Control cubicle fault early warning method and device and electronic equipment

Abstract

The invention provides a fault early warning method and device for a control cubicle and electronic equipment. The method comprises the following steps: acquiring operation data of a contact in a control cubicle in a set duration; performing multi-scale segmentation on the operation data to obtain a data sequence of a plurality of time spans; for the data sequence corresponding to each time span in the plurality of time spans, carrying out denoising processing on the data sequence by adopting a wavelet packet analysis algorithm to obtain a denoised data sequence corresponding to the time span; inputting each denoised data sequence into a pre-trained temperature prediction model to obtain a first prediction temperature corresponding to each time span; determining a second predicted temperature of the contact based on the first predicted temperature and a weight of the corresponding time span; and when the second predicted temperature is greater than a set temperature threshold value, determining the contact as a potential abnormal contact, and carrying out fault early warning on the control cubicle. The method is used for achieving the effects of accurately predicting the contact temperature and further performing accurate early warning on the fault of the control cubicle.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of smart grid monitoring, and in particular to a power collection and control cabinet fault early warning method and device and electronic equipment. BACKGROUND

[0002] With the development of smart grid and the improvement of voltage level of power system, the operation state of power collection and control cabinet as a core electrical equipment directly affects the safety and reliability of power grid. The power collection and control cabinet is composed of cable room, busbar room and circuit breaker room, and mainly plays a breaking role in the process of power generation, transmission and distribution of power grid system, and also has the functions of control and protection, and plays an important role in the safe, coordinated and reliable operation of power system. Therefore, it is very important to accurately predict the fault of the power collection and control cabinet.

[0003] In the related art, the change of contact temperature in the power collection and control cabinet is mainly used for fault early warning. Specifically, the running data of the contact in the power collection and control cabinet within a set time length is obtained, a trained machine learning model is applied to process the running data, and the predicted temperature data of the contact at a future time is output. If the early warning temperature data is greater than a set temperature threshold, the power collection and control cabinet is early warned of fault. However, the accuracy of the predicted contact temperature by this method is low, which leads to inaccurate early warning of the fault of the power collection and control cabinet. SUMMARY

[0004] The present application provides a power collection and control cabinet fault early warning method, device and electronic equipment to accurately predict the contact temperature and accurately early warn the fault of the power collection and control cabinet, improve the operation and maintenance efficiency, and ensure the safe and stable operation of the power collection and control cabinet.

[0005] In a first aspect, the present application provides a power collection and control cabinet fault early warning method, comprising:

[0006] obtaining running data of a contact in a power collection and control cabinet within a set time length;

[0007] performing multi-scale segmentation on the running data to obtain data sequences of multiple time spans;

[0008] for each data sequence corresponding to each time span in the multiple time spans, performing denoising processing on the data sequence by using a wavelet packet analysis algorithm to obtain a denoised data sequence corresponding to the time span;

[0009] inputting each denoised data sequence into a pre-trained temperature prediction model to obtain a first predicted temperature corresponding to each time span, the temperature prediction model being obtained by pre-training a machine learning model based on historical running data of the contact of the power collection and control cabinet, and the machine learning model comprising a support vector machine;

[0010] Based on the weights of the first predicted temperature and the corresponding time span, the second predicted temperature of the contact point is determined. The weights represent the degree of influence of the corresponding time span on the predicted temperature.

[0011] When the second predicted temperature exceeds the set temperature threshold, the contact is identified as a potential abnormal contact, and a fault warning is issued to the control cabinet.

[0012] In one possible implementation, a wavelet packet analysis algorithm is used to denoise the data sequence, obtaining a denoised data sequence corresponding to the time span, including:

[0013] The data sequence is decomposed into three-level wavelet packet decomposition to obtain the wavelet packet coefficients of the eight nodes corresponding to the frequency bands.

[0014] The wavelet packet coefficients of each node are processed by the hard thresholding method to obtain the denoised wavelet packet coefficients of the corresponding frequency band. The threshold in the hard thresholding method is dynamically determined based on the signal energy.

[0015] The denoised wavelet packet coefficients are reconstructed to obtain the denoised data sequence corresponding to the time span.

[0016] In one possible implementation, the threshold satisfies the following formula:

[0017]

[0018] In the formula, In a three-layer wavelet packet decomposition tree, the first... Threshold for each node; Basic threshold; For the first Each node corresponds to a frequency band of signal energy, and the signal energy represents the sum of squares of the wavelet packet coefficients within the frequency band; This represents the total energy of all frequency bands in the three-layer wavelet packet decomposition tree. This is an adjustment coefficient used to control the intensity of the energy's influence on the threshold. .

[0019] In one possible implementation, the runtime data is segmented into multiple scales to obtain data sequences spanning multiple time spans, including:

[0020] Based on a defined segmentation algorithm, the running data is segmented at multiple scales to obtain data sequences spanning multiple time spans. The defined segmentation algorithm satisfies the following form:

[0021]

[0022] In the formula, For the first The length of the data sequence spanning the time layer; The preset initial segment length; a maximum number of multi-scale segments, , a maximum number of multi-scale segments, , a length of the operation data.

[0023] In a possible implementation, the weight satisfies the following calculation formula:

[0024]

[0025] wherein, is the weight of the i-th time span; is the weight of the i-th time span; is a maximum number of multi-scale segments.

[0026] In a possible implementation, the weight corresponding to each time span is obtained based on the data sequence of the plurality of time spans by using an entropy weight method.

[0027] In a possible implementation, the power control cabinet fault early warning method further comprises:

[0028] determining a deviation of the second predicted temperature relative to the set temperature threshold value;

[0029] if the deviation is within a first threshold range, a first fault early warning strategy is executed, the first fault early warning strategy being configured to output first early warning information;

[0030] if the deviation is within a second threshold range, a second fault early warning strategy is executed, the second fault early warning strategy being configured to trigger a regulation and control mechanism, the regulation and control mechanism comprising enabling a heat dissipation system or reducing a load, and outputting second early warning information;

[0031] if the deviation is within a third threshold range, a third fault early warning is performed on the power control cabinet, the third fault early warning being configured to automatically cut off power supply of the power control cabinet, and output third early warning information.

[0032] In a second aspect, the present application provides a power control cabinet fault early warning device, comprising:

[0033] an acquisition module configured to acquire operation data of contacts in the power control cabinet within a set time length;

[0034] a first processing module configured to perform multi-scale segmentation on the operation data to obtain a plurality of data sequences of time spans;

[0035] The first processing module is further configured to, for each data sequence corresponding to each time span in the plurality of time spans, perform denoising processing on the data sequence by using a wavelet packet analysis algorithm to obtain a denoised data sequence corresponding to the time span.

