Relay adhesion fault prediction method and device

Abstract

The invention provides a relay adhesion fault prediction method and device, and relates to the technical field of relay detection. Relay operation data and relay parameter time sequence samples are respectively collected; carrying out anomaly detection on the relay parameter time sequence sample based on an anomaly detection algorithm, obtaining an abnormal sample, and carrying out adhesion fault type labeling on the abnormal sample; preprocessing the abnormal sample and the adhesion fault type corresponding to the abnormal sample to obtain an adhesion target parameter sample; constructing a fault prediction model based on the adhesion target parameter sample and an adhesion fault judgment function; and inputting the relay operation data into the fault prediction model, and obtaining the probability that the relay corresponding to the abnormal sample has an adhesion fault, and the relay adhesion fault prediction method and device are beneficial to improving the accuracy of relay adhesion fault prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of relay detection, and in particular to a method and device for predicting relay adhesion faults. Background Art

[0002] As global environmental pollution and energy issues become increasingly severe, pure electric-powered construction machinery is increasingly common in the market. Coupled with increasing demands for battery life, battery capacities and voltages are increasing. This places stricter demands on electrical system safety testing, placing a crucial role in electrical control and safety protection.

[0003] The Chinese patent application number CN117214688A discloses a relay adhesion detection method, device, electronic device and storage medium, the method comprising: determining the adhesion detection type of the relay, wherein the adhesion detection type represents the power-on adhesion detection or the power-off adhesion detection of the relay; determining the adhesion discrimination parameter for the adhesion detection type, wherein the adhesion discrimination parameter includes at least two of the bus voltage of the circuit where the relay is located, the relay voltage of the relay and a first voltage threshold; and determining whether the relay is adhered based on the adhesion discrimination parameter. However, the data collected in the above application is relatively single and heavily relies on the threshold range set by the prior experience of the operation and maintenance personnel. It is necessary to set it separately for relays of different manufacturers and relays in different states, which can easily lead to insufficient accuracy in relay adhesion fault prediction. Therefore, it is very necessary to provide a relay adhesion fault prediction method and device to improve the accuracy of relay adhesion fault prediction. Summary of the Invention

[0004] In light of this, the present invention proposes a relay adhesion fault prediction method and device. By performing anomaly detection on relay parameter time series samples and labeling the abnormal samples with fault types, combined with time series feature methods, an accurate relay adhesion fault prediction model is constructed, thereby achieving automated prediction of relay adhesion faults and improving the accuracy of relay adhesion fault prediction.

[0005] The present invention provides a method and device for predicting relay adhesion failure, the method comprising:

[0006] Collect relay operation data and relay parameter timing samples respectively;

[0007] Perform anomaly detection on relay parameter time series samples based on an anomaly detection algorithm, obtain anomaly samples and mark the abnormal samples as adhesion fault types;

[0008] Preprocessing the abnormal sample and the adhesion fault type corresponding to the abnormal sample to obtain an adhesion target parameter sample;

[0009] Building a fault prediction model based on the adhesion target parameter sample and the adhesion fault judgment function;

[0010] The relay operation data is input into the fault prediction model to obtain the probability of the relay corresponding to the abnormal sample having a sticking fault.

[0011] On the basis of the above technical solution, preferably, the relay operation data includes the main circuit current, magnetic flux data, temperature data, number of actions and response delay of the relay.

[0012] On the basis of the above technical solution, preferably, the relay parameter timing samples are obtained by dividing the relay operation data into time series at preset time intervals according to the type of the relay operation data.

[0013] More preferably, the preprocessing of the abnormal sample and the adhesion fault type corresponding to the abnormal sample to obtain an adhesion target parameter sample specifically includes:

[0014] The abnormal samples are sequentially subjected to removing outliers, processing missing data, and eliminating noise interference to obtain standard abnormal samples, and time domain features, frequency domain features, and state features are extracted from the standard abnormal samples;

[0015] performing feature selection on the time domain features, the frequency domain features, and the state features respectively, to screen out key features most relevant to the adhesion fault;

[0016] Parameter normalization and time series alignment are performed on the key features, and the processed key features are combined with corresponding fault type labels to obtain adhesion target parameter samples.

