquisition period.

[0064] H reflects the frequency of temperature decreases in the current cycle; the higher the value, the more frequently the inverter's cooling system adjusts. J reflects whether the overall temperature has increased compared to the previous period; the lower the value, the higher the average temperature of the current period. L reflects the degree of temperature fluctuation in the current period; the higher the value, the greater the temperature fluctuation. Temperature precursor W reflects whether the temperature change of the inverter's internal IGBT module has been abnormal in the current period; the higher the value, the greater the possibility of abnormal temperature data changes in the current period.

[0065] Furthermore, based on the load asynchronous coefficients of the current sequences of phases A, B, and C in the current acquisition cycle, and the temperature precursor coefficients in the current acquisition cycle, the fault precursor coefficients for the current acquisition cycle are obtained. These coefficients are used to characterize the probability that the operating data characteristics of the IGBT modules inside the inverter in the current acquisition cycle belong to a breakdown fault.

[0066] In a preferred implementation, a fault precursor coefficient for the current acquisition period is obtained based on the temperature precursor and the average level of the load asynchrony coefficient of all phase current sequences. This coefficient characterizes the likelihood of a breakdown fault occurring in the inverter during the current acquisition period. The flowchart for obtaining the fault precursor coefficient for the current acquisition period is shown below. Figure 2 As shown.

[0067] In this embodiment, the fault precursor coefficient of the current acquisition period is denoted as... Its specific expression is: In the formula, R is the fault precursor coefficient for the current acquisition period. W represents the average load asynchronous coefficient of the current sequences of phases A, B, and C in the current acquisition period, and W represents the temperature precursor in the current acquisition period.

[0068] The fault precursor coefficient comprehensively considers the changes in current load and temperature of the frequency converter, and can fully analyze the health status of the frequency converter. The larger the value, the more the operating data characteristics of the IGBT module inside the frequency converter are consistent with the data characteristics before the breakdown fault, and the greater the possibility of abnormal operation of the frequency converter.

[0069] The fault precursor coefficients for the current acquisition cycle are normalized using the sigmoid function, mapping them to the range [0,1]. If the normalized fault precursor coefficient is greater than or equal to a preset threshold, the inverter is considered to have a fault. The more the operating data collected by the sensors in the current acquisition cycle matches the precursor characteristics of an IGBT breakdown fault, the more necessary it is to repair the inverter promptly. If the normalized fault precursor coefficient is less than the preset threshold, the inverter is considered to be in good health, showing no signs of failure, and can continue to operate. In this embodiment, the preset threshold is 0.5, but the implementer can choose a value according to the actual situation.

[0070] Based on the same inventive concept as the above methods, this application also provides a sensor-based inverter operation fault diagnosis system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above sensor-based inverter operation fault diagnosis methods.

[0071] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0072] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0073] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.

Claims

1. A sensor-based method for diagnosing inverter operation faults, characterized in that, The method includes the following steps: Obtain the phase current sequence, temperature sequence, and speed sequence of the frequency converter in each acquisition cycle; The data with a value of 0 in each phase current sequence are recorded as zero-point elements. Based on the slope of the fitted line corresponding to the peak value in each phase current sequence of the current acquisition period and the correlation of the data in the neighborhood window corresponding to all adjacent zero-point elements in each phase current sequence, the current precursor of each phase current sequence in the current acquisition period is obtained. The current sequence of each phase is divided into multiple current subsequences, and the rotational speed subsequence with the same position sequence as each current subsequence is obtained. Based on the position sequence range corresponding to the neighborhood window of the zero-point element with high-frequency energy in each phase current sequence of the current period, the high-frequency subsequence of each phase current sequence of the current period is obtained. Based on the proportion of high-frequency subsequences with a smaller dispersion than the rotational speed sequence in all high-frequency subsequences of each phase current sequence of the current period, the correlation between the average level of each phase current subsequence of the current period and the average level of its corresponding rotational speed subsequence, and the current precursor of each phase current sequence, the load asynchrony coefficient of each phase current sequence of the current period is obtained. Based on the proportion of negative numbers in the first-order difference sequence of the temperature sequence in the current acquisition period, the dispersion of the temperature sequence, and the difference between the average level of the temperature sequence in the current acquisition period and the previous acquisition period, the temperature precursor of the current acquisition period is obtained. Combined with the average level of the load asynchronous coefficient of all phase current sequences in the current acquisition period, the fault precursor coefficient of the current acquisition period is obtained, thereby determining whether the frequency converter has a fault.

