Remote fault diagnosis method and system of frequency converter combined with internet of things

Abstract

The application discloses a frequency converter remote fault diagnosis method and system combined with the Internet of Things and relates to the field of electrical measurement. The method comprises the following steps: receiving a frequency converter control mode for calibration, obtaining reference current fluctuation information and reference voltage fluctuation information; performing dimensionality reduction reconstruction on the reference current fluctuation information and the reference voltage fluctuation information to generate a reference current feature matrix and a reference voltage feature matrix; obtaining a comparison current feature matrix and a comparison voltage feature matrix; obtaining a fault prediction probability through a fault probability prediction component; and when the fault prediction probability is greater than or equal to a fault probability threshold, generating a fault early warning signal and sending the signal to a user end. The method solves the technical problem that, in the prior art, fault analysis needs to be performed according to complex semantic monitoring data of a frequency converter, the hierarchy is more, the complexity is greater, and the amount of calculation is larger, achieves the technical effect of reducing the amount of calculation and efficiently and quickly completing frequency converter fault diagnosis.

Description

Technical Field

[0001] This application relates to the field of electrical measurement technology, and more particularly to the field of electrical measurement diagnosis, specifically to a method and system for remote fault diagnosis of frequency converters combined with the Internet of Things. Background Technology

[0002] As a core piece of equipment in industrial production, the stability and reliability of frequency converters directly affect the operating efficiency of production lines and product quality. Therefore, remote fault diagnosis methods for frequency converters combined with the Internet of Things (IoT) have significant application value. Traditional frequency converter fault diagnosis methods mainly rely on the experience and technical skills of on-site personnel, identifying potential faults through manual inspections and periodic maintenance. However, this method has many shortcomings, such as low diagnostic efficiency, low accuracy, and slow response speed.

[0003] In summary, existing technologies suffer from the problem of requiring fault analysis based on complex semantic monitoring data of frequency converters, which involves multiple layers and significant complexity, resulting in a large computational burden. Summary of the Invention

[0004] Therefore, it is necessary to provide a method and system for remote fault diagnosis of frequency converters that integrates the Internet of Things, which can reduce the amount of computation and efficiently and quickly complete the fault diagnosis of frequency converters, in order to address the above-mentioned technical problems.

[0005] Firstly, a remote fault diagnosis method for frequency converters combined with the Internet of Things (IoT) is provided, comprising: receiving frequency converter control modes; calibrating the frequency converter control modes to obtain reference current fluctuation information and reference voltage fluctuation information; performing dimensionality reduction and reconstruction on the reference current fluctuation information to generate a reference current feature matrix; performing dimensionality reduction and reconstruction on the reference voltage fluctuation information to generate a reference voltage feature matrix; receiving monitoring current fluctuation information and monitoring voltage fluctuation information and performing dimensionality reduction and reconstruction to obtain a comparison current feature matrix and a comparison voltage feature matrix; analyzing the reference current feature matrix, the comparison current feature matrix, the reference voltage feature matrix, and the comparison voltage feature matrix through a fault probability prediction component to obtain a fault prediction probability; and generating a fault warning signal and sending it to the user terminal when the fault prediction probability is greater than or equal to a fault probability threshold.

[0006] Secondly, a remote fault diagnosis system for frequency converters integrating the Internet of Things (IoT) is provided, comprising: a frequency converter control mode receiving module for receiving frequency converter control modes; a calibration module for calibrating the frequency converter control modes to obtain reference current fluctuation information and reference voltage fluctuation information; a reference current feature matrix generation module for performing dimensionality reduction and reconstruction of the reference current fluctuation information to generate a reference current feature matrix; and a reference voltage feature matrix generation module for performing dimensionality reduction and reconstruction of the reference voltage fluctuation information to generate a reference voltage feature matrix. The system includes: a quasi-voltage feature matrix; a dimensionality reduction and reconstruction module, which receives monitoring current fluctuation information and monitoring voltage fluctuation information, performs dimensionality reduction and reconstruction, and obtains a comparison current feature matrix and a comparison voltage feature matrix; a fault prediction probability module, which analyzes the reference current feature matrix, the comparison current feature matrix, the reference voltage feature matrix, and the comparison voltage feature matrix through a fault probability prediction component to obtain a fault prediction probability; and a fault warning signal generation module, which generates a fault warning signal and sends it to the user terminal when the fault prediction probability is greater than or equal to a fault probability threshold.

[0007] Thirdly, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps described in the first aspect.

[0008] Fourthly, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps described in the first aspect.

[0009] The above-mentioned method and system for remote fault diagnosis of frequency converters combined with the Internet of Things solves the technical problem in the prior art that fault analysis is required based on complex semantic monitoring data of frequency converters, which has many layers and great complexity, resulting in a large amount of analysis calculation. This method achieves the technical effect of reducing the amount of calculation and completing the fault diagnosis of frequency converters efficiently and quickly.

[0010] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating a remote fault diagnosis method for frequency converters that incorporates the Internet of Things in one embodiment.

[0012] Figure 2 This is a schematic diagram illustrating the process of obtaining reference current fluctuation information and reference voltage fluctuation information in a remote fault diagnosis method for frequency converters that incorporates the Internet of Things in one embodiment.

[0013] Figure 3 This is a block diagram of a remote fault diagnosis system for frequency converters that incorporates the Internet of Things in one embodiment.

[0014] Figure 4 This is an internal structural diagram of a computer device in one embodiment.