[0036] The second processing module is used to input each denoised data sequence into the pre-trained temperature prediction model to obtain the first predicted temperature corresponding to each time span. The temperature prediction model is pre-trained on a machine learning model based on the historical operating data of the control cabinet contacts. The machine learning model includes a support vector machine.

[0037] The determination module is used to determine the second predicted temperature of the contact point based on the first predicted temperature and the weight of the corresponding time span. The weight represents the degree of influence of the corresponding time span on the predicted temperature.

[0038] The early warning module is used to identify the contact as a potential abnormal contact and provide a fault warning to the control cabinet when the second predicted temperature is greater than the set temperature threshold.

[0039] In one possible implementation, the first processing module is specifically used to: perform three-level wavelet packet decomposition on the data sequence to obtain wavelet packet coefficients corresponding to the frequency bands of the eight nodes; process each wavelet packet coefficient based on the hard thresholding method to obtain the denoised wavelet packet coefficients of the frequency bands corresponding to each node, wherein the threshold in the hard thresholding method is dynamically determined based on the signal energy; and reconstruct the denoised wavelet packet coefficients to obtain the denoised data sequence corresponding to the time span.

[0040] In one possible implementation, the threshold satisfies the following formula:

[0041]

[0042] In the formula, In a three-layer wavelet packet decomposition tree, the first... Threshold for each node; Basic threshold; For the first Each node corresponds to a frequency band of signal energy, and the signal energy represents the sum of squares of the wavelet packet coefficients within the frequency band; This represents the total energy of all frequency bands in the three-layer wavelet packet decomposition tree. This is an adjustment coefficient used to control the intensity of the energy's influence on the threshold. .

[0043] In one possible implementation, the first processing module is further configured to: perform multi-scale segmentation on the running data based on a set segmentation algorithm to obtain data sequences with multiple time spans, wherein the set segmentation algorithm satisfies the following form:

[0044]

[0045] In the formula, For the first The length of the data sequence spanning the time layer; The preset initial segment length; It is a multi-scale segmented hierarchy. , The maximum number of levels in multi-scale segmentation. , The length of the data being processed.

[0046] In one possible implementation, the weights satisfy the following calculation formula:

[0047]

[0048] In the formula, For the first Weights for the time span of each layer; This represents the maximum number of levels in the multi-scale segmentation.

[0049] In one possible implementation, the weights corresponding to each time span are obtained by assigning weights to each time span based on data sequences of multiple time spans using the entropy weight method.

[0050] In one possible implementation, the early warning module is specifically used to: determine the deviation of the second predicted temperature from a set temperature threshold; if the deviation is within a first threshold range, execute a first fault early warning strategy, which is used to output a first early warning message; if the deviation is within a second threshold range, execute a second fault early warning strategy, which is used to trigger a control mechanism, including activating the heat dissipation system or reducing the load, and output a second early warning message; if the deviation is within a third threshold range, issue a third fault early warning for the control cabinet, which is used to automatically cut off the power supply to the control cabinet and output a third early warning message.

[0051] Thirdly, this application provides an electronic device, including: a memory and a processor;

[0052] The memory stores the instructions that the computer executes;

[0053] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0054] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible embodiments of the first aspect.

[0055] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0056] The method, device, and electronic equipment for fault early warning of the control cabinet provided in this application acquire the operating data of the contacts in the control cabinet within a set time period; perform multi-scale segmentation on the operating data to obtain data sequences of multiple time spans; for the data sequences corresponding to each time span in the multiple time spans, a wavelet packet analysis algorithm is used to denoise the data sequences to obtain denoised data sequences corresponding to the time spans; input each denoised data sequence into a pre-trained temperature prediction model to obtain the first predicted temperature corresponding to each time span. The temperature prediction model is a machine learning model pre-trained based on the historical operating data of the control cabinet contacts, and the machine learning model includes a support vector machine; based on the first predicted temperature and the weight of the corresponding time span, a second predicted temperature of the contact is determined, and the weight represents the degree of influence of the corresponding time span on the predicted temperature; when the second predicted temperature is greater than a set temperature threshold, the contact is identified as a potential abnormal contact, and a fault early warning is issued for the control cabinet. This process involves segmenting the operational data and applying the segmented data sequences to predict temperature. This allows for analysis of contact point operational data across different time spans, uncovering multi-layered patterns, and further weighted fusion to obtain the predicted temperature of the contact point at future times. By integrating information from different time spans, balancing local and global features, and reducing the impact of abnormal trend fluctuations in data within a set time period, the accuracy of temperature prediction is improved. Wavelet packet denoising effectively suppresses noise interference, reducing its impact on model prediction accuracy. Furthermore, the application of a support vector machine regression model effectively captures nonlinear relationships in the operational data, enhancing the model's adaptability to complex nonlinear operational data changes and further improving the accuracy of temperature prediction. This significantly improves the accuracy and reliability of the control cabinet fault early warning system, enabling precise early warning of control cabinet faults, timely and effectively preventing fault escalation and deterioration, improving operational efficiency, and ensuring the safe and stable operation of the control cabinet. Attached Figure Description

[0057] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0058] Figure 1 A schematic diagram illustrating a scenario for the control cabinet fault early warning method provided in this application embodiment;

[0059] Figure 2 A flowchart illustrating the control cabinet fault early warning method provided in this application embodiment. Figure 1 ;

[0060] Figure 3 A flowchart illustrating the control cabinet fault early warning method provided in this application embodiment. Figure 2 ;

[0061] Figure 4A structural schematic diagram of the fault early warning device of the switch cabinet provided in the embodiments of the present application is shown.

[0062] Figure 5 A structural schematic diagram of the electronic device provided in the embodiments of the present application is shown.

[0063] The specific embodiments of the present application have been shown in the above-described drawings, and will be described in more detail hereinafter. These drawings and the written description are not intended to restrict the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0064] The exemplary embodiments will be described in detail herein with reference to the attached drawings. In the following description, the same numbers are used to indicate the same or similar components. The embodiments described in the following exemplary embodiments are not meant to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present application as detailed in the appended claims.

[0065] The terms "first", "second", and the like in the description and in the claims of the present application are used for distinguishing between similar objects and not necessarily for describing a specific sequential or chronological order. It is to be understood that the use of the terms so construed can be interchanged, such that the embodiments of the present application described herein are capable of operation in other sequences than those described or otherwise illustrated herein. Further, the terms "comprise" and "include", and variations thereof, are intended to cover a non-exclusive inclusion such that a process, method, system, product, or apparatus that comprises a list of steps or units are not necessarily limited to those steps or units but can include other not expressly listed steps or units.