[0017] More preferably, the constructing of a fault prediction model based on the adhesion target parameter sample and the adhesion fault judgment function specifically includes:

[0018] Performing a time series analysis on the adhesion target parameter samples to determine the abnormal time of each adhesion fault type, and obtaining a maximum value of the abnormal time corresponding to each adhesion fault type;

[0019] Taking the maximum value as a prediction period, geometrically scaling the adhesion target parameter sample to obtain a prediction vector, wherein the prediction vector represents a probability of the relay having a adhesion failure;

[0020] Determining input parameters of the fault prediction model according to the prediction vector;

[0021] The fault prediction model is constructed based on the input parameters of the fault prediction model and the adhesion fault judgment function.

[0022] More preferably, obtaining the probability of a sticking failure of the relay corresponding to the abnormal sample specifically includes:

[0023] According to the relay operation data and the prediction period, a slope vector within a prediction period from a start time to a current time is obtained;

[0024] The slope vector is matched with the prediction vector output by the fault prediction model, and the probability of a sticking fault occurring in the relay corresponding to the abnormal sample is determined based on the matching result.

[0025] More preferably, the adhesion fault judgment function is expressed as:

[0026]

[0027] Among them, I C (t) represents the current characteristic value at time t, n1 represents the static current weight value, I(t) represents the measured current value at time t, I ref Indicates the rated current value, n2 indicates the current change rate weight value, dI(t) / dt indicates the current change rate, n3 indicates the current fluctuation weight value, I ave Indicates the average current value, represents the characteristic value of magnetic flux at time t, k1 represents the static characteristic weight, represents the measured magnetic flux value at time t, represents the reference magnetic flux value, k2 represents the dynamic characteristic weight, dI(t) / dt represents the rate of change of magnetic flux, Indicates the nominal magnetic flux change rate, T C (t) represents the time characteristic value at time t, m1 represents the duration weight, T(t) represents the duration of the current relay adhesion, T ref represents the reference time of relay adhesion, m2 represents the cumulative effect weight, F(t) represents the adhesion fault judgment function, w1 represents the current characteristic weight, w2 represents the magnetic flux characteristic weight, and w3 represents the time characteristic weight.

[0028] In a second aspect of the present application, a relay adhesion fault prediction device is provided, the relay adhesion fault prediction device comprising a data acquisition module, a model building module and a fault prediction module, wherein:

[0029] The data acquisition module is used to collect relay operation data and relay parameter timing samples respectively;

[0030] The model building module is used to perform anomaly detection on relay parameter time series samples based on an anomaly detection algorithm, obtain anomaly samples and label the abnormal samples with adhesion fault types; pre-process the abnormal samples and the adhesion fault types corresponding to the abnormal samples to obtain adhesion target parameter samples; and build a fault prediction model based on the adhesion target parameter samples and the adhesion fault judgment function;

[0031] The fault prediction module is used to input the relay operation data into the fault prediction model to obtain the probability of the relay corresponding to the abnormal sample having a sticking fault.

[0032] In a third aspect of the present application, an electronic device is provided, comprising a processor, a memory, a user interface and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory.

[0033] In a fourth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. The computer program is executed by a processor to implement the steps of an air-conditioning system optimization control method based on a load point table and a genetic algorithm.

[0034] The present invention provides a method and device for predicting relay adhesion failures, which have the following advantages over the prior art:

[0035] (1) By performing anomaly detection on relay parameter time series samples and marking the fault types of abnormal samples, combined with the time series feature method, an accurate relay adhesion fault prediction model is constructed, thereby realizing the automatic prediction of relay adhesion faults and improving the accuracy of relay adhesion fault prediction. Compared with the existing technology in which operation and maintenance personnel predict relay adhesion faults based on prior experience, the efficiency of relay adhesion fault prediction is improved. At the same time, by constructing a fault prediction model, fault prediction of different power panels can be realized, which improves the applicability of relay adhesion fault prediction.

[0036] (2) Through systematic time series analysis of adhesion target parameter samples, the abnormal time characteristics of different types of adhesion faults are accurately identified, and the optimal prediction period is determined by the maximum value, which realizes the dynamic tracking of the fault development process. The prediction vector is obtained by geometric scaling processing, which not only ensures the comparability of data at different time scales, but also improves the accuracy of fault probability prediction. At the same time, the model input parameters are determined based on the prediction vector, and the prediction model is constructed in combination with the adhesion fault judgment function. A complete fault prediction system is established, which significantly improves the accuracy and reliability of relay adhesion fault prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1 A schematic diagram of the steps of a relay adhesion fault prediction method provided by the present invention;

[0039] Figure 2 A schematic structural diagram of a relay adhesion fault prediction device provided by the present invention;

[0040] Figure 3 This is a schematic structural diagram of the electronic device provided by the present invention.