[0015] Figure labeling: 11 Inverter control mode receiving module, 12 Calibration module, 13 Reference current feature matrix generation module, 14 Reference voltage feature matrix generation module, 15 Dimension reduction and reconstruction module, 16 Fault prediction probability module, 17 Fault early warning signal generation module. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0017] like Figure 1 As shown, this application provides a remote fault diagnosis method for frequency converters combined with the Internet of Things, including:

[0018] Receive inverter control modes;

[0019] Traditional inverter fault analysis requires fault parsing based on complex semantic monitoring data of the inverter, resulting in multiple layers and significant complexity, leading to a large computational load. This application provides an inverter remote fault diagnosis method combined with the Internet of Things (IoT). By integrating IoT with remote inverter fault diagnosis methods, it enables remote monitoring and fault diagnosis of the inverter's operating status. The IoT acquires various operating parameters of the inverter in real time through sensor technology, such as voltage, current, temperature, and vibration. This data is processed by the transmission layer and then sent to the control center. At the control center, through big data analysis and fault diagnosis algorithms, the operating status of the inverter can be monitored in real time, and alarms can be issued promptly when anomalies are detected. This achieves efficient and accurate analysis of inverter faults, reduces the computational load, and improves the efficiency and accuracy of fault diagnosis.

[0020] A frequency converter is a power control device that controls the speed, torque, or other process parameters of a motor by changing the frequency and voltage of the power supply. Based on the control mode of the frequency converter, corresponding input parameters are configured. One control mode corresponds to a set of control parameters, including voltage range, signal type, etc., thus acquiring the preset control parameters of the frequency converter. Receiving the frequency converter's control mode provides a foundation for subsequent remote fault diagnosis methods for the frequency converter.

[0021] The inverter control mode is calibrated to obtain reference current fluctuation information and reference voltage fluctuation information;

[0022] Choose a suitable calibration environment to ensure that external factors such as ambient temperature and humidity do not affect the calibration results. Determine the load type and load conditions for calibration, such as using a standard motor or simulated load. Based on the characteristics of the control mode, set the corresponding operating parameters, such as frequency, voltage, and current. For current and voltage, set reasonable reference values ​​according to the inverter's working principle and application requirements. Run the inverter under control mode and record the current and voltage fluctuation data in real time. Perform statistical analysis on the collected current and voltage fluctuation data, extract key features, identify and eliminate abnormal data, and ensure the reliability of the calibration results. Obtain reference current fluctuation information and reference voltage fluctuation information. Reference current fluctuation information refers to the stable fluctuation range of the inverter's output current under specific conditions during normal operation. This range is usually determined through the calibration process to ensure that the inverter can output a stable and expected current under normal operating conditions. Reference voltage fluctuation information refers to the stable fluctuation range of the inverter's output voltage under specific operating conditions during normal operation. This information is obtained through calibration and analysis to ensure that the inverter can provide a stable and required output voltage under the predetermined operating mode. By calibrating the control mode of the frequency converter, reference current fluctuation information and reference voltage fluctuation information of the frequency converter under a specific control mode can be obtained, ensuring the stable operation of the frequency converter, optimizing performance, and laying the groundwork for subsequent fault diagnosis.

[0023] like Figure 2 As shown, positive sample backtracking is performed based on the inverter control mode to obtain current fluctuation record information set and voltage fluctuation record information set, where positive sample refers to fault-free sample of the same inverter control mode;

[0024] The reference current fluctuation information is obtained by performing feature hierarchical aggregation on the current fluctuation record information set.

[0025] The voltage fluctuation record information set is subjected to feature hierarchical aggregation to obtain the reference voltage fluctuation information.

[0026] Positive sample backtracking refers to collecting current and voltage fluctuation records from fault-free samples under the same inverter control mode to form current fluctuation record sets and voltage fluctuation record sets. This ensures the collected data comes from inverters of the same model and under the same control mode, verifying that the data originates from inverter operation records under fault-free conditions. During normal inverter operation, current measurement devices are used to record changes in output current in real time. Based on these changes, a current fluctuation record set is generated, containing the current value at each time point and its corresponding timestamp. Similarly, during normal inverter operation, voltage measurement devices are used to record changes in output voltage in real time. Based on these changes, a voltage fluctuation record set is generated, containing the voltage value at each time point and its corresponding timestamp. Feature hierarchical aggregation includes feature extraction and feature fusion. By aggregating at different levels, the inherent structure and patterns of the data can be better understood and described. Preprocessing of the current fluctuation record set, such as outlier removal and smoothing, extracts key features of the current fluctuation, such as mean, standard deviation, and peak value. For each feature, feature aggregation is performed to describe the shape and characteristics of the current distribution, obtaining reference current fluctuation information. Similarly, reference voltage fluctuation information is obtained using the same method. Obtaining accurate reference current and voltage fluctuation information provides fundamental support for subsequent performance evaluation, fault diagnosis, and optimization of the frequency converter.

[0027] Based on the first current fluctuation record information in the current fluctuation record information set, extract the amplitude record information set, frequency record information set, phase record information set, kurtosis record information set, skewness record information set, year-on-year change rate record information set, and month-on-month change rate record information set from the first current fluctuation record information set. The year-on-year change rate represents the current amplitude change rate at an interval of one cycle, and the month-on-month change rate represents the current amplitude change rate of adjacent currents in the same cycle.

[0028] Traverse the amplitude record information set, the frequency record information set, the phase record information set, and the kurtosis record information set to perform first-type feature hierarchical aggregation to obtain the reference current amplitude, reference current frequency, reference current phase, and reference current kurtosis;

[0029] The second type of feature hierarchy aggregation is performed by traversing the skewness record information set, the year-on-year change rate record information set, and the month-on-month change rate record information set to obtain the reference current skewness, the reference current year-on-year change rate, and the reference current month-on-month change rate.