[0066] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of the relevant data need to comply with relevant laws, regulations and standards, and provide corresponding operation portal for the user to choose authorization or refusal.

[0067] The working environment of the control cabinet is usually harsh, and the measurement device for collecting the contact operation data in the control cabinet is easily affected by different working conditions (complex working conditions) such as strong electromagnetic field, high temperature, large current, high voltage, etc. Therefore, the collected operation data has many noise signals, and the data has fluctuations in different degrees. However, in the related art, when predicting the temperature at a future time, the noise interference on the operation data is not considered, and the operation data obtained is processed by applying a machine learning model, resulting in poor accuracy of the predicted temperature data. In addition, in complex working conditions, it is difficult for traditional denoising methods to separate noise and useful signals. The working conditions in the control cabinet are complex, and only based on the operation data in a single complete time period, the temperature is directly predicted, and the prediction result is easily affected by the abnormal trend fluctuations in the time period, thereby producing deviation, for example, at some time points, the operation data suddenly changes and abnormally fluctuates, which greatly affects the overall prediction result. Moreover, under a single time window, the model cannot capture both short-term sudden changes and long-term trends, therefore, the related art has low contact temperature prediction accuracy, and thus cannot accurately predict the control cabinet fault.

[0068] To solve the above technical problems, the control cabinet fault early warning method provided by the present application analyzes the operation data of the contact from different time spans, and further weights and fuses to obtain the predicted temperature of the contact at a future time, which comprehensively considers the information of different time spans, reduces the influence of the trend fluctuations of data anomalies in a set time period, and improves the accuracy of temperature prediction. By wavelet packet denoising, noise interference is effectively suppressed, and support vector machine is applied to effectively capture the nonlinear relationship existing in the operation data, further improve the accuracy of the predicted temperature, and significantly improve the accuracy and reliability of the control cabinet fault early warning.

[0069] The technical solutions of the present application and how the technical solutions of the present application solve the above technical problems will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present application will be described below with reference to the accompanying drawings.

[0070] Figure 1 The scene schematic diagram of the control cabinet fault early warning method provided by the embodiments of the present application is as follows Figure 1As shown, the application scenario includes a computing device 11, a measurement device 12, and a control cabinet 13, wherein the control cabinet 13 has at least one contact point, which is an interface or contact surface between conductors in the control cabinet for current to pass through, such as a normally open contact point, a normally closed contact point, an auxiliary contact point, and the like. The operation data of the contact point is collected by the measurement device 12, such as a temperature sensor, a thermocouple, a fiber bragg grating sensor, a current transformer, a Hall current sensor, a voltage transformer, a micro-ohmmeter, and the like. The measurement device 12 can report the operation data of the contact point to the computing device 12, and the computing device 12 is deployed with a control cabinet fault early warning method to realize the fault early warning of the control cabinet 13.

[0071] It should be noted that, Figure 1 Only as an example, the application scenario and the embodiment of the present application are not limited to Figure 1 The computing device 11 can be a server, a server cluster, a virtual resource, or a terminal device with certain computing power, such as a desktop computer, a notebook computer, and the like. The computing device 11 can be a local device or a cloud device. The type and number of the computing device 11 are not shown in the embodiments of the present application.

[0072] Figure 2 The flow of the control cabinet fault early warning method provided in the embodiments of the present application is shown in Figure 1 As shown, Figure 2 The control cabinet fault early warning method includes:

[0073] S201, obtaining the operation data of the contact point in the control cabinet within a set time length.

[0074] The operation data includes electrical parameters (such as current, voltage, power factor), environmental parameters (such as ambient temperature, humidity), and state parameters (such as contact point temperature, contact resistance, pressure). It should be understood that the operation data will directly or indirectly affect the predicted temperature of the contact point at a future time. The set time length is, for example, 1 minute, 5 minutes, 10 minutes, and the like.

[0075] For example, the operation data can be directly obtained from each measurement device, or obtained from a power grid automation system through a communication interface, the power grid automation system integrates the monitoring and control of each control cabinet, including the data collection of the measurement device, or obtained from a cloud platform, the cloud platform can integrate and store data from different systems and different measurement devices.

[0076] It should be noted that for scenarios with high real-time requirements, the set time length is preferably set to be small to quickly respond to abnormalities; for scenarios with low real-time requirements, the set time length can be set to be large to capture more data information. According to the scene and the demand, the calculation efficiency and the trend capturing ability are balanced.

[0077] S202, performing multi-scale segmentation on the operation data to obtain data sequences of multiple time spans.

[0078] It can be understood that each data in the operation data is a continuous time sequence, and each data is synchronously collected, that is, the data timestamps are consistent, and the subsequent prediction results are avoided from being affected by phase error. The operation data can be regarded as a multi-dimensional data sequence, and the dimension is equal to the number of data types. For example, the operation data includes three types of data of voltage, current and temperature, and the dimension of the operation data is 3. After segmentation, the dimension remains unchanged. Through shorter data sequences, the model can focus on local features and capture fine-grained dependencies, and longer data sequences can help the model focus on a larger time range and capture coarse-grained dependencies. In order to reduce the influence of the data sequence length parameter on the model, the data is segmented by using multiple different time spans, and multiple data sequences obtained by segmentation together constitute a multi-scale representation of the operation data.

[0079] For example, when the operation data is segmented, the time sequence correlation is ensured, different time windows are sequentially used for segmentation in time sequence, for example, for operation data with a set time length of 6 minutes, the time windows are 1 minute, 2 minutes and 3 minutes, respectively, and 3 data sequences of different time spans are obtained, that is, under the condition that the sampling rate remains unchanged, the lengths of the 3 data sequences are also different, and the dimensions are the same.

[0080] S203, for the data sequences respectively corresponding to each time span in the multiple time spans, a wavelet packet analysis algorithm is used to perform denoising processing on the data sequences to obtain denoised data sequences corresponding to the time spans.

[0081] Since the working condition environment of the control and collection cabinet is relatively complex, the data sequence carries more noise signals, thereby generating different frequency signals. The wavelet packet analysis algorithm can separate signal characteristics of different frequency bands, retain low-frequency useful signals while suppressing high-frequency noise, and is especially suitable for non-stationary signals (such as temperature, current, etc.) generated during the operation of the control and collection cabinet contacts, thereby providing higher-quality input data for the temperature prediction model.

[0082] It should be noted that when the data sequence is denoised, the wavelet packet analysis algorithm is used to denoise different dimensions of data contained in the data sequence, for example, by channel processing (one channel processes one type of data), selecting appropriate wavelet basis functions, decomposition levels and threshold strategies according to the data type.

