Analytical judgment method for monitoring failure point of filter and equipment medium

Abstract

The embodiment of the invention discloses an analytical judgment method and device for monitoring a failure point of a filter and a medium, belongs to the technical field of filter detection, and solves the problem that time and labor are consumed due to the fact that a plurality of filter cartridges need to be troubleshot one by one when the filter is subjected to troubleshooting. Acquiring corresponding gas output end data samples of the to-be-detected multi-filter-cartridge filter under the condition that different filter cartridges have faults; feature data extraction is carried out on the gas output end data sample, and first feature data matrix sets corresponding to different filter cartridge faults are constructed based on the extracted feature data; obtaining a first feature data matrix set distribution association relationship corresponding to different filter cartridge faults through the neural network graph; performing data enhancement processing on the first feature data matrix set to obtain second feature data matrix sets corresponding to different filter cartridge faults; and constructing a fault point detection model based on the second feature data matrix set so as to perform fault point detection on the multi-cartridge filter.

Description

Technical Field

[0001] The present application relates to the technical field of filter detection, and in particular to an analysis and judgment method, device and medium for monitoring filter failure points. Background Art

[0002] In today's industrial production and environmental purification fields, multi-cartridge filters, as a highly efficient gas purification equipment, play a vital role. Through multiple built-in filter cartridges, they can effectively remove particulate matter, harmful gases or other pollutants in the air to ensure that the emitted gas meets environmental standards or production process requirements. However, as the use time increases, the filter cartridge may gradually fail due to blockage, damage or aging, which in turn affects the performance and efficiency of the entire filtration system. Therefore, timely and accurate detection of the fault status of multi-cartridge filters is of great significance for maintaining production safety, environmental protection and cost control.

[0003] Most existing multi-filter cartridge fault detection methods rely on installing sensors at the gas output end to monitor the concentration changes of the output gas in real time. However, when the system prompts a filter failure, maintenance personnel often take a one-by-one investigation approach to check the condition of each filter cartridge one by one. This process is not only time-consuming and labor-intensive, reducing the timeliness of fault response, but may also increase the risk of filter cartridge damage due to frequent disassembly and assembly operations, further increasing maintenance costs. Summary of the invention

[0004] The embodiments of the present application provide an analysis and judgment method, device and medium for monitoring filter failure points, which are used to solve the following technical problems: When troubleshooting filters, maintenance personnel often check multiple filter cartridges one by one. This process is not only time-consuming and labor-intensive, but also reduces the timeliness of fault response. It may also increase the risk of filter cartridge damage due to frequent disassembly and assembly operations, further increasing maintenance costs.

[0005] The present application embodiment adopts the following technical solutions:

[0006] An embodiment of the present application provides a method for analyzing and determining failure points of monitoring filter failures, including obtaining gas output end data samples corresponding to different filter cartridge failures of a multi-filter cartridge filter to be tested; wherein a plurality of gas detection devices are evenly spaced at the gas output end; the plurality of gas detection devices correspond to detecting different filter cartridge areas, respectively; feature data is extracted from the gas output end data samples, and a first feature data matrix set corresponding to different filter cartridge failures is constructed based on the extracted feature data; the first feature data matrix set is input into a neural network graph, and a distribution correlation relationship of the first feature data matrix sets corresponding to different filter cartridge failures is obtained through the neural network graph; based on the distribution correlation relationship of the first feature data matrix set, data enhancement processing is performed on the first feature data matrix set to obtain second feature data matrix sets corresponding to different filter cartridge failures; a fault point detection model is constructed based on the second feature data matrix set, so as to perform fault point detection on the multi-filter cartridge filter based on the fault point detection model and the gas output end data currently obtained.

[0007] The embodiment of the present application evenly arranges multiple detection devices at the gas output end to independently monitor different filter cartridge areas, thereby more accurately locating the faulty filter cartridge and reducing the possibility of false alarms and missed alarms. By constructing a feature data matrix set, the data characteristics of different filter cartridge failures can be accurately analyzed and compared, providing convenience for subsequent data processing and model training. The feature data matrix set is processed by a neural network to determine the potential correlation between different filter cartridge failures, providing a more accurate basis for subsequent fault point detection. The diversity and richness of the data are increased by data enhancement processing, and the generalization ability and robustness of the model are further improved, so that it can more accurately identify the fault point when facing new data. By constructing a model based on the second feature data matrix set, the extracted features and correlations are fully utilized to achieve accurate detection of the fault points of multi-filter cartridge filters. This not only improves the efficiency and accuracy of fault detection, but also reduces maintenance costs and time costs.