[0030] The reference current amplitude, reference current frequency, reference current phase, reference current kurtosis, reference current skewness, year-on-year change rate of reference current, and year-on-year change rate of reference current are added to the reference current fluctuation information.

[0031] The amplitude data of the current is extracted from the first current fluctuation record information, wherein the first current fluctuation record information refers to the arbitrarily selected record from the set of current fluctuation record information. Based on the amplitude record information set, frequency record information set, phase record information set, kurtosis record information set, skewness record information set, year-on-year change rate record information set, and month-on-month change rate record information set of the first current fluctuation record information, the amplitude data of the current is extracted to form the amplitude record information set, the amplitude reflects the magnitude of the current. The frequency data of the current is extracted to form the frequency record information set. Frequency represents the rate of change of current. Phase data of the current is extracted to form a phase record information set. Phase reflects the time offset between the current and the reference signal. Kurtosis of the current is calculated to form a kurtosis record information set. Kurtosis describes the sharpness of the current distribution pattern. Skewness of the current is calculated to form a skewness record information set. Skewness describes the degree of skewness of the current distribution pattern. The rate of change of current amplitude over one cycle is calculated to form a year-on-year rate of change record information set. The year-on-year rate of change is used to compare current changes over different cycles. The rate of change of current amplitude between adjacent currents within the same cycle is calculated to form a cycle-on-cycle rate of change record information set. The cycle-on-cycle rate of change is used to compare current changes between adjacent cycles. The year-on-year rate of change characterizes the current changes over one cycle. The current amplitude change rate over a period of time is used, and the year-on-year change rate represents the change rate of adjacent current amplitudes within the same period. A first type of feature-level aggregation is performed by traversing the amplitude record set, the frequency record set, the phase record set, and the kurtosis record set. This first type of feature-level aggregation refers to the aggregation of traditional feature information, i.e., basic features, such as amplitude, frequency, and phase. This includes calculating the mean, median, mode, or other statistics to reflect the overall situation of these features. The aggregated results include the reference current amplitude, reference current frequency, reference current phase, and reference current kurtosis, respectively. The first type of feature-level aggregation focuses on the aggregation of basic features. A second type of feature-level aggregation is performed by traversing the skewness record set, the year-on-year change rate record set, and the year-on-year change rate record set. This second type of feature-level aggregation focuses on capturing the complex relationships of the distribution and changing trends of these features. The aggregated results are the reference current skewness, the reference current year-on-year change rate, and the reference current year-on-year change rate, respectively. Compared to the first type of feature hierarchy aggregation, this method adds year-on-year and month-on-month change rates, enabling analysis of refined current amplitude variations and improving the accuracy of fault analysis. All the aforementioned benchmark feature values, including benchmark current amplitude, benchmark current frequency, benchmark current phase, benchmark current kurtosis, benchmark current skewness, year-on-year change rate, and month-on-month change rate, are integrated into the benchmark current fluctuation information. This provides a comprehensive description of current fluctuation characteristics for subsequent performance evaluation, fault diagnosis, and optimized control strategies.

[0032] Cluster analysis is performed on the amplitude recording information set based on the amplitude deviation threshold to obtain multi-cluster amplitude recording information, wherein the multi-cluster amplitude recording information has multiple trigger frequencies within the first cluster;

[0033] Based on the multiple trigger frequencies within the first cluster, the average value of the intra-cluster amplitude records with trigger frequencies greater than or equal to the threshold of the trigger frequency within the first cluster is calculated to obtain several intra-cluster amplitude characteristic values, which are then added to the reference current amplitude.

[0034] The calculation process for the frequency recording information set, the phase recording information set, and the kurtosis recording information set is the same as that for the amplitude recording information set.

[0035] Based on practical application requirements and the characteristics of current fluctuations, an amplitude deviation threshold is set, which can be determined by those skilled in the art, to determine which amplitude records belong to the same cluster, i.e., have similar amplitude ranges. A hierarchical clustering algorithm is used to cluster the amplitude record information set. During clustering, similar amplitude records are grouped into one cluster according to the amplitude deviation threshold, resulting in multi-cluster amplitude record information. Each cluster represents a specific amplitude range, and the multi-cluster amplitude record information has multiple intra-cluster trigger frequencies. For each cluster of amplitude record information, its intra-cluster trigger frequency is calculated, which can be defined as the number of times the amplitude record appears within that cluster. This is used to identify which clusters are the main, frequent amplitude fluctuation patterns. An intra-cluster trigger frequency threshold is set, which can be determined by those skilled in the art; only clusters with an intra-cluster trigger frequency greater than or equal to this threshold are further processed. For clusters that meet the intra-cluster trigger frequency threshold, the mean of their intra-cluster amplitude is calculated, representing a typical value of the current amplitude within that cluster. This mean value can be used as an intra-cluster amplitude feature value and added to the reference current amplitude. For frequency, phase, and kurtosis record sets, the same processing procedure as for amplitude record sets can be used. First, cluster analysis is performed based on the corresponding deviation thresholds, and then feature values ​​are extracted based on the trigger frequency within each cluster. These feature values ​​are added to the reference current frequency, reference current phase, and reference current kurtosis, respectively. Since amplitude, frequency, phase, and kurtosis are the most fundamental features, all feature values ​​with a frequency higher than a certain threshold are counted as evaluation benchmarks. All extracted benchmark feature values, including amplitude, frequency, phase, and kurtosis, are integrated to form complete reference current fluctuation information, which is used for subsequent performance evaluation, fault diagnosis, or control strategy optimization. This method based on cluster analysis and intra-cluster trigger frequency allows for more effective extraction of representative feature values ​​from a large number of current fluctuation records, providing strong support for inverter performance analysis and optimization.