[0041] Explanation of the accompanying symbols: 1. Relay adhesion fault prediction device; 11. Data acquisition module; 12. Model building module; 13. Fault prediction module; 2. Electronic device; 21. Processor; 22. Communication bus; 23. User interface; 24. Network interface; 25. Memory. DETAILED DESCRIPTION

[0042] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0043] The present invention discloses a method for predicting relay adhesion failure, referring to Figure 1 , the steps of the method include S1 to S5.

[0044] Step S1, respectively collecting relay operation data and relay parameter time series samples.

[0045] In this embodiment, the relay operation data can be collected according to needs, such as collecting the operating voltage and operating current of the relay in real time through voltage and current sensors; using displacement sensors to collect mechanical parameters such as contact displacement and contact pressure; collecting external environmental parameters such as ambient temperature and humidity; recording operating status parameters such as the number of switches and action time; and collecting electrical characteristic parameters such as contact resistance and insulation resistance.

[0046] The collection of relay parameter time series samples can be achieved by establishing a historical database, recording the parameter changes during the long-term operation of the relay, collecting and storing various parameter data at fixed time intervals, recording the trend information of parameter changes over time, saving the complete parameter sequence before and after the fault occurs, and establishing a time series model of parameter changes.

[0047] In one example, we can record the trend of parameter changes over time. For example, for relay contact resistance, the values can be recorded hourly to observe the changing trend. Under normal circumstances, the contact resistance increases slowly with use, but may suddenly increase when an abnormality occurs. For example, if the contact resistance of a relay is approximately 20mΩ during normal operation and the values are recorded every hour, after a week of continuous monitoring, the resistance value gradually rises to 25mΩ and suddenly rises to 40mΩ at a certain point. This trend change indicates the possibility of a sticking failure.

[0048] Preserve the complete parameter sequence before and after a fault occurs. For example, for relay actuation time, record complete parameter data for the 24 hours before and 6 hours after a fault occurs. For example, a relay's normal actuation time is 15ms. Within 12 hours before the fault, the actuation time gradually increases to 18ms, 20ms, and 25ms. It reaches 30ms when the fault occurs, and remains within the 28-30ms range after the fault occurs. This complete sequence clearly demonstrates the development of parameter anomalies.

[0049] To build a time series model of parameter changes, for example, a relay coil current can be used to construct a time series model encompassing multiple related parameters. The model records the changing patterns of parameters such as coil current, coil temperature, and ambient temperature. During normal operation, the coil current and temperature exhibit a stable correlation. However, this correlation changes when an anomaly occurs. For example, under normal circumstances, the coil current increases slightly with increasing temperature. However, when a sticking fault occurs, the coil current exhibits abnormal fluctuations even when the temperature remains unchanged. The model can capture these abnormal fluctuations.

[0050] Furthermore, the relay operation data includes the main circuit current, magnetic flux data, temperature data, number of actions and response delay of the relay. The relay parameter timing samples are obtained by dividing the relay operation data into time series at preset time intervals according to the type of the relay operation data.

[0051] Step S2: performing anomaly detection on the relay parameter time series samples based on an anomaly detection algorithm, obtaining abnormal samples and marking the abnormal samples as adhesion fault types.

[0052] In this step, the anomaly detection algorithm may adopt a moving average detection method or a LOF local anomaly factor algorithm.

[0053] In an example, taking the contact resistance parameter of a relay contact as an example, the process of implementing anomaly detection using the LOF local anomaly factor algorithm is as follows:

[0054] Assume that the contact resistance values collected at five consecutive time points are: time t1: 22 mΩ; time t2: 23 mΩ; time t3: 24 mΩ; time t4: 35 mΩ; time t5: 23 mΩ.

[0055] Set the minimum number of points MinPts to 3, select Euclidean distance as the distance metric, and set the LOF threshold to 1.5. Take time t4 (35mΩ) as an example and calculate the distance to other points.