[0008] In one implementation of the present application, feature data is extracted from gas output data samples, and first feature data matrix sets corresponding to different filter cartridge failures are constructed based on the extracted feature data, specifically including: time domain feature extraction of gas output data samples; and gas concentration feature extraction of gas output data samples through Hilbert transform; based on feature type, time domain features and gas concentration features are divided; wherein the time domain features include at least one of peak time, root mean square value and waveform factor; the gas concentration features include at least instantaneous amplitude extraction value and gas concentration frequency extraction value; when the current filter cartridge to be tested fails, different types of features corresponding to multiple gas detection devices are determined, and based on different types of features, multiple feature data matrices are constructed to obtain a first feature data matrix set corresponding to the current faulty filter cartridge to be tested.

[0009] In one implementation of the present application, a first feature data matrix set is input into a neural network graph, and the distribution association relationship of the first feature data matrix sets corresponding to different filter cartridge failures is obtained through the neural network graph, specifically including: inputting the first feature data matrix set corresponding to the first time period into the graph neural network; in the graph neural network, determining the gas concentration difference between the key points of each preset filter cartridge corresponding to the first time period; wherein the key points of the preset filter cartridge are evenly arranged in the graph neural network, and the positions of the key points of the preset filter cartridge are related to the actual positions of each filter cartridge; inputting the first feature data matrix set corresponding to the next time period into the graph neural network; in the graph neural network, determining the gas concentration difference between the key points of each preset filter cartridge corresponding to the next time period; until the first feature data matrix sets are input into the graph neural network in the order of the extraction time of the data samples at the gas output end, and the gas concentration difference corresponding to multiple time periods is obtained; determining the key point relationship score based on the gas concentration difference, and determining the distribution association relationship of the first feature data matrix sets corresponding to different filter cartridge failures based on the changes in the gas concentration difference and the key point relationship score corresponding to the multiple time periods.

[0010] In one implementation of the present application, based on the changes in the gas concentration differences and key point relationship scores corresponding to multiple time periods, the distribution association relationships of the first characteristic data matrix sets corresponding to different filter cartridge failures are determined, specifically including: determining the node distances between the key points of each pre-set filter cartridge, and obtaining the first node score based on the node distance; determining the second node score based on the gas concentration differences between the key points of each pre-set filter cartridge; obtaining the key point relationship score based on the first node score and the second node score; constructing a gas concentration change curve between the key points of each pre-set filter cartridge based on the gas concentration differences corresponding to multiple time periods; and constructing a key point relationship score change curve between the key points of each pre-set filter cartridge based on the key point relationship scores corresponding to multiple time periods; determining the distribution association relationships of the first characteristic data matrix sets corresponding to different filter cartridge failures based on the gas concentration change curve and the relationship score change curve.

[0011] In one implementation of the present application, based on the distribution association relationship of the first feature data matrix set, data enhancement processing is performed on the first feature data matrix set, specifically including: normalizing the first feature data matrix set to scale the feature data; performing feature polynomial combination processing on multiple feature data matrices corresponding to the same fault filter cartridge to generate new features; pre-setting the GANs model with the distribution association relationship of the first feature data matrix set and the new feature input to generate a first new data sample; pre-setting the VAEs model with the distribution association relationship of the first feature data matrix set and the new feature input to generate a second new data sample; and performing data enhancement processing on the first feature data matrix set based on the new features, the first new data sample and the second new data sample.

[0012] In one implementation of the present application, a fault point detection model is constructed based on a second feature data matrix set, specifically including: training a preset neural network model based on the second feature data matrix set; determining the derivative vector groups corresponding to each second feature data matrix set through a global self-attention mechanism, multiplying and summing the derivative vector groups to obtain the global most relevant information corresponding to each second feature data matrix set, so as to adjust the preset neural network model based on the global most relevant information; updating the weights of the adjusted preset neural network model through a back propagation algorithm; and obtaining a fault point detection model when the training results meet the requirements.