[0083] Optionally, before the wavelet packet denoising processing is performed, the data sequence can be preprocessed, and the preprocessing includes data cleaning and normalization. The data cleaning is to eliminate abnormal values or missing values caused by equipment failure, communication anomaly, human operation, etc., and the normalization is to eliminate data dimension differences.

[0084] S204, input each denoised data sequence to the pre-trained temperature prediction model to obtain a first predicted temperature corresponding to each time span.

[0085] The temperature prediction model is obtained by training a machine learning model based on historical operation data of the control cabinet contact point in advance, and the machine learning model includes a support vector machine.

[0086] Considering that the operation data of the control cabinet contact point has the characteristics of high dimension, time variation and nonlinearity, a support vector machine is selected as the basis of the temperature prediction model in the embodiment. The support vector machine can effectively solve the problems of nonlinearity and high dimension by mapping the input to a high-dimensional feature space through a kernel trick to construct an optimal classification hyperplane, and is not prone to overfitting, so that the temperature prediction model trained based on the support vector machine has good generalization ability and prediction ability.

[0087] In this step, the denoised data sequences of different time spans correspond to a prediction task respectively, and each prediction task corresponds to a predicted temperature. For example, a plurality of independent temperature prediction models or a unified multi-output model (i.e. a multi-task learning model) can be used for temperature prediction to generate the first predicted temperature.

[0088] For training of the temperature prediction model, the sample data includes operation data of a plurality of time periods in history and predicted temperature data corresponding to the time periods. The temperature prediction model is trained to learn the correlation between the historical operation data and the temperature data, so as to accurately predict the temperature at a future time. It can be understood that the end of the model training process can be similar to the related art, that is, the temperature prediction model is considered to be trained when a set number of training times is reached, or the temperature prediction model is obtained when the target regression function value is less than a set threshold.

[0089] S205, based on the first predicted temperature and the weight of the corresponding time span, determine the second predicted temperature of the contact point, and the weight represents the influence degree of the corresponding time span on the predicted temperature.

[0090] It should be understood that this step is to weight and fuse the prediction results of a plurality of time spans. The weight of the time span can be determined by various implementation manners.

[0091] In an implementation, the weight corresponding to each time span is obtained by assigning weights to each time span based on the data sequence of the plurality of time spans using an entropy weight method. Specifically, for the data sequence of each time span, the information entropy of the data sequence is calculated, which reflects the discrete degree of the data sequence. The higher the discrete degree, the smaller the information entropy, and the greater the information utility value, indicating that the data sequence provides a greater amount of effective information. Then, the weight of each data sequence is calculated according to the information utility value of the data sequence of each time span. This implementation uses the entropy weight method to reasonably assign weights to each time span by calculating the information entropy and information utility value of each data sequence, which can reflect the amount of information provided based on the discrete degree of the data itself, and then objectively reflect the influence of each time span on the predicted temperature, avoiding the deviation that may be caused by subjective weighting, and improving the objectivity and accuracy of the second predicted temperature.

[0092] In another implementation, the weight can be determined according to the length of the data sequence. The greater the length of the data sequence, the greater the amount of information it contains, and the greater the weight.

[0093] By weightedly averaging the prediction results of different time spans, for example, the number of time spans is 4, the weights corresponding to each time span are 0.2, 0.3, 0.1 and 0.4 respectively, and the first predicted temperatures corresponding to each time span are 30℃, 35℃, 34℃ and 29℃ respectively, then the second predicted temperature T .

[0094] In S206, when the second predicted temperature is greater than the set temperature threshold, the contact is determined as a potential abnormal contact, and a fault warning is given to the control cabinet.

[0095] In this step, when the second predicted temperature is greater than the set temperature threshold, it indicates that the operating state of the contact may have deviated from the normal state, reflecting that the line, component or other equipment or line associated with the contact in the control cabinet may have an abnormality. Therefore, the contact is determined as a potential abnormal contact.

[0096] The warning can be realized by pushing the corresponding fault warning information to the operation and maintenance personnel. For example, it can be pushed to the mobile terminal (such as a mobile phone) of the operation and maintenance personnel, or to the monitoring large screen of the central control room, etc. For example, the warning information can include the potential abnormal contact, the control cabinet and the occurrence time of the potential abnormality, etc.

[0097] The embodiments of the present application realize analysis of the running data of the contact from different time spans, mine multi-level rules, and further weighted fusion to obtain the predicted temperature of the contact at a future time, by segmenting the running data, applying the data sequence processed by segmentation to predict the temperature, comprehensively integrating information of different time spans, balancing local and global features, reducing the trend fluctuation of data anomalies within a set time, thereby improving the accuracy of temperature prediction. Through wavelet packet denoising, the influence of noise on the prediction accuracy of the model is effectively suppressed, and the support vector machine regression model is applied to effectively capture the nonlinear relationship existing in the running data, improve the adaptability of the model to complex nonlinear running data changes, further improve the accuracy of the predicted temperature, and significantly improve the accuracy and reliability of the fault early warning of the control and distribution cabinet, realize accurate early warning of the fault of the control and distribution cabinet, effectively avoid the expansion and deterioration of the fault in time, improve the operation and maintenance efficiency, and ensure the safe and stable operation of the control and distribution cabinet.

[0098] During wavelet packet decomposition, a suitable decomposition level should be selected. A smaller decomposition level cannot effectively extract different spike frequencies of the running data, and a larger decomposition level will make the calculation redundant, disperse the frequency band energy, and even overfit, affecting the generalization ability of the subsequent temperature prediction model. Therefore, in some embodiments, a wavelet packet analysis algorithm is used to denoise the data sequence to obtain a denoised data sequence corresponding to the time span, specifically including:

[0099] S2031, performing three-layer wavelet packet decomposition on the data sequence to obtain eight wavelet packet coefficients corresponding to the frequency bands of the nodes.

[0100] It should be understood that wavelet packet analysis decomposes the signal (data sequence) into different frequency bands by constructing a complete binary tree structure, each frequency band corresponds to a wavelet packet node, and the wavelet packet node contains the wavelet packet coefficients of the frequency band. The wavelet packet coefficients are the projection of the signal in the corresponding frequency band, reflecting the energy distribution and time-frequency characteristics of the frequency band.

[0101] S2032, processing each wavelet packet coefficient based on the hard threshold method to obtain the denoised wavelet packet coefficients of each node corresponding to the frequency band, and the threshold in the hard threshold method is dynamically determined based on the signal energy.

[0102] In the wavelet packet analysis algorithm, compared with the soft threshold method, the hard threshold method has stronger tracking ability for the change characteristics of the signal, and better restoration ability for signals with drastic changes. Therefore, the hard threshold method is used for denoising in the present embodiment.