[0056] The distance between time t4 and point t1 is: |35-22|=13;

[0057] The distance between time t4 and point t2 is: |35-23|=12;

[0058] The distance between time t4 and point t3 is: |35-24|=11;

[0059] The distance between time t4 and point t5 is: |35-23|=12;

[0060] The k distance value (k=3) is 12 (the third closest distance);

[0061] The reachable distance from point t4 to point t1 is: max(12, 13) = 13;

[0062] The reachable distance from point t4 to point t2 is: max(12, 12) = 12;

[0063] The reachable distance from point t4 to point t3 is: max(12, 11) = 12;

[0064] The reachable distance from point t4 to point t5 is: max(12, 12) = 12;

[0065] Since the local reachability density of point t4 is the number of points in the neighborhood divided by the sum of the reachable distances, the local reachability density = 3 / (12+12+12) = 0.083.

[0066] The local reachability density of the reference point, taking point t3 as an example, the distance to the adjacent points is within the range of 1-2, and the local reachability density is about 0.5.

[0067] The LOF value of point t4 = the average local reachability density of the neighboring points / the local reachability density of point t4 (ie, LOF value = 0.5 / 0.083 = 6.02).

[0068] The LOF value at time t4 (6.02) is significantly greater than the set threshold (1.5), indicating that the contact resistance at t4 is an outlier. The contact resistance at t4 (35mΩ) deviates significantly from the values at nearby time points, and the local density at this point is significantly lower than that of surrounding points. This is typical of a local anomaly, indicating the onset of a relay contact adhesion failure. The LOF algorithm can accurately identify outliers in relay parameter sequences, particularly those data points that significantly deviate from their local neighborhood. This method is particularly suitable for detecting sudden anomalies during relay operation.

[0069] Furthermore, specific conditions for sample extraction were set, including a time window centered on the outlier point, extending forward 24 hours and backward 6 hours. During this time, key parameters such as contact resistance, actuation time, and coil current were collected, with a sampling frequency set to once per hour. Any time point with a LOF value exceeding 1.5 or a statistical anomaly was identified as an outlier.

[0070] For each abnormal sample, complete information is recorded, including the timestamp, the specific value of the abnormal parameter, the values of other related parameters, the degree of abnormality, and the parameter change trend characteristics. For example, when an abnormality is detected at time t4, the contact resistance value of 35mΩ is recorded at that moment. Also, related parameters such as the actuation time of 25ms and the coil current of 2.8A are recorded. The calculated LOF abnormality index value is 6.02.

[0071] Adhesion faults are classified into three levels based on the severity of the fault: mild, moderate, and severe. Mild adhesion is characterized by a 20%-30% increase in contact resistance and a 10%-20% increase in operating time, with strong recoverability. Moderate adhesion is characterized by a 30%-50% increase in contact resistance and a 20%-40% increase in operating time, with some recoverability. Severe adhesion is characterized by an increase in contact resistance exceeding 50%, a 40% increase in operating time, and difficulty in self-recovery.

[0072] Each abnormal sample is systematically recorded, including the sample's unique identifier, occurrence time, fault type, fault level, key characteristic parameters (such as contact resistance deviation rate, actuation time deviation rate, coil current fluctuation rate, etc.), recoverability assessment, and treatment recommendations. This information is stored in a structured format to facilitate subsequent analysis and use.

[0073] Step S3: pre-process the abnormal samples and the adhesion fault types corresponding to the abnormal samples to obtain adhesion target parameter samples.

[0074] This step also includes steps S31 to S33.

[0075] In step S31 , outliers are removed, missing data are processed, and noise interference is eliminated in sequence for the abnormal samples to obtain standard abnormal samples, and time domain features, frequency domain features, and state features are extracted from the standard abnormal samples.

[0076] In this step, the IQR (interquartile range) method is used to remove outliers. The Q1 (25th percentile) and Q3 (75th percentile) of the data are calculated, with IQR = Q3 - Q1. Upper and lower bounds are set, where the upper bound = Q3 + 1.5*IQR and the lower bound = Q1 - 1.58*IQR. Data points outside these bounds are marked as outliers and removed. For time series data, linear interpolation is used to fill short-term missing values, and spline interpolation is used to handle longer-term missing values. For non-time series data, when the missing ratio is <5%, the mean / median is used for filling; when the missing ratio is >5%, the KNN interpolation method is used. For key features, multiple interpolation is considered. Wavelet transform is used to eliminate noise interference. An appropriate wavelet basis function (such as db4 or sym8) is selected, and multi-scale wavelet decomposition is performed. Soft thresholding is used for denoising at each scale, and wavelet reconstruction is used to obtain the denoised signal.