2. The sensor-based inverter operation fault diagnosis method as described in claim 1, characterized in that, The process for obtaining the current precursor of each phase current sequence in the current acquisition period is as follows: Based on the correlation of the data within the neighborhood window corresponding to all adjacent zero-point elements in each phase current sequence, the first correlation of each phase current sequence in the current acquisition period is obtained. The formula for calculating the current precursor of each phase current sequence in the current acquisition period is: In the formula, The current precursor of the A-phase current sequence in the current acquisition period; This represents the slope of the fitted straight line corresponding to all peak values ​​in the A-phase current sequence during the current acquisition period. It is an exponential function with the natural constant e as its base. The first correlation of the A-phase current sequence in the current acquisition period. This is a preset constant.

3. The sensor-based inverter operation fault diagnosis method as described in claim 2, characterized in that, The process of obtaining the first correlation of each phase current sequence in the current acquisition period is as follows: construct a window with each zero element in each phase current sequence as the center, the width of the window is the odd number closest to T / 2, and T is the mean of the interval between all adjacent peaks in each phase current sequence; calculate the correlation coefficient between all adjacent window data in each phase current sequence in the current acquisition period, and record the mean of the absolute values ​​of all correlation coefficients as the first correlation of each phase current sequence in the current acquisition period.

4. The sensor-based inverter operation fault diagnosis method as described in claim 3, characterized in that, The process of obtaining the high-frequency subsequences of each phase current sequence in the current acquisition period is as follows: Perform a Fourier transform on each phase current sequence in the current acquisition period to obtain the fundamental frequency of each phase current sequence; obtain the spectral sequence of the window centered on each zero-point element in each phase current sequence; record the sum of all energy components with frequencies greater than the corresponding fundamental frequency in the spectral sequence of each window as the high-frequency energy value of each window; record the zero-point elements corresponding to all windows with high-frequency energy values ​​greater than a preset segmentation threshold in each phase current sequence as local high-frequency zeros in each phase current sequence; extract the corresponding subsequence from the rotational speed sequence based on the positional range of the window data of all local high-frequency zeros in each phase current sequence in the current acquisition period, and record it as the high-frequency subsequence of each phase current sequence in the current acquisition period.

5. The sensor-based inverter operation fault diagnosis method as described in claim 1, characterized in that, The process for obtaining the load asynchronous coefficient of each phase current sequence in the current acquisition period is as follows: The first ratio of each phase current sequence in the current acquisition period is obtained based on the proportion of high-frequency subsequences with a dispersion degree less than that of the rotational speed sequence among all high-frequency subsequences of each phase current sequence in the current acquisition period. Calculate the mean of each current subsequence and the mean of each speed subsequence of each phase current sequence respectively, and construct the current mean sequence and speed mean sequence of each phase current according to the position order of each subsequence. Calculate the load asynchrony coefficient for each phase current sequence in the current acquisition cycle: In the formula, The load asynchrony coefficient is the current sequence of phase A in the current acquisition period. This represents the current precursor of the A-phase current sequence in the current acquisition period. This is the absolute value of the correlation coefficient between the mean current sequence of phase A current and its mean rotational speed sequence during the current acquisition period. The first ratio of the A-phase current sequence in the current acquisition period.

6. The sensor-based inverter operation fault diagnosis method as described in claim 5, characterized in that, The process of obtaining the first ratio of each phase current sequence in the current acquisition period is as follows: calculate the variance of the rotation speed sequence in the current acquisition period, denoted as z; calculate the variance of each high-frequency subsequence of each phase current sequence in the current acquisition period, and record the proportion of data with variances less than z as the first ratio of each phase current sequence in the current acquisition period.

7. The sensor-based inverter operation fault diagnosis method as described in claim 1, characterized in that, The formula for calculating the temperature precursor of the current acquisition cycle is: In the formula, W is the temperature precursor of the current acquisition period; H is the ratio of the number of negative numbers in the first-order difference sequence of the temperature sequence to the total number of data in the first-order difference sequence in the current acquisition period. L represents the variance of the temperature sequence in the current acquisition period; J represents the ratio of the mean of the temperature sequence in the previous acquisition period to the mean of the temperature sequence in the current acquisition period.

8. The sensor-based inverter operation fault diagnosis method as described in claim 1, characterized in that, The formula for calculating the fault precursor coefficient in the current acquisition period is: In the formula, R is the fault precursor coefficient for the current acquisition period. W represents the average load asynchronous coefficient of the current sequences of phases A, B, and C in the current acquisition period, and W represents the temperature precursor in the current acquisition period.

9. The sensor-based inverter operation fault diagnosis method as described in claim 1, characterized in that, The specific process for determining whether the frequency converter has malfunctioned is as follows: if the normalized fault precursor coefficient is greater than or equal to the preset threshold, the frequency converter is determined to have malfunctioned; otherwise, the frequency converter is determined not to have malfunctioned.

10. A sensor-based inverter operation fault diagnosis system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the sensor-based inverter operation fault diagnosis method as described in any one of claims 1-9.

Images