[0103] Specifically, the hard threshold method retains the wavelet packet coefficients with an absolute value greater than or equal to the threshold, and directly sets the wavelet packet coefficients with an absolute value less than the threshold to 0, thereby suppressing noise.

[0104] In the traditional hard threshold method, a same threshold (i.e. the basic threshold in the following) is used for all wavelet packet coefficients, which is usually based on statistical theory or empirical formula (such as the minimum maximum variance method) and is related to the signal length and noise level, but once calculated, it will not be changed in the processing process, which may lead to over-denoising, insufficient denoising or over-smoothing. In order to overcome the shortcomings of the traditional method, an adaptive threshold is introduced in this embodiment, i.e. the threshold is dynamically determined based on the signal energy, and different thresholds are set for each frequency band. The wavelet packet coefficients of each frequency band are denoised using the adaptive threshold. The hard threshold processing formula is as follows:

[0105]

[0106] In the formula, c jk is the jth wavelet packet coefficient corresponding to the kth node in the three-layer wavelet packet decomposition tree; is the threshold of the kth node in the three-layer wavelet packet decomposition tree.

[0107] For example, in a possible implementation, the threshold in this step satisfies the following formula:

[0108]

[0109] In the formula, c jk is the jth wavelet packet coefficient corresponding to the kth node in the three-layer wavelet packet decomposition tree; is the basic threshold; is the signal energy of the kth node corresponding frequency band, which represents the sum of squares of all wavelet packet coefficients in the frequency band; is the total energy of all frequency bands in the three-layer wavelet packet decomposition tree; is an adjustment coefficient, which is used to control the influence strength of energy on the threshold.

[0110] For each wavelet packet node, its signal energy can be calculated by the following method: for all wavelet packet coefficients corresponding to each node obtained in step S2031, the sum of squares of all wavelet packet coefficients is calculated to obtain the signal energy of the frequency band corresponding to the node.

[0111] S2033, reconstruct the denoised wavelet packet coefficients to obtain the denoised data sequence corresponding to the time span.

[0112] For example, the denoised wavelet packet coefficients are synthesized layer by layer through inverse wavelet packet transform to finally restore the time domain signal and obtain the denoised data sequence.

[0113] ​​​​​​​​The embodiment of the application adjusts the threshold value through signal energy, improves the processing capability for complex signals while maintaining the sparsity of the hard threshold value, and uses the adaptive threshold value to perform the denoising processing on the signals, thereby significantly improving the adaptability and effect of signal denoising, and better adapting to the noise environment under different working conditions, and providing more accurate and reliable data basis for the prediction of the contact temperature of the power grid control cabinet.

[0114] In some examples, the running data is multi-scale segmented to obtain data sequences of multiple time spans, including: multi-scale segmenting the running data based on a set segmentation algorithm to obtain data sequences of multiple time spans, the set segmentation algorithm satisfying the following form:

[0115]

[0116] In the formula, is the length of the data sequence of the i-th time span; is the length of the data sequence of the i-th time span; is a preset initial segmentation length; is a level of multi-scale segmentation, , is a maximum number of levels of multi-scale segmentation, , is the length of the running data.

[0117] The embodiment of the application can cover different scale ranges from fine granularity to coarse granularity in a limited number of segments through the geometric growth manner. Meanwhile, the segmentation length increases rapidly with the increase of the level, which avoids the high calculation complexity caused by too many segments, and the calculation complexity is relatively low in the way of gradually adding by the exponential power of 2, thereby achieving the effective coverage of the time series information and the balance of the calculation efficiency.

[0118] Further, in a possible implementation manner, the weight satisfies the following calculation formula:

[0119]

[0120] In the formula, is the weight of the i-th time span; is the weight of the i-th time span; is a maximum number of levels of multi-scale segmentation.

[0121] For example, the running data is multi-scale segmented to obtain data sequences of four time spans, i.e. , , , That is, the greater the length of the data sequence, the greater the amount of information contained, and the greater the weight.

[0122] On the basis of the above-mentioned embodiments, in some embodiments, the power control cabinet fault early warning method can further include: determining the deviation of the second predicted temperature relative to the set temperature threshold; if the deviation is within a first threshold range, a first fault early warning strategy is executed, the first fault early warning strategy is used to output first early warning information; if the deviation is within a second threshold range, a second fault early warning strategy is executed, the second fault early warning strategy is used to trigger a regulation mechanism, the regulation mechanism includes enabling a heat dissipation system or reducing a load, and outputting second early warning information; if the deviation is within a third threshold range, a third fault early warning is performed on the power control cabinet, the third fault early warning is used to automatically cut off the power supply of the power control cabinet, and output third early warning information.

[0123] The set temperature threshold can be a safe upper temperature limit determined according to the equipment manufacturer's specifications, industry standards or historical fault data. For example, the maximum temperature allowed at the contact point is 80℃, considering the safety margin, the early warning threshold is set to 75℃, and the fault threshold is 80℃.

[0124] For example, assume that the first threshold range is 0-5℃, the second threshold range is 6-10℃, and the third threshold range is greater than 10℃. Correspondingly, if the deviation is between 0-5℃, the first early warning information is sent to the operation and maintenance personnel through SMS, email or system log, for example, "Please check whether the power control cabinet heat dissipation system (such as fan, air vent) is normal"; optionally, a yellow warning information can also be displayed on the power control cabinet display screen or indicator light. If the deviation is between 6-10℃, start the standby heat dissipation system (such as increase the fan speed) or reduce the load, automatically push the second early warning information to the operation and maintenance platform, and require confirmation of receipt, at the same time trigger the sound and light alarm, the red indicator light flashes. If the deviation is greater than 10℃, the power supply of the power control cabinet is automatically cut off to prevent equipment damage, and relevant responsible persons are notified through telephone, broadcast and other means to prompt relevant personnel to start the emergency plan.

[0125] The embodiments of the present application can realize early detection and accurate intervention of power control cabinet faults by quantifying the deviation of the second predicted temperature from the set threshold, and implementing early warning strategies in stages, which can significantly reduce the failure rate of power control cabinets caused by overheating, and improve the stability of power system operation.

[0126] Figure 3 Flowchart of the power control cabinet fault early warning method provided by the embodiments of the present application Figure 2 As shown in Figure 3 The power control cabinet fault early warning method includes:

[0127] S301, obtaining the running data of the contact points in the power control cabinet within a set time.

[0128] For example, the running data in the current 6 minutes is obtained from the power grid automation system through the communication interface, and the power grid automation system integrates the monitoring and control of each control cabinet, including the data acquisition of the measuring device.

[0129] In S302, the running data is segmented in multiple scales to obtain data sequences of multiple time spans.