[0077] In step S32 , feature selection is performed on the time domain features, frequency domain features, and state features respectively to screen out the key features most relevant to the adhesion fault.

[0078] In step S33 , parameter normalization and time series alignment are performed on the key features, and the processed key features are combined with corresponding fault type labels to obtain adhesion target parameter samples.

[0079] In this step, for features with relatively concentrated distributions, the minimum-maximum normalization method is used to map the data to the interval [0, 1]. For features with large fluctuations, the Z-score normalization method is used to set the data to a mean of 0 and a standard deviation of 1. For features containing outliers, the quartile normalization method is used. First, the statistics of each feature (maximum, minimum, mean, standard deviation, etc.) are calculated and then converted according to the selected normalization method.

[0080] Time series alignment involves determining the target sampling rate, interpolating sequences with lower sampling rates (e.g., linear interpolation), downsampling sequences with higher sampling rates, ensuring that all sequences have the same time interval, determining a standard sequence length, truncating longer sequences, padding shorter sequences (e.g., zero padding or repeating trailing values), and verifying the consistency of all sequence lengths. Time alignment involves selecting an appropriate reference sequence as a benchmark, identifying key time points or feature points in the sequence, aligning the sequences based on these feature points, and, if necessary, using the dynamic time warping (DTW) algorithm for precise alignment.

[0081] By removing outliers, processing missing data, and eliminating noise interference, the quality and reliability of sample data are significantly improved. Furthermore, by extracting time, frequency, and state features, a comprehensive characterization of fault characteristics is achieved. Feature selection is then used to identify the key features most relevant to adhesion faults, reducing data redundancy and computational complexity while also increasing the correlation between features and faults. Finally, parameter normalization and time series alignment eliminate dimensional differences between features, ensuring sample consistency across the temporal dimension and forming a standardized, high-quality training sample set. This series of processes forms a complete data processing chain, providing a reliable data foundation for the fault prediction model and effectively improving the model's accuracy, generalization capability, and interpretability of prediction results.

[0082] Step S4: constructing a fault prediction model based on the adhesion target parameter sample and the adhesion fault judgment function.

[0083] This step also includes steps S41 to S44.

[0084] Step S41 , performing time series analysis on the adhesion target parameter samples, determining the abnormal time of each adhesion fault type, and obtaining the maximum value of the abnormal time corresponding to each adhesion fault type.

[0085] In this step, the sample data is sorted in chronological order, the continuity of the timestamp is ensured and the integrity of the data is checked, possible missing values are processed, the samples are classified by fault type, a time series index is established to determine the analysis time window, and reference baseline data is prepared. In the anomaly detection analysis, the sliding mean and standard deviation changes are calculated respectively to analyze the data distribution characteristics. For change point detection, the CUSUM algorithm can be used to detect cumulative changes, the window segmentation method is applied, the statistical significance is calculated, and potential change points are marked. For the anomaly judgment standard, first set the threshold range, identify the time when the threshold is first exceeded, calculate the duration and determine the key time points, record the anomaly characteristics, analyze the time characteristics of different fault types, compare the differences in anomaly patterns, and establish a feature mapping relationship.

[0086] Step S42 : Taking the maximum value as the prediction period, geometrically scaling the adhesion target parameter samples to obtain a prediction vector, wherein the prediction vector represents the probability of a adhesion fault occurring in the relay.

[0087] Step S43: Determine input parameters of the fault prediction model according to the prediction vector.

[0088] Step S44: constructing a fault prediction model based on the input parameters of the fault prediction model and the adhesion fault judgment function.

[0089] By conducting a systematic time series analysis of adhesion target parameter samples, the abnormal time characteristics of different types of adhesion faults are accurately identified, and the optimal prediction period is determined through the maximum value, thus realizing dynamic tracking of the fault development process. The prediction vector is obtained by geometric scaling processing, which not only ensures the comparability of data at different time scales, but also improves the accuracy of fault probability prediction. At the same time, the model input parameters are determined based on the prediction vector, and the prediction model is constructed in combination with the adhesion fault judgment function. A complete fault prediction system is established, which significantly improves the accuracy and reliability of relay adhesion fault prediction.