[0036] Perform central tendency analysis on the skewness record information set to obtain the central tendency value of the skewness record information;

[0037] Cluster analysis is performed on the concentrated values ​​of the skewness record information based on the skewness deviation threshold to obtain multi-cluster skewness record information.

[0038] The intra-cluster mean is calculated by traversing the multi-cluster skewness record information to obtain multiple intra-cluster skewness feature values;

[0039] Based on the maximum and minimum values ​​of the multiple intra-cluster skewness characteristic values, a skewness fluctuation constraint interval is constructed and set as the reference current skewness.

[0040] The calculation process for the year-on-year change rate of the reference current and the month-on-month change rate of the reference current is the same as that for the skewness record information set.

[0041] Central tendency analysis of skewness records involves calculating statistical measures such as the mean and median of skewness to understand the distribution of skewness data. The median reflects the central location of the dataset and is not affected by extreme values. Cluster analysis is performed on the concentrated values ​​of skewness records based on the skewness deviation threshold. Cluster analysis groups similar skewness values ​​into one category, forming multi-cluster skewness records. The skewness deviation threshold is based on the actual situation of the data and the purpose of analysis to ensure sufficient similarity of skewness values ​​within a cluster. Each skewness cluster is traversed, and the mean of the skewness values ​​within each cluster is calculated. This mean represents the typical value of the skewness within the cluster and can be considered as a characteristic value of the skewness within the cluster, obtaining multiple characteristic values ​​of skewness within the cluster. Based on all the characteristic values ​​of skewness within the cluster, the maximum and minimum values ​​are taken to construct a skewness fluctuation constraint interval. This skewness fluctuation constraint interval reflects the fluctuation range of the skewness value within the entire dataset and is set as the reference current skewness. Following the above method, the year-on-year change rate and the month-on-month change rate of the reference current are processed to obtain the year-on-year change rate and the month-on-month change rate of the reference current, respectively. The processed reference current skewness, the year-on-year change rate, and the month-on-month change rate of the reference current are integrated together and can be used for subsequent comprehensive evaluation of inverter performance and fault prediction.

[0042] The reference current fluctuation information is reconstructed by dimensionality reduction to generate a reference current feature matrix;

[0043] The reference voltage fluctuation information is reconstructed by dimensionality reduction to generate a reference voltage feature matrix;

[0044] The reference current fluctuation information includes the reference current amplitude, reference current frequency, reference current phase, reference current kurtosis, reference current skewness, year-on-year change rate, and year-on-year change rate. Data preprocessing is performed on the reference current fluctuation information to eliminate the influence of dimensions and ranges between different features, ensuring that each feature has the same unit during dimensionality reduction. The reference current fluctuation information is then dimensionality-reduced using a dimensionality reduction reconstruction function to obtain a dimensionality-reduced current feature vector. This dimensionality-reduced current feature vector is integrated to obtain a matrix form, i.e., the reference current feature matrix. The reference voltage fluctuation information is then dimensionality-reduced and reconstructed using the same method to obtain the reference voltage feature matrix. After dimensionality reduction and reconstruction, the reference current feature matrix and the reference voltage feature matrix will serve as the basis for subsequent analysis and modeling, enabling tasks such as fault prediction, performance evaluation, and anomaly detection.

[0045] Receive monitoring current fluctuation information and monitoring voltage fluctuation information, perform dimensionality reduction and reconstruction, and obtain the comparison current feature matrix and the comparison voltage feature matrix;

[0046] The system receives real-time current and voltage fluctuation information from the monitoring system, which is the current and voltage fluctuation information of the inverter acquired by the monitoring device. This includes real-time values, rates of change, and fluctuation ranges. The system preprocesses this information and extracts key features, including the average value, standard deviation, peak value, skewness, and kurtosis of the current and voltage. It then performs dimensionality reduction on the monitoring current and voltage fluctuation information using a dimensionality reduction reconstruction function to obtain low-dimensional feature vectors. These reduced-dimensional current and voltage feature vectors are integrated into a comparison current feature matrix and a comparison voltage feature matrix, respectively. Each row of these matrices represents a sample, and each column represents a dimensionality-reduced feature value, used for comparison with the reference current and voltage feature matrices to perform tasks such as inverter fault prediction. This transformation of high-dimensional monitoring current and voltage fluctuation information into low-dimensional comparison current and voltage feature matrices provides a convenient and efficient data representation for subsequent analysis and comparison.

[0047] The user terminal receives a feature multi-valued interval, wherein the feature multi-valued interval includes a current feature interval sequence.

[0048] Based on the current characteristic interval sequence, construct a dimension-reduction reconstruction function:

[0049] ;

[0050] in, Characterize the eigenvalues ​​by dimensionality reduction and reconstruction. The current characteristic characterizing the j-th attribute. , , , The current characteristic interval sequence configured for the user terminal, where N represents the interval counting sign, N is an integer, and N≥2;

[0051] According to the dimensionality reduction and reconstruction function, the reference current fluctuation information is dimensionality reduced and reconstructed to generate the reference current feature matrix. Each column of the reference current feature matrix is ​​a dimensionality reduction and reconstruction feature value of the same attribute. When an element is empty, it is padded with N+1.