[0013] In one implementation of the present application, before performing fault point detection on a multi-cartridge filter based on a fault point detection model and currently acquired gas output end data, the method also includes: inputting the currently acquired gas output end data into a first time series detection model to obtain a time feature sequence of the multi-cartridge filter to be tested within a current time period; and performing frequency domain feature extraction on the currently acquired gas output end data, inputting the frequency domain features into a second time series prediction model to obtain a frequency domain feature sequence corresponding to the multi-cartridge filter to be tested in the current time period; obtaining gas composition data corresponding to the air inlet end of the multi-cartridge filter to be tested in the current time period, and determining a warning threshold based on the gas composition data; obtaining the total number of time features in the time feature sequence and frequency domain features in the frequency domain feature sequence, and generating warning information when the ratio between the number of abnormal features and the total number is greater than the warning threshold.

[0014] In one implementation of the present application, fault point detection is performed on a multi-cartridge filter based on a fault point detection model and currently acquired gas output end data, specifically including: inputting the currently acquired gas output end data into the fault point detection model to output multiple reference fault filter cartridge information based on the fault point detection model; and performing a fault alarm based on the priority of multiple reference fault filter cartridge information.

[0015] The embodiment of the present application provides an analysis and judgment device for monitoring filter failure points, including: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can: obtain gas output end data samples corresponding to different filter cartridge failures of the multi-filter cartridge filter to be tested; wherein a plurality of gas detection devices are evenly spaced at the gas output end; the plurality of gas detection devices correspond to detecting different filter cartridge areas respectively; feature data is extracted from the gas output end data samples, and a first feature data matrix set corresponding to different filter cartridge failures is constructed based on the extracted feature data; the first feature data matrix set is input into a neural network graph, and the distribution correlation relationship of the first feature data matrix set corresponding to different filter cartridge failures is obtained through the neural network graph; based on the distribution correlation relationship of the first feature data matrix set, the first feature data matrix set is data enhanced to obtain second feature data matrix sets corresponding to different filter cartridge failures; a fault point detection model is constructed based on the second feature data matrix set, so as to perform fault point detection on the multi-filter cartridge filter based on the fault point detection model and the currently acquired gas output end data.

[0016] A non-volatile computer storage medium provided in an embodiment of the present application stores computer executable instructions, wherein the computer executable instructions are configured to: obtain gas output end data samples corresponding to different filter cartridge failures of a multi-filter cartridge filter to be tested; wherein a plurality of gas detection devices are evenly spaced at the gas output end; and the plurality of gas detection devices correspond to detecting different filter cartridge areas, respectively; extract characteristic data from the gas output end data samples, and construct first characteristic data matrix sets corresponding to different filter cartridge failures based on the extracted characteristic data; input the first characteristic data matrix set into a neural network graph, and obtain distribution association relationships of the first characteristic data matrix sets corresponding to different filter cartridge failures through the neural network graph; perform data enhancement processing on the first characteristic data matrix set based on the distribution association relationship of the first characteristic data matrix set, and obtain second characteristic data matrix sets corresponding to different filter cartridge failures; and construct a fault point detection model based on the second characteristic data matrix set, so as to perform fault point detection on the multi-filter cartridge filter based on the fault point detection model and the gas output end data currently obtained.

[0017] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects: the embodiments of the present application evenly set multiple detection devices at the gas output end to independently monitor different filter cartridge areas, thereby more accurately locating the faulty filter cartridge and reducing the possibility of false alarms and missed alarms. By constructing a feature data matrix set, the data characteristics of different filter cartridge failures can be accurately analyzed and compared, providing convenience for subsequent data processing and model training. The feature data matrix set is processed by a neural network to determine the potential correlation between different filter cartridge failures, providing a more accurate basis for subsequent fault point detection. By increasing the diversity and richness of data through data enhancement processing, the generalization ability and robustness of the model are further improved, so that it can more accurately identify fault points when facing new data. By constructing a model based on the second feature data matrix set, the extracted features and correlations are fully utilized to achieve accurate detection of fault points of multi-filter cartridge filters. This not only improves the efficiency and accuracy of fault detection, but also reduces maintenance costs and time costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings:

[0019] Figure 1 A flow chart of a method for analyzing and determining failure points of a monitoring filter provided in an embodiment of the present application;

[0020] Figure 2 A schematic diagram of the structure of an analysis and judgment device for monitoring filter failure points provided in an embodiment of the present application.