[0130] For example, when the running data is segmented, the time sequence correlation is ensured, and different time windows are used in time sequence to segment, for example, for running data with a set time length of 6 minutes, the time windows are 1 minute, 2 minutes and 3 minutes, respectively, to obtain data sequences of 3 time spans, that is, under the condition that the sampling rate is unchanged, the lengths of the 3 data sequences are also different, and the dimensions are the same. The running data can be regarded as a multi-dimensional data sequence, and the dimension is equal to the number of data types. For example, the running data contains three types of data, i.e. voltage, current and temperature, so the dimension of the running data is 3, and the dimension does not change after segmentation.

[0131] In S303, the data sequence corresponding to each time span in the multiple time spans is subjected to three-layer wavelet packet decomposition to obtain wavelet packet coefficients corresponding to eight nodes of frequency bands.

[0132] In S304, each wavelet packet coefficient is processed based on a hard threshold method to obtain a denoised wavelet packet coefficient corresponding to each node of the frequency band, and the threshold in the hard threshold method is dynamically determined based on signal energy.

[0133] Specifically, the wavelet packet coefficients with absolute values greater than or equal to the threshold are retained, and the wavelet packet coefficients with absolute values less than the threshold are directly set to 0, thereby suppressing noise.

[0134] In S305, the denoised wavelet packet coefficients are reconstructed to obtain a denoised data sequence corresponding to the time span.

[0135] For example, the denoised wavelet packet coefficients are synthesized layer by layer through inverse wavelet packet transform to finally restore the time domain signal to obtain the denoised data sequence.

[0136] In S306, each denoised data sequence is input into a pre-trained temperature prediction model to obtain a first predicted temperature corresponding to each time span.

[0137] The temperature prediction model is obtained by training a machine learning model based on historical running data of the control cabinet contact, and the machine learning model includes a support vector machine.

[0138] For example, the target regression function of the temperature prediction model is as follows:

[0139]

[0140] Constraints:

[0141] wherein, is the th input training sample sequence, is the number of training sample sequences; is the th output training sample sequence, is a kernel function used to calculate the similarity between sample sequences and is the allowed deviation between the predicted value and the true value; is a regularization parameter used to balance the complexity of the model and the error. A larger value will result in a tighter fitting of the model to the training data, which may overfit, and a smaller value will make the model smoother, which may underfit. The value is obtained through experimental tuning or selected based on data characteristics; and are Lagrange multipliers used to introduce constraints in optimization problems. By solving the above optimization problem, the optimal Lagrange multipliers and of the temperature prediction model can be obtained.

[0142] Substituting into the decision regression equation, it is used to predict new input samples . The decision regression equation is:

[0143]

[0144] wherein, the decision regression equation combines the optimal Lagrange multipliers, kernel functions and bias terms to predict new samples using the information of training samples. It can be understood that the decision regression equation and the contactor contact operating data are used to predict the contact temperature at a future time point.

[0145] The denoised data sequences of different time spans correspond to a prediction task respectively, and each prediction task corresponds to a predicted temperature. For example, multiple independent temperature prediction models or a unified multi-output model (i.e. a multi-task learning model) can be used for temperature prediction to generate a first predicted temperature.

[0146] S307, based on the first predicted temperature and the weight of the corresponding time span, a second predicted temperature of the contact is determined, and the weight represents the influence degree of the corresponding time span on the predicted temperature.

[0147] ​​The prediction results of multiple time spans are weighted and fused, details of which can be found in step S205, which will not be repeated here.

[0148] S308, determine the deviation of the second predicted temperature relative to the set temperature threshold.

[0149] S309, if the deviation is within the first threshold range, execute the first fault warning strategy; if the deviation is within the second threshold range, execute the second fault warning strategy; if the deviation is within the third threshold range, perform the third fault warning on the control cabinet.

[0150] The first fault warning strategy is used to output the first warning information. For example, send the first warning information to the operation and maintenance personnel through SMS, email or system log, such as "please check whether the heat dissipation system (such as fan, air vent) of the control cabinet is normal"; optionally, a yellow warning information can also be displayed on the display screen or indicator light of the control cabinet.

[0151] The second fault warning strategy is used to trigger the regulation mechanism, which includes enabling the heat dissipation system or reducing the load, and outputting the second warning information. For example, start the standby heat dissipation system (such as increase the fan speed) or reduce the load, automatically push the second warning information to the operation and maintenance platform, and require confirmation of receipt, while triggering the audible and light alarm, and the red indicator light flashes.

[0152] The third fault warning is used to automatically cut off the power supply of the control cabinet. For example, automatically cut off the power supply of the control cabinet to prevent equipment damage, and notify the responsible person through telephone, broadcast and other means to prompt the relevant personnel to start the emergency plan.

[0153] In summary, the present application has at least the following advantages:

[0154] I. By segmenting the running data, applying the segmented data sequence to predict the temperature, analyzing the running data of the contact from different time spans, mining multi-level rules, and further weighting and fusing to obtain the predicted temperature of the contact at future time, the information of different time spans is integrated, the local and global features are balanced, the trend fluctuation of data anomalies in the set time is reduced, thereby improving the accuracy of temperature prediction. By wavelet packet denoising, the noise interference is effectively suppressed, the influence of noise on the prediction accuracy of the model is reduced, and the support vector machine regression model is applied to effectively capture the nonlinear relationship existing in the running data, improve the adaptability of the model to complex nonlinear running data changes, further improve the accuracy of the predicted temperature, and thus significantly improve the accuracy and reliability of the control cabinet fault warning, realize the precise early warning of the control cabinet fault, avoid the expansion and deterioration of the fault in time and effectively, improve the operation efficiency, and ensure the safe and stable operation of the control cabinet.

[0155] II. By adaptively adjusting the threshold value based on signal energy, the processing capability for complex signals is improved while maintaining the sparsity of hard threshold, and the adaptability and effectiveness of signal denoising are significantly improved by using an adaptive threshold for signal denoising, thereby better adapting to different noise environments under different working conditions and providing more accurate and reliable data basis for power grid control cabinet contact temperature prediction.

[0156] III. By quantifying the deviation of the second predicted temperature from the set threshold and implementing a warning strategy in stages, early detection and precise intervention of the control cabinet fault can be achieved, which can significantly reduce the failure rate of the control cabinet caused by overheating and improve the stability of the power system operation.

[0157] Figure 4 The structure diagram of the control cabinet fault early warning device provided by the embodiments of the present application is shown in Figure 4 The control cabinet fault early warning device 40 provided by the embodiments of the present application includes an acquisition module 41, a first processing module 42, a second processing module 43, a determination module 44, and a warning module 45. Among them:

[0158] The acquisition module 41 is configured to acquire the running data of the contacts in the control cabinet within a set time period.