[0052] The feature multi-valued interval is received through a user terminal, where the user terminal refers to the port through which the user performs configuration operations. The feature multi-valued interval refers to the feature value interval configured by the user. Multi-valued means that the feature is transformed at the user terminal, i.e., the feature is converted into 0, 1, 2, etc., and the interval to which the feature belongs is determined based on the number. The feature multi-valued interval includes a current feature interval sequence, which is defined by the user based on the upper and lower limits of the interval values. The received current feature interval sequence will be used in the subsequent dimensionality reduction and reconstruction process to convert the original current data into a series of labels or categories, each label or category corresponding to a specific interval. Based on the current feature interval sequence, a dimensionality reduction and reconstruction function is constructed, wherein... Characterize the eigenvalues ​​by dimensionality reduction and reconstruction. The current characteristic characterizing the j-th attribute. , , , A current feature interval sequence is configured for the user end, where N represents the interval count sign, and N is an integer, N≥2. A defined dimensionality reduction reconstruction function is used to transform the baseline current fluctuation information. For each sample in the dataset, its current feature value is compared with the user-defined current feature interval sequence, and the dimensionality reduction reconstruction feature value is calculated based on the comparison result. This process iterates through each sample in the dataset, checking if each current feature value falls within a user-defined current feature interval. If it does, the corresponding count is added to the accumulator of the dimensionality reduction reconstruction feature value. After iterating through all current feature values, the value in the accumulator is the dimensionality reduction reconstruction feature value, transforming the original high-dimensional current feature data into low-dimensional dimensionality reduction reconstruction feature values. Each feature value represents the distribution of the sample within a specific current feature interval. The obtained dimensionality reduction reconstruction feature values ​​are then organized into a matrix form, i.e., the baseline current feature matrix. Each row of this matrix corresponds to a sample, and each column corresponds to a dimensionality reduction and reconstruction feature. If a sample is missing data for a certain feature (i.e., the element is empty), it can be filled with N+1 elements according to regulations to ensure the integrity of the matrix and facilitate subsequent data analysis and processing. Dimensionality reduction and reconstruction were performed on the reference current fluctuation information, generating a reference current feature matrix. This reference current feature matrix not only reduces the dimensionality of the data but also retains key information from the original data, providing convenience for subsequent analysis and modeling.

[0053] The fault prediction probability is obtained by analyzing the reference current feature matrix, the comparison current feature matrix, the reference voltage feature matrix, and the comparison voltage feature matrix using the fault probability prediction component.

[0054] A fault probability prediction component is constructed. This component analyzes a reference current feature matrix, a comparison current feature matrix, a reference voltage feature matrix, and a comparison voltage feature matrix. For example, based on a neural network model, the component is constructed by providing these matrixes as input data. The component analyzes and processes the reference current feature matrix, the comparison current feature matrix, the reference voltage feature matrix, and the comparison voltage feature matrix, and outputs a fault prediction probability. This fault prediction probability refers to the probability of a fault based on historical IoT data. The fault probability prediction component effectively utilizes the information in the reference and comparison feature matrices to predict and assess the fault risk of equipment, providing strong support for fault prevention and maintenance.

[0055] By comparing the reference current feature matrix and the comparison current feature matrix, the current deviation reconstruction feature value is obtained;

[0056] Based on the deviation reconstruction feature value, frequency statistics are performed on the monitored current fluctuation information to obtain the current deviation trigger frequency;

[0057] By comparing the reference voltage feature matrix and the comparison voltage feature matrix, the voltage deviation reconstruction feature value is obtained;

[0058] Based on the voltage deviation reconstruction feature value, frequency statistics are performed on the monitored voltage fluctuation information to obtain the voltage deviation trigger frequency;

[0059] Based on the current deviation reconstruction feature value, the current deviation trigger frequency, the voltage deviation reconstruction feature value, and the voltage deviation trigger frequency, historical operation records are retrieved using the Internet of Things, and the fault detection rate is statistically analyzed and set as the fault prediction probability.

[0060] By comparing the reference current feature matrix and the comparison current feature matrix, current deviation reconstruction feature values ​​are obtained. These two matrices represent the current characteristics during normal operation and the current characteristics of the equipment in its current operating state, respectively. The comparison process can be achieved by calculating the difference, ratio, or other similarity measure between the two matrices to obtain the current deviation reconstruction feature values. The current deviation reconstruction feature values ​​refer to the differences in feature values ​​between the reference current feature matrix and the comparison current feature matrix, quantifying the degree of change in current characteristics and helping to identify potential faults. Based on the current deviation reconstruction feature values, frequency statistics are performed on the monitored current fluctuation information to analyze the number and frequency of occurrence of current deviation feature values ​​in the monitoring data. By statistically analyzing the occurrence frequency of different deviation feature values, To understand the prevalence and severity of current deviations and assess their impact on equipment operation, the higher the frequency of different deviation characteristic values, the greater the current deviation and the more likely a fault will occur. Using the above method, the reference voltage characteristic matrix and the comparison voltage characteristic matrix are compared to obtain voltage deviation reconstruction characteristic values. These voltage deviation reconstruction characteristic values ​​reflect changes in voltage characteristics and are crucial for identifying voltage-related faults. Frequency statistics are performed on the monitored voltage fluctuation information based on these voltage deviation reconstruction characteristic values ​​to obtain the voltage deviation trigger frequency. By analyzing the frequency of voltage deviation occurrences, voltage stability and its impact on equipment performance can be assessed. After obtaining the current deviation reconstruction characteristic values, current deviation trigger frequency, voltage deviation reconstruction characteristic values, and voltage deviation trigger frequency, IoT technology is used to retrieve the equipment's historical operating records. These historical operating records include various data and events from the equipment's past operation. Based on the retrieved historical operating records, the percentage of faults detected that match the current current and voltage deviation characteristics is statistically analyzed. This percentage represents the probability of equipment failure under the current operating conditions. Therefore, it can be used as the failure prediction probability. By combining real-time monitoring data with historical failure data, a quantitative assessment of equipment failure risk can be achieved.