[0090] In this step, the expression of the adhesion fault judgment function is:

[0091]

[0092] Among them, I C (t) represents the current characteristic value at time t, n1 represents the static current weight value, I(t) represents the measured current value at time t, I ref Indicates the rated current value, n2 indicates the current change rate weight value, dI(t) / dt indicates the current change rate, n3 indicates the current fluctuation weight value, I ave Indicates the average current value, represents the characteristic value of magnetic flux at time t, k1 represents the static characteristic weight, represents the measured magnetic flux value at time t, represents the reference magnetic flux value, k2 represents the dynamic characteristic weight, dI(t) / dt represents the rate of change of magnetic flux, Indicates the nominal magnetic flux change rate, T C (t) represents the time characteristic value at time t, m1 represents the duration weight, T(t) represents the duration of the current relay adhesion, T ref represents the reference time of relay adhesion, m2 represents the cumulative effect weight, F(t) represents the adhesion fault judgment function, w1 represents the current characteristic weight, w2 represents the magnetic flux characteristic weight, and w3 represents the time characteristic weight.

[0093] Step S5: input the relay operation data into the fault prediction model to obtain the probability of the relay corresponding to the abnormal sample having a sticking fault.

[0094] This step also includes steps S51 to S52.

[0095] Step S51 : obtaining a slope vector within a prediction period from a start time to a current time according to the relay operation data and the prediction period.

[0096] Step S52: Match the slope vector with the prediction vector output by the fault prediction model, and determine the probability of a sticking fault occurring in the relay corresponding to the abnormal sample based on the matching result.

[0097] In this embodiment, a dual verification mechanism combining slope vectors and prediction vectors significantly improves the accuracy and reliability of relay adhesion fault predictions. By dynamically capturing fault development trends and updating prediction results in real time, the system can promptly detect abnormal conditions and provide early warnings, providing ample lead time for preventive maintenance. This probabilistic output not only provides a more detailed basis for risk assessment but also supports more scientific maintenance decision-making, effectively reducing the risk of false positives and missed warnings.

[0098] By performing anomaly detection on relay parameter time series samples and labeling the fault types of abnormal samples, combined with the time series feature method, an accurate relay adhesion fault prediction model is constructed, thereby realizing the automatic prediction of relay adhesion faults and improving the accuracy of relay adhesion fault prediction. Compared with the existing technology in which operation and maintenance personnel predict relay adhesion faults based on prior experience, the efficiency of relay adhesion fault prediction is improved. At the same time, by constructing a fault prediction model, fault prediction of different power panels can be realized, thereby improving the applicability of relay adhesion fault prediction.

[0099] Based on the above method, the embodiment of the present application discloses a relay adhesion fault prediction device, referring to Figure 2 The relay adhesion fault prediction device 1 includes a data acquisition module 11, a model building module 12 and a fault prediction module 13, wherein

[0100] The data acquisition module 11 is used to collect relay operation data and relay parameter time series samples respectively;

[0101] The model building module 12 is used to perform anomaly detection on the relay parameter time series samples based on the anomaly detection algorithm, obtain the abnormal samples and mark the abnormal samples with the adhesion fault type; pre-process the abnormal samples and the adhesion fault type corresponding to the abnormal samples to obtain adhesion target parameter samples; and build a fault prediction model based on the adhesion target parameter samples and the adhesion fault judgment function;

[0102] The fault prediction module 13 is used to input the relay operation data into the fault prediction model to obtain the probability of the relay adhesion fault corresponding to the abnormal sample.

[0103] In one example, the relay operation data includes the main circuit current, magnetic flux data, temperature data, operation times, and response delay of the relay.

[0104] In one example, the relay parameter time series samples are obtained by dividing the relay operation data into time series at preset time intervals according to the type of the relay operation data.

[0105] In one example, the model building module 12 is used to remove outliers, process missing data, and eliminate noise interference from abnormal samples in sequence to obtain standard abnormal samples, and extract time domain features, frequency domain features, and state features from the standard abnormal samples; perform feature selection on the time domain features, frequency domain features, and state features respectively to screen out the key features most relevant to the adhesion fault; perform parameter normalization and time series alignment on the key features, and combine the processed key features with the corresponding fault type labels to obtain adhesion target parameter samples.