[0159] The first processing module 42 is configured to perform multi-scale segmentation on the running data to obtain a plurality of data sequences of time spans.

[0160] The first processing module 42 is further configured to, for each data sequence corresponding to each time span in the plurality of time spans, perform denoising processing on the data sequence using a wavelet packet analysis algorithm to obtain a denoised data sequence corresponding to the time span.

[0161] The second processing module 43 is configured to input each denoised data sequence into a pre-trained temperature prediction model to obtain a first predicted temperature corresponding to each time span, wherein the temperature prediction model is obtained by training a machine learning model based on historical running data of the contacts of the control cabinet, and the machine learning model includes a support vector machine.

[0162] The determination module 44 is configured to determine a second predicted temperature of the contacts based on the first predicted temperature and a weight of the corresponding time span, wherein the weight represents the influence degree of the corresponding time span on the predicted temperature.

[0163] The warning module 45 is configured to determine the contacts as potential abnormal contacts and perform fault warning on the control cabinet when the second predicted temperature is greater than a set temperature threshold.

[0164] In a possible implementation, the first processing module 42 is specifically configured to perform three-layer wavelet packet decomposition on the data sequence to obtain eight wavelet packet coefficients corresponding to frequency bands of eight nodes; perform processing on the wavelet packet coefficients based on a hard threshold method to obtain wavelet packet coefficients of the nodes corresponding to the frequency bands after noise reduction, wherein a threshold in the hard threshold method is dynamically determined based on signal energy; and reconstruct the wavelet packet coefficients after noise reduction to obtain a data sequence after noise reduction corresponding to a time span.

[0165] In a possible implementation, the threshold satisfies the following formula:

[0166]

[0167] wherein is a threshold of an i th node in a three-layer wavelet packet decomposition tree; is a threshold of an i th node in a three-layer wavelet packet decomposition tree; is a basic threshold; is a signal energy of a frequency band corresponding to an i th node, and the signal energy represents a sum of squares of wavelet packet coefficients in the frequency band; is a total energy of all frequency bands in the three-layer wavelet packet decomposition tree; is an adjustment coefficient, used to control an influence strength of the energy on the threshold.

[0168] In a possible implementation, the first processing module 42 is further configured to perform multi-scale segmentation on the running data based on a set segmentation algorithm to obtain data sequences of multiple time spans, and the set segmentation algorithm satisfies the following form:

[0169]

[0170] wherein is a length of a data sequence of an i th time span; is a preset initial segmentation length; is a level of multi-scale segmentation, is a maximum level number of multi-scale segmentation, is a length of the running data.

[0171] In a possible implementation, the weight satisfies the following calculation formula:

[0172]

[0173] wherein is a weight of an i th time span; is a maximum level number of multi-scale segmentation. In a possible implementation, the first processing module 42 is specifically configured to perform three-layer wavelet packet decomposition on the data sequence to obtain eight wavelet packet coefficients corresponding to frequency bands of eight nodes; perform processing on the wavelet packet coefficients based on a hard threshold method to obtain wavelet packet coefficients of the nodes corresponding to the frequency bands after noise reduction, wherein a threshold in the hard threshold method is dynamically determined based on signal energy; and reconstruct the wavelet packet coefficients after noise reduction to obtain a data sequence after noise reduction corresponding to a time span.

[0165] In a possible implementation, the threshold satisfies the following formula:

[0166]

[0167] wherein is a threshold of an i th node in a three-layer wavelet packet decomposition tree; is a threshold of an i th node in a three-layer wavelet packet decomposition tree; is a basic threshold; is a signal energy of a frequency band corresponding to an i th node, and the signal energy represents a sum of squares of wavelet packet coefficients in the frequency band; is a total energy of all frequency bands in the three-layer wavelet packet decomposition tree; is an adjustment coefficient, used to control an influence strength of the energy on the threshold.

[0168] In a possible implementation, the first processing module 42 is further configured to perform multi-scale segmentation on the running data based on a set segmentation algorithm to obtain data sequences of multiple time spans, and the set segmentation algorithm satisfies the following form:

[0169]

[0170] wherein is a length of a data sequence of an i th time span; is a preset initial segmentation length; is a level of multi-scale segmentation, is a maximum level number of multi-scale segmentation, is a length of the running data.

[0171] In a possible implementation, the weight satisfies the following calculation formula:

[0172]

[0173] wherein is a weight of an i th time span; is a maximum level number of multi-scale segmentation.

[0174] In a possible implementation, the weight corresponding to each time span is obtained based on the data sequence of the plurality of time spans and an entropy weight method.

[0175] In a possible implementation, the early warning module 45 is specifically configured to: determine a deviation of the second predicted temperature from a set temperature threshold; if the deviation is within a first threshold range, execute a first fault early warning strategy, the first fault early warning strategy being configured to output first early warning information; if the deviation is within a second threshold range, execute a second fault early warning strategy, the second fault early warning strategy being configured to trigger a regulation mechanism, the regulation mechanism including enabling a heat dissipation system or reducing a load, and output second early warning information; and if the deviation is within a third threshold range, perform a third fault early warning on the control cabinet, the third fault early warning being configured to automatically cut off power supply of the control cabinet, and output third early warning information.

[0176] The control cabinet fault early warning device provided in the embodiment can execute the method provided in the method embodiment, and has similar implementation principles and technical effects, which will not be described here again in the embodiment.

[0177] Figure 5 A structural schematic diagram of an electronic device provided in the embodiment is shown in FIG. 1. Figure 5 As shown in FIG. 1, the electronic device 50 provided in the embodiment includes at least one processor 501 and a memory 502. Optionally, the electronic device 50 further includes a communication component 503. The processor 501, the memory 502, and the communication component 503 are connected through a bus 504.

[0178] In the implementation process, the at least one processor 501 executes computer execution instructions stored in the memory 502, so that the at least one processor 501 executes the method described above.

[0179] The specific implementation process of the processor 501 can refer to the method embodiment described above, which has similar implementation principles and technical effects, and will not be described here again in the embodiment.

[0180] In the above embodiment, it should be understood that the processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The steps of the method disclosed in the application can be directly embodied as execution completed by a hardware processor, or executed by a combination of hardware and software modules in the processor.

[0181] The memory can include a Random Access Memory (RAM) and can also include a Non-volatile Memory (NVM), such as at least one disk memory.

[0182] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, the bus in the drawings of the present application does not limit to only one bus or one type of bus.

[0183] The embodiment of the present application further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the method described above.

[0184] The embodiment of the present application further provides a computer readable storage medium, which stores computer execution instructions, and when a processor executes the computer execution instructions, the method described above is implemented.