[0021] Reference numerals:

[0022] 200: analysis and judgment device for monitoring filter failure points, 201: processor, 202: memory. DETAILED DESCRIPTION

[0023] The embodiments of the present application provide a method, device and medium for analyzing and determining failure points of a monitoring filter.

[0024] In order to enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0025] The technical solution proposed in the embodiment of the present invention is described in detail below with reference to the accompanying drawings.

[0026] Figure 1 A flow chart of a method for analyzing and judging the failure point of a monitoring filter provided in an embodiment of the present application, such as Figure 1 As shown, the analysis and judgment method of monitoring the failure point of the filter includes the following steps:

[0027] S101, obtaining gas output end data samples corresponding to each of the multi-filter cartridge filters under test when different filter cartridges fail.

[0028] In one embodiment of the present application, a plurality of gas detection devices are evenly spaced apart at the gas output end of the multi-cartridge filter; the plurality of gas detection devices correspond to and detect different filter cartridge areas, respectively.

[0029] In one embodiment of the present application, multiple gas detection devices need to be evenly spaced at the gas output end of the multi-cartridge filter. The positions of these devices should ensure that the exhaust area of ​​all filter cartridges can be fully covered so as to accurately detect the working status of each filter cartridge. Each gas detection device should detect a different filter cartridge area, usually by dividing the output end into multiple independent detection areas, each area corresponding to one or more filter cartridges.

[0030] Furthermore, in order to obtain data samples when different filter cartridges fail, various fault conditions need to be simulated. For example, a filter cartridge can be artificially blocked, its filter medium can be damaged, or its working pressure can be adjusted to simulate a real fault scenario. In the process of simulating the fault, a gas detection device is used to record the data at the gas output end in real time. These data should include key parameters such as gas concentration, flow rate, temperature, etc. to fully reflect the working status of the filter cartridge. Label each data sample to indicate its corresponding filter cartridge and fault type. For example, numeric or letter codes can be used to represent different filter cartridges and fault types for identification in subsequent data analysis and model training.

[0031] S102, extracting characteristic data from the gas output end data sample, and constructing first characteristic data matrix sets corresponding to different filter cartridge failures based on the extracted characteristic data.

[0032] In one embodiment of the present application, time domain features are extracted from the data samples of the gas output end; and gas concentration features are extracted from the data samples of the gas output end through Hilbert transform. Based on the feature type, the time domain features and gas concentration features are divided; wherein the time domain features include at least one of the peak time, the root mean square value, and the waveform factor; the gas concentration features include at least the instantaneous amplitude extraction value and the gas concentration frequency extraction value. When the current filter cartridge to be tested fails, different types of features corresponding to the multiple gas detection devices are determined, and based on the different types of features, multiple feature data matrices are constructed to obtain the first feature data matrix set corresponding to the current failed filter cartridge to be tested.

[0033] Specifically, the time domain features extracted in the embodiments of the present application may include peak time, root mean square value and waveform factor. The peak time is the time point when the signal reaches its maximum value. In gas concentration monitoring, the peak time indicates the peak moment of pollutant emission; the root mean square value is the square root of the average value of the sum of the squares of the signal amplitudes, which can reflect the energy of the signal. For gas concentration data, it indicates the average concentration level of pollutants; the waveform factor can reflect the waveform characteristics of the signal, such as the degree of sharpness or flatness.

[0034] Furthermore, the embodiment of the present application uses the instantaneous amplitude of the Hilbert transform to react to the instantaneous change of the gas concentration, and performs frequency domain analysis on the instantaneous amplitude to obtain a frequency component or characteristic value, which reflects the frequency characteristics of the gas concentration changing with time. For each gas detection device, a set of time domain features and gas concentration features can be extracted, and these features can be organized according to the feature type to construct a first feature data matrix set. For example, in the case of a fault in the current filter cartridge to be tested, the instantaneous amplitude extraction values ​​corresponding to each detection device obtained are organized to obtain a corresponding feature data matrix, whose instantaneous amplitude extraction value changes with time. Therefore, in the current time period, multiple time-continuous feature data matrices can be obtained to construct the first feature data matrix set corresponding to the current faulty filter cartridge to be tested.