[0106] In one example, the model construction module 12 is used to perform time series analysis on the adhesion target parameter samples, determine the abnormal time of each adhesion fault type, and obtain the maximum value of the abnormal time corresponding to each adhesion fault type; use the maximum value as the prediction period, and perform geometric scaling on the adhesion target parameter samples to obtain a prediction vector, wherein the prediction vector represents the probability of a adhesion fault occurring in the relay; determine the input parameters of the fault prediction model based on the prediction vector; and construct a fault prediction model based on the input parameters of the fault prediction model and the adhesion fault judgment function.

[0107] In one example, the fault prediction module 13 is used to obtain a slope vector within a prediction period from the start time to the current time based on the relay operation data and the prediction period; match the slope vector with the prediction vector output by the fault prediction model, and determine the probability of a adhesion fault occurring in the relay corresponding to the abnormal sample based on the matching result.

[0108] In one example, the expression of the adhesion fault judgment function is:

[0109]

[0110] Among them, I C (t) represents the current characteristic value at time t, n1 represents the static current weight value, I(t) represents the measured current value at time t, I ref Indicates the rated current value, n2 indicates the current change rate weight value, dI(t) / dt indicates the current change rate, n3 indicates the current fluctuation weight value, I ave Indicates the average current value, represents the characteristic value of magnetic flux at time t, k1 represents the static characteristic weight, represents the measured magnetic flux value at time t, represents the reference magnetic flux value, k2 represents the dynamic characteristic weight, dI(t) / dt represents the rate of change of magnetic flux, Indicates the nominal magnetic flux change rate, T C (t) represents the time characteristic value at time t, m1 represents the duration weight, T(t) represents the duration of the current relay adhesion, T ref represents the reference time of relay adhesion, m2 represents the cumulative effect weight, F(t) represents the adhesion fault judgment function, w1 represents the current characteristic weight, w2 represents the magnetic flux characteristic weight, and w3 represents the time characteristic weight.

[0111] See Figure 3 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 3 As shown, the electronic device 2 may include: at least one processor 21 , at least one network interface 24 , a user interface 23 , a memory 25 , and at least one communication bus 22 .

[0112] The communication bus 22 is used to realize the connection and communication between these components.

[0113] The user interface 23 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 23 may also include a standard wired interface and a wireless interface.

[0114] The network interface 24 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0115] The processor 21 may include one or more processing cores. The processor 21 utilizes various interfaces and lines to connect various parts of the entire server. By running or executing instructions, programs, code sets, or instruction sets stored in the memory 25, and calling data stored in the memory 25, the processor 21 performs various server functions and processes data. Optionally, the processor 21 may be implemented in at least one hardware form selected from digital signal processing (DSP), field-programmable gate array (FPGA), and programmable logic array (PLA). The processor 21 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display; and the modem is used to handle wireless communications. It is understood that the modem may not be integrated into the processor 21 and may be implemented separately on a single chip.

[0116] Among them, the memory 25 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 25 includes a non-transitory computer-readable storage medium. The memory 25 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 25 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 25 may also be optionally at least one storage device located away from the aforementioned processor 21. As Figure 3 As shown, the memory 25 as a computer storage medium may include an operating system, a network communication module, a user interface module, and an application program of an air conditioning system optimization control method based on a load point table and a genetic algorithm.

[0117] exist Figure 3 In the electronic device 2 shown, the user interface 23 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 21 can be used to call the application program stored in the memory 25 for the air-conditioning system optimization control method based on the load point table and the genetic algorithm. When executed by one or more processors, the electronic device executes one or more methods in the above-mentioned embodiments.

[0118] A computer-readable storage medium stores instructions, which, when executed by one or more processors, cause the computer to execute one or more methods in the above-mentioned embodiments.

[0119] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for this application.

[0120] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0121] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely schematic, such as the division of units, which is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interface, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0122] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0123] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0124] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes various media that can store program codes, such as USB flash drives, mobile hard drives, magnetic disks or optical disks.

[0125] The above are merely exemplary embodiments of the present disclosure and are not intended to limit the scope of the present disclosure. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present disclosure. This application is intended to cover any variations, uses, or adaptive changes of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not described in the present disclosure.

[0126] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for predicting relay adhesion failure, characterized in that: The method comprises: Collect relay operation data and relay parameter timing samples respectively; Perform anomaly detection on relay parameter time series samples based on an anomaly detection algorithm, obtain anomaly samples and mark the abnormal samples as adhesion fault types; Preprocessing the abnormal sample and the adhesion fault type corresponding to the abnormal sample to obtain an adhesion target parameter sample; Building a fault prediction model based on the adhesion target parameter sample and the adhesion fault judgment function; The relay operation data is input into the fault prediction model to obtain the probability of the relay corresponding to the abnormal sample having a sticking fault.