[0061] When the predicted fault probability is greater than or equal to the fault probability threshold, a fault warning signal is generated and sent to the user terminal.

[0062] A fault probability threshold, set by those skilled in the art, is used to determine whether the frequency converter has malfunctioned. When the predicted fault probability meets the warning condition (i.e., is greater than or equal to the fault probability threshold), it indicates that the frequency converter has malfunctioned, and a fault warning signal is generated, such as generating a specific warning code, triggering an alarm sound, or displaying a light indicator. The warning signal should contain sufficient information so that the user can quickly understand the nature and severity of the fault. By generating and sending fault warning signals in a timely manner, the system can help users promptly detect and respond to potential equipment faults, thereby improving the operational safety and reliability of the equipment.

[0063] like Figure 3 As shown, embodiments of this application include a remote fault diagnosis system for frequency converters incorporating the Internet of Things, comprising:

[0064] Inverter control mode receiving module 11, the inverter control mode receiving module 11 is used to receive inverter control modes;

[0065] Calibration module 12 is used to calibrate the control mode of the frequency converter and obtain reference current fluctuation information and reference voltage fluctuation information;

[0066] A reference current feature matrix generation module 13 is used to perform dimensionality reduction and reconstruction on the reference current fluctuation information to generate a reference current feature matrix.

[0067] A reference voltage feature matrix generation module 14 is used to perform dimensionality reduction and reconstruction on the reference voltage fluctuation information to generate a reference voltage feature matrix.

[0068] Dimension reduction and reconstruction module 15 is used to receive monitoring current fluctuation information and monitoring voltage fluctuation information, perform dimension reduction and reconstruction, and obtain comparison current feature matrix and comparison voltage feature matrix.

[0069] The fault prediction probability module 16 is used to analyze the reference current feature matrix, the comparison current feature matrix, the reference voltage feature matrix and the comparison voltage feature matrix through the fault probability prediction component to obtain the fault prediction probability.

[0070] The fault warning signal generation module 17 is used to generate a fault warning signal and send it to the user terminal when the fault prediction probability is greater than or equal to the fault probability threshold.

[0071] Furthermore, embodiments of this application also include:

[0072] The positive sample backtracking module is used to perform positive sample backtracking based on the inverter control mode to obtain a current fluctuation record information set and a voltage fluctuation record information set. Here, a positive sample refers to a fault-free sample with the same inverter control mode.

[0073] A feature-level aggregation module is used to perform feature-level aggregation on the current fluctuation record information set to obtain the reference current fluctuation information.

[0074] A reference voltage fluctuation information acquisition module is used to perform feature-level aggregation on the voltage fluctuation record information set to obtain the reference voltage fluctuation information.

[0075] Furthermore, embodiments of this application also include:

[0076] A current fluctuation information extraction module is used to extract amplitude record information set, frequency record information set, phase record information set, kurtosis record information set, skewness record information set, year-on-year change rate record information set and month-on-month change rate record information set from the first current fluctuation record information set of the current fluctuation record information set. The year-on-year change rate represents the current amplitude change rate at an interval of one period, and the month-on-month change rate represents the current amplitude change rate of adjacent currents in the same period.

[0077] The first type of feature hierarchy aggregation module is used to traverse the amplitude record information set, the frequency record information set, the phase record information set, and the kurtosis record information set to perform first type of feature hierarchy aggregation, and obtain the reference current amplitude, reference current frequency, reference current phase, and reference current kurtosis.

[0078] The second type of feature hierarchy aggregation module is used to traverse the skewness record information set, the year-on-year change rate record information set and the month-on-month change rate record information set to perform second type of feature hierarchy aggregation, and obtain the reference current skewness, the reference current year-on-year change rate and the reference current month-on-month change rate.

[0079] A current information adding module is used to add the reference current amplitude, the reference current frequency, the reference current phase, the reference current kurtosis, the reference current skewness, the year-on-year change rate of the reference current, and the month-on-month change rate of the reference current to the reference current fluctuation information.

[0080] Furthermore, embodiments of this application also include:

[0081] A multi-cluster amplitude recording information acquisition module is used to perform cluster analysis on the amplitude recording information set based on an amplitude deviation threshold to obtain multi-cluster amplitude recording information, wherein the multi-cluster amplitude recording information has multiple trigger frequencies within a first cluster;

[0082] The intra-cluster amplitude feature value addition module is used to calculate the average value of intra-cluster amplitude record information with intra-cluster trigger frequencies greater than or equal to the threshold of the first intra-cluster trigger frequency based on the plurality of first intra-cluster trigger frequencies, obtain a number of intra-cluster amplitude feature values, and add them to the reference current amplitude.

[0083] The calculation module is used in which the calculation process of the frequency recording information set, the phase recording information set, and the kurtosis recording information set is the same as that of the amplitude recording information set.

[0084] Furthermore, embodiments of this application also include:

[0085] A central tendency analysis module is used to perform central tendency analysis on the skewness record information set to obtain the central tendency value of the skewness record information;

[0086] A multi-cluster skewness record information acquisition module is used to perform cluster analysis on the concentrated values ​​of the skewness record information according to the skewness deviation threshold to obtain multi-cluster skewness record information.

[0087] The mean calculation module is used to traverse the multi-cluster skewness record information to calculate the intra-cluster mean and obtain multiple intra-cluster skewness feature values.

[0088] A reference current skewness setting module is used to construct a skewness fluctuation constraint interval based on the maximum and minimum values ​​of the multiple cluster intra-cluster skewness characteristic values, and set it as the reference current skewness.

[0089] The rate of change calculation module is used in which the calculation process of the year-on-year change rate of the reference current and the month-on-month change rate of the reference current is the same as that of the skewness record information set.