[0185] The readable storage medium described above can be realized by any type of volatile or non-volatile storage device or a combination thereof, such as a Static Random Access Memory (SRAM), an Electrically Erasable Programmable Read-Only Memory (EEPROM), an Erasable Programmable Read-Only Memory (EPROM), a Programmable Read-Only Memory (PROM), a Read-Only Memory (ROM), a magnetic storage, a flash memory, a magnetic disk or an optical disk. The readable storage medium can be any available medium that can be accessed by a general or special purpose computer.

[0186] An exemplary readable storage medium is coupled to the processor, so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be an integral part of the processor. The processor and the readable storage medium can be located in an Application Specific Integrated Circuit (ASIC). Of course, the processor and the readable storage medium can also exist as discrete components in the device.

[0187] The division of units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0188] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, can be located in one place, or can be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0189] In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit.

[0190] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the part of the present application that essentially contributes to the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0191] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware. The aforementioned program can be stored in a computer readable storage medium. The program executes the steps including the above-mentioned method embodiments when executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, and various program code storage media.

[0192] It should be understood that many of the materials and devices exemplified in this disclosure are articles of manufacture (i.e., articles of manufacture) according to this disclosure. The articles of manufacture can be manufactured as such or can be manufactured by combining the materials and devices exemplified in this disclosure. It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. It should be understood that, in some embodiments, equivalents to the specific electrode structures and / or methods described herein can be employed without departing from the scope of the application. Accordingly, the phraseology and terminology used herein is for the purpose of description and should not be regarded as limiting. The use of "including," "comprising," "having," "containing," "involving," "characterized by," "characterized into," and variations thereof herein, is meant to encompass the items listed thereafter, and equivalents thereof as well as additional items. Although the foregoing application has been described in some detail by way of illustration and example, it is not to be limited thereby, but rather, only by the scope of the appended claims.

Claims

1. A method for early warning of faults in a control cabinet, characterized in that, include: Acquire the operating data of the contacts in the control cabinet within a set time period; The operational data is segmented into multiple scales to obtain data sequences spanning multiple time spans; For the data sequences corresponding to each of the multiple time spans, a wavelet packet analysis algorithm is used to denoise the data sequences to obtain the denoised data sequences corresponding to the time spans. Each of the denoised data sequences is input into a pre-trained temperature prediction model to obtain the first predicted temperature corresponding to each of the time spans. The temperature prediction model is a machine learning model pre-trained based on the historical operating data of the control cabinet contacts. The machine learning model includes a support vector machine. Based on the weight of the first predicted temperature and the corresponding time span, the second predicted temperature of the contact point is determined, whereby the weight represents the degree of influence of the corresponding time span on the predicted temperature. When the second predicted temperature is greater than the set temperature threshold, the contact point is identified as a potential abnormal contact point, and a fault warning is issued to the control cabinet.

2. The method for early warning of control cabinet faults according to claim 1, characterized in that, The data sequence is denoised using a wavelet packet analysis algorithm to obtain the denoised data sequence corresponding to the time span, including: The data sequence is subjected to three-level wavelet packet decomposition to obtain wavelet packet coefficients for each of the eight nodes corresponding to the frequency bands. The wavelet packet coefficients of each node are processed by the hard thresholding method to obtain the denoised wavelet packet coefficients of the corresponding frequency bands. The threshold in the hard thresholding method is dynamically determined based on the signal energy. The denoised wavelet packet coefficients are reconstructed to obtain the denoised data sequence corresponding to the time span.

3. The method for early warning of control cabinet faults according to claim 2, characterized in that, The threshold satisfies the following formula: In the formula, In a three-layer wavelet packet decomposition tree, the first... Threshold for each node; Basic threshold; For the first The signal energy of each node corresponds to a frequency band, and the signal energy represents the sum of squares of the wavelet packet coefficients within the frequency band; This represents the total energy of all frequency bands in the three-layer wavelet packet decomposition tree. This is an adjustment coefficient used to control the intensity of the energy's influence on the threshold. .

4. The method for early warning of control cabinet faults according to any one of claims 1 to 3, characterized in that, The process of segmenting the operational data into multiple scales yields data sequences spanning multiple time spans, including: Based on a defined segmentation algorithm, the running data is segmented into multiple scales to obtain data sequences spanning multiple time spans. The defined segmentation algorithm satisfies the following form: In the formula, For the first The length of the data sequence spanning the time layer; The preset initial segment length; It is a multi-scale segmented hierarchy. , The maximum number of levels in multi-scale segmentation. , The length of the data being processed.

5. The method for early warning of control cabinet faults according to claim 4, characterized in that, The weights satisfy the following calculation formula: In the formula, For the first Weights for the time span of each layer; This represents the maximum number of levels in the multi-scale segmentation.

6. The method for early warning of control cabinet faults according to any one of claims 1 to 3, characterized in that, The weights corresponding to each time span are obtained by assigning weights to each time span based on the data sequences of the multiple time spans using the entropy weight method.

7. The method for early warning of control cabinet faults according to any one of claims 1 to 3, characterized in that, Also includes: Determine the deviation of the second predicted temperature from the set temperature threshold; If the deviation is within the first threshold range, then the first fault warning strategy is executed, and the first fault warning strategy is used to output the first warning information; If the deviation is within the second threshold range, a second fault warning strategy is executed. The second fault warning strategy is used to trigger a control mechanism, which includes activating the heat dissipation system or reducing the load, and outputting a second warning message. If the deviation is within the third threshold range, a third fault warning is issued for the control cabinet. The third fault warning is used to automatically cut off the power supply to the control cabinet and output the third warning information.

8. A fault early warning device for a control cabinet, characterized in that, include: The acquisition module is used to acquire the operating data of the contacts in the control cabinet within a set time period; The first processing module is used to perform multi-scale segmentation on the running data to obtain data sequences with multiple time spans; The first processing module is further configured to perform denoising processing on the data sequences corresponding to each time span in the plurality of time spans using a wavelet packet analysis algorithm, so as to obtain the denoised data sequences corresponding to the time spans. The second processing module is used to input the denoised data sequences into the pre-trained temperature prediction model to obtain the first predicted temperature corresponding to each time span. The temperature prediction model is pre-trained on a machine learning model based on the historical operating data of the control cabinet contacts. The machine learning model includes a support vector machine. The determination module is used to determine the second predicted temperature of the contact point based on the first predicted temperature and the weight of the corresponding time span, wherein the weight represents the degree of influence of the corresponding time span on the predicted temperature; The early warning module is used to identify the contact as a potential abnormal contact when the second predicted temperature is greater than the set temperature threshold, and to provide a fault warning for the control cabinet.

9. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed, are used to implement the method as described in any one of claims 1 to 7.

Images