2. The method according to claim 1, wherein The relay operation data includes the relay's main circuit current, magnetic flux data, temperature data, number of actions, and response delay.

3. The method according to claim 1, wherein The relay parameter time series samples are obtained by dividing the relay operation data into time series at preset time intervals according to the type of the relay operation data.

4. The method according to claim 1, wherein The preprocessing of the abnormal sample and the adhesion fault type corresponding to the abnormal sample to obtain an adhesion target parameter sample specifically includes: The abnormal samples are sequentially subjected to removing outliers, processing missing data, and eliminating noise interference to obtain standard abnormal samples, and time domain features, frequency domain features, and state features are extracted from the standard abnormal samples; performing feature selection on the time domain features, the frequency domain features, and the state features respectively, to screen out key features most relevant to the adhesion fault; Parameter normalization and time series alignment are performed on the key features, and the processed key features are combined with corresponding fault type labels to obtain adhesion target parameter samples.

5. The method according to claim 4, wherein The constructing of a fault prediction model based on the adhesion target parameter sample and the adhesion fault judgment function specifically includes: Performing a time series analysis on the adhesion target parameter samples to determine the abnormal time of each adhesion fault type, and obtaining a maximum value of the abnormal time corresponding to each adhesion fault type; Taking the maximum value as a prediction period, geometrically scaling the adhesion target parameter sample to obtain a prediction vector, wherein the prediction vector represents a probability of the relay having a adhesion failure; Determining input parameters of the fault prediction model according to the prediction vector; The fault prediction model is constructed based on the input parameters of the fault prediction model and the adhesion fault judgment function.

6. The method according to claim 5, wherein The obtaining of the probability of a sticking fault occurring in the relay corresponding to the abnormal sample specifically includes: According to the relay operation data and the prediction period, a slope vector within a prediction period from a start time to a current time is obtained; The slope vector is matched with the prediction vector output by the fault prediction model, and the probability of a sticking fault occurring in the relay corresponding to the abnormal sample is determined based on the matching result.

7. The method according to claim 1, wherein The expression of the adhesion fault judgment function is: Among them, I C (t) represents the current characteristic value at time t, n1 represents the static current weight value, I(t) represents the measured current value at time t, I ref Indicates the rated current value, n2 indicates the current change rate weight value, dI(t) / dt indicates the current change rate, n3 indicates the current fluctuation weight value, I ave Indicates the average current value, represents the characteristic value of magnetic flux at time t, k1 represents the static characteristic weight, represents the measured magnetic flux value at time t, represents the reference magnetic flux value, k2 represents the dynamic characteristic weight, dI(t) / dt represents the rate of change of magnetic flux, Indicates the nominal magnetic flux change rate, T C (t) represents the time characteristic value at time t, m1 represents the duration weight, T(t) represents the duration of the current relay adhesion, T ref represents the reference time of relay adhesion, m2 represents the cumulative effect weight, F(t) represents the adhesion fault judgment function, w1 represents the current characteristic weight, w2 represents the magnetic flux characteristic weight, and w3 represents the time characteristic weight.

8. A relay adhesion fault prediction device, characterized in that: The relay adhesion fault prediction device (1) comprises a data acquisition module (11), a model construction module (12) and a fault prediction module (13), wherein: The data acquisition module (11) is used to respectively acquire relay operation data and relay parameter time sequence samples; The model construction module (12) is used to perform anomaly detection on relay parameter time series samples based on an anomaly detection algorithm, obtain anomaly samples and mark the abnormal samples with adhesion fault types; pre-process the abnormal samples and the adhesion fault types corresponding to the abnormal samples to obtain adhesion target parameter samples; and construct a fault prediction model based on the adhesion target parameter samples and the adhesion fault judgment function; The fault prediction module (13) is used to input the relay operation data into the fault prediction model to obtain the probability of the relay corresponding to the abnormal sample having a sticking fault.

9. An electronic device, characterized in that: The electronic device (2) comprises a processor (21), a memory (25), a user interface (23) and a network interface (24), wherein the memory (25) is used to store instructions, the user interface (23) and the network interface (24) are used to communicate with other devices, and the processor (21) is used to execute the instructions stored in the memory (25) so that the electronic device (2) executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images