[0090] Furthermore, embodiments of this application also include:

[0091] A feature multi-valued interval receiving module is used to receive a feature multi-valued interval through a user terminal, wherein the feature multi-valued interval includes a current feature interval sequence.

[0092] A dimension reduction reconstruction function construction module is used to construct a dimension reduction reconstruction function based on the current characteristic interval sequence.

[0093] ;

[0094] in, Characterize the eigenvalues ​​by dimensionality reduction and reconstruction. The current characteristic characterizing the j-th attribute. , , , The current characteristic interval sequence configured for the user terminal, where N represents the interval counting sign, N is an integer, and N≥2;

[0095] The dimension reduction and reconstruction module is used to perform dimension reduction and reconstruction on the reference current fluctuation information according to the dimension reduction and reconstruction function to generate the reference current feature matrix. Each column of the reference current feature matrix is ​​a dimension reduction and reconstruction feature value of the same attribute. When an element is empty, it is padded with N+1.

[0096] Furthermore, embodiments of this application also include:

[0097] A current deviation reconstruction feature value acquisition module is used to compare the reference current feature matrix and the comparison current feature matrix to obtain the current deviation reconstruction feature value.

[0098] A current deviation trigger frequency acquisition module is used to perform frequency statistics on the monitored current fluctuation information based on the deviation reconstruction feature value to obtain the current deviation trigger frequency.

[0099] A voltage deviation reconstruction feature value acquisition module is used to compare the reference voltage feature matrix and the comparison voltage feature matrix to obtain voltage deviation reconstruction feature values.

[0100] A voltage deviation trigger frequency acquisition module is used to perform frequency statistics on the monitored voltage fluctuation information based on the voltage deviation reconstruction feature value to obtain the voltage deviation trigger frequency.

[0101] The fault detection percentage statistics module is used to retrieve historical operation records based on the current deviation reconstruction feature value, the current deviation trigger frequency, the voltage deviation reconstruction feature value, and the voltage deviation trigger frequency, and to calculate the fault detection percentage, which is set as the fault prediction probability.

[0102] For specific embodiments of the inverter remote fault diagnosis system combined with the Internet of Things (IoT), please refer to the embodiments of the inverter remote fault diagnosis method combined with IoT described above, which will not be repeated here. The above modules can be embedded in hardware or independent of the processor in the computer device, or stored in software in the memory of the computer device, so that the processor can call and execute the corresponding operations of each module.

[0103] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores news data and data such as time decay factors. The network interface communicates with external terminals via a network connection. The computer program is executed by the processor to implement a remote fault diagnosis method for frequency converters integrated with the Internet of Things (IoT).

[0104] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0105] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of a method for remote fault diagnosis of frequency converters in conjunction with the Internet of Things.

[0106] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of a method for remote fault diagnosis of frequency converters in conjunction with the Internet of Things.

[0107] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0108] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A remote fault diagnosis method for a frequency converter combined with the Internet of Things, characterized in that, include: Receive inverter control modes; The inverter control mode is calibrated to obtain reference current fluctuation information and reference voltage fluctuation information; The reference current fluctuation information is reconstructed by dimensionality reduction to generate a reference current feature matrix; The reference voltage fluctuation information is reconstructed by dimensionality reduction to generate a reference voltage feature matrix; Receive monitoring current fluctuation information and monitoring voltage fluctuation information, perform dimensionality reduction and reconstruction, and obtain the comparison current feature matrix and the comparison voltage feature matrix; The fault prediction probability is obtained by analyzing the reference current feature matrix, the comparison current feature matrix, the reference voltage feature matrix, and the comparison voltage feature matrix using the fault probability prediction component. When the fault prediction probability is greater than or equal to the fault probability threshold, a fault warning signal is generated and sent to the user terminal. The fault prediction component analyzes the reference current feature matrix, the comparison current feature matrix, the reference voltage feature matrix, and the comparison voltage feature matrix to obtain the fault prediction probability, including: By comparing the reference current feature matrix and the comparison current feature matrix, the current deviation reconstruction feature value is obtained; Based on the current deviation reconstruction feature value, frequency statistics are performed on the monitored current fluctuation information to obtain the current deviation trigger frequency; By comparing the reference voltage feature matrix and the comparison voltage feature matrix, the voltage deviation reconstruction feature value is obtained; Based on the voltage deviation reconstruction feature value, frequency statistics are performed on the monitored voltage fluctuation information to obtain the voltage deviation trigger frequency; Based on the current deviation reconstruction feature value, the current deviation trigger frequency, the voltage deviation reconstruction feature value, and the voltage deviation trigger frequency, historical operation records are retrieved using the Internet of Things, and the fault detection rate is statistically analyzed and set as the fault prediction probability.

2. The method of claim 1, wherein, The inverter control mode is calibrated to obtain reference current fluctuation information and reference voltage fluctuation information, including: Based on the inverter control mode, positive sample backtracking is performed to obtain current fluctuation record information set and voltage fluctuation record information set, where positive sample refers to fault-free sample of the same inverter control mode; The reference current fluctuation information is obtained by performing feature-level aggregation on the current fluctuation record information set. The voltage fluctuation record information set is subjected to feature hierarchical aggregation to obtain the reference voltage fluctuation information.