[0035] S103, inputting the first characteristic data matrix set into a neural network graph, and obtaining the distribution correlation relationship of the first characteristic data matrix sets corresponding to different filter cartridge failures through the neural network graph.

[0036] In one embodiment of the present application, the first feature data matrix set corresponding to the first time period is input into the graph neural network. In the graph neural network, the gas concentration difference between the key points of each preset filter cartridge corresponding to the first time period is determined; wherein, the key points of the preset filter cartridge are evenly arranged in the graph neural network, and the positions of the key points of the preset filter cartridge are related to the actual positions of each filter cartridge. The first feature data matrix set corresponding to the next time period is input into the graph neural network. In the graph neural network, the gas concentration difference between the key points of each preset filter cartridge corresponding to the next time period is determined. Until the first feature data matrix set is input into the graph neural network in the order of the extraction time of the data samples at the gas output end, the gas concentration difference corresponding to multiple time periods is obtained. The key point relationship score is determined based on the gas concentration difference, and the distribution correlation relationship of the first feature data matrix set corresponding to different filter cartridge failures is determined based on the change of the gas concentration difference and the key point relationship score corresponding to the multiple time periods.

[0037] Specifically, in the graph neural network, the preset filter cartridge key points are related to the positions of each filter cartridge, the edges between nodes represent the spatial relationship or connection between the filter cartridges, and the eigenvector of the node is composed of the eigenvalues ​​in the first eigendata matrix set. The first eigendata matrix set of each time period is used as the input of the graph neural network, and each eigendata matrix contains the time domain features and gas concentration features of all filter cartridges in that time period.

[0038] Furthermore, features related to gas concentration, such as instantaneous amplitude extraction values, are extracted from the node feature vectors of the graph neural network. For each time period, the gas concentration differences between the key points (i.e., nodes) of each preset filter cartridge are calculated. These differences reflect the relative changes in gas concentration between the filter cartridges. According to the time sequence of the data sample extraction at the gas output end, the first feature data matrix set of each time period is input into the graph neural network in sequence. For each input time period, the graph neural network will output the gas concentration difference between the key points of each filter cartridge in that time period.

[0039] Furthermore, based on the gas concentration difference, the relationship score between the key points of each filter cartridge in each time period can be calculated, and the relationship score reflects the degree of mutual influence or the strength of association between the filter cartridges. Based on the changes in the gas concentration difference and the key point relationship score in multiple time periods, the distribution association relationship of the first characteristic data matrix set corresponding to different filter cartridge failures is determined.

[0040] In one embodiment of the present application, the node distance between the key points of each pre-set filter cartridge is determined, and a first node score is obtained based on the node distance. The second node score is determined based on the gas concentration difference between the key points of each pre-set filter cartridge. Based on the first node score and the second node score, a key point relationship score is obtained. Based on the gas concentration difference corresponding to a plurality of time periods, a gas concentration change curve between the key points of each pre-set filter cartridge is constructed; and based on the key point relationship scores corresponding to a plurality of time periods, a key point relationship score change curve between the key points of each pre-set filter cartridge is constructed. Based on the gas concentration change curve and the relationship score change curve, the distribution correlation relationship of the first characteristic data matrix set corresponding to different filter cartridge failures is determined.

[0041] Specifically, in the graph neural network, each pre-set filter cartridge key point (node) represents a filter cartridge or a key position of the filter cartridge. Based on the node distance, a score is assigned to each node, namely the first node score. For each time period, the gas concentration difference between the pre-set filter cartridge key points is calculated. According to the gas concentration difference, another score is assigned to each node, namely the second node score. Combining the first node score and the second node score, the key point relationship score between the pre-set filter cartridge key points in each time period can be obtained.

[0042] Furthermore, based on the gas concentration differences corresponding to multiple time periods, gas concentration change curves between key points of each pre-set filter cartridge are constructed, and these curves show the changing trend of gas concentration over time. Similarly, based on the key point relationship scores corresponding to multiple time periods, key point