3. The method of claim 2, wherein, The reference current fluctuation information is obtained by performing feature-level aggregation on the current fluctuation record information set, including: Based on the first current fluctuation record information in the current fluctuation record information set, extract the amplitude record information set, frequency record information set, phase record information set, kurtosis record information set, skewness record information set, year-on-year change rate record information set, and month-on-month change rate record information set from the first current fluctuation record information set. The year-on-year change rate represents the current amplitude change rate at an interval of one cycle, and the month-on-month change rate represents the current amplitude change rate of adjacent currents in the same cycle. Traverse the amplitude record information set, the frequency record information set, the phase record information set, and the kurtosis record information set to perform first-type feature hierarchical aggregation to obtain the reference current amplitude, reference current frequency, reference current phase, and reference current kurtosis; The second type of feature hierarchy aggregation is performed by traversing the skewness record information set, the year-on-year change rate record information set, and the month-on-month change rate record information set to obtain the reference current skewness, the reference current year-on-year change rate, and the reference current month-on-month change rate. The reference current amplitude, reference current frequency, reference current phase, reference current kurtosis, reference current skewness, year-on-year change rate of reference current, and year-on-year change rate of reference current are added to the reference current fluctuation information.

4. The method of claim 3, wherein, Traversing the amplitude record information set, the frequency record information set, the phase record information set, and the kurtosis record information set to perform first-type feature hierarchical aggregation, the reference current amplitude, reference current frequency, reference current phase, and reference current kurtosis are obtained, including: Cluster analysis is performed on the amplitude recording information set based on the amplitude deviation threshold to obtain multi-cluster amplitude recording information, wherein the multi-cluster amplitude recording information has multiple trigger frequencies within the first cluster; Based on the multiple trigger frequencies within the first cluster, the average value of the intra-cluster amplitude records with trigger frequencies greater than or equal to the threshold of the trigger frequency within the first cluster is calculated to obtain several intra-cluster amplitude characteristic values, which are then added to the reference current amplitude. The calculation process for the frequency recording information set, the phase recording information set, and the kurtosis recording information set is the same as that for the amplitude recording information set.

5. The method of claim 3, wherein, Traversing the skewness record information set, the year-on-year change rate record information set, and the month-on-month change rate record information set, a second type of feature hierarchy aggregation is performed to obtain the reference current skewness, the reference current year-on-year change rate, and the reference current month-on-month change rate, including: Perform central tendency analysis on the skewness record information set to obtain the central tendency value of the skewness record information; Cluster analysis is performed on the concentrated values ​​of the skewness record information based on the skewness deviation threshold to obtain multi-cluster skewness record information; The intra-cluster mean is calculated by traversing the multi-cluster skewness record information to obtain multiple intra-cluster skewness feature values; Based on the maximum and minimum values ​​of the multiple intra-cluster skewness characteristic values, a skewness fluctuation constraint interval is constructed and set as the reference current skewness. The calculation process for the year-on-year change rate of the reference current and the month-on-month change rate of the reference current is the same as that for the skewness record information set.

6. The method of claim 1, wherein, The reference current fluctuation information is reconstructed by dimensionality reduction to generate a reference current feature matrix, including: The user terminal receives a feature multi-valued interval, wherein the feature multi-valued interval includes a current feature interval sequence. Based on the current characteristic interval sequence, construct a dimension-reduction reconstruction function: ； wherein, characterizing the dimensionality reduction reconstructed eigenvalues, characterizing the current feature of the jth attribute, , … , … , a current feature interval sequence configured for the user end, characterizing the interval notation symbol, N is an integer, N≥2; According to the dimensionality reduction and reconstruction function, the reference current fluctuation information is dimensionality reduced and reconstructed to generate the reference current feature matrix. Each column of the reference current feature matrix is ​​a dimensionality reduction and reconstruction feature value of the same attribute. When an element is empty, it is padded with N+1.

7. A remote fault diagnosis system for a frequency converter connected to the Internet, characterized in that include: A frequency converter control mode receiving module, wherein the frequency converter control mode receiving module is used to receive frequency converter control modes; A calibration module is used to calibrate the control mode of the frequency converter and obtain reference current fluctuation information and reference voltage fluctuation information. A reference current feature matrix generation module is used to perform dimensionality reduction and reconstruction on the reference current fluctuation information to generate a reference current feature matrix. A reference voltage feature matrix generation module is used to perform dimensionality reduction and reconstruction of the reference voltage fluctuation information to generate a reference voltage feature matrix. A dimension reduction and reconstruction module is used to receive monitoring current fluctuation information and monitoring voltage fluctuation information, perform dimension reduction and reconstruction, and obtain a comparison current feature matrix and a comparison voltage feature matrix. The fault prediction probability module is used to analyze the reference current feature matrix, the comparison current feature matrix, the reference voltage feature matrix and the comparison voltage feature matrix through the fault probability prediction component to obtain the fault prediction probability. A fault warning signal generation module is used to generate a fault warning signal and send it to the user terminal when the fault prediction probability is greater than or equal to the fault probability threshold. A current deviation reconstruction feature value acquisition module is used to compare the reference current feature matrix and the comparison current feature matrix to obtain the current deviation reconstruction feature value. A current deviation trigger frequency acquisition module is used to perform frequency statistics on the monitored current fluctuation information based on the current deviation reconstruction feature value to obtain the current deviation trigger frequency. A voltage deviation reconstruction feature value acquisition module is used to compare the reference voltage feature matrix and the comparison voltage feature matrix to obtain voltage deviation reconstruction feature values. A voltage deviation trigger frequency acquisition module is used to perform frequency statistics on the monitored voltage fluctuation information based on the voltage deviation reconstruction feature value to obtain the voltage deviation trigger frequency. The fault detection percentage statistics module is used to retrieve historical operation records based on the current deviation reconstruction feature value, the current deviation trigger frequency, the voltage deviation reconstruction feature value, and the voltage deviation trigger frequency, and to calculate the fault detection percentage, which is set as the fault prediction probability.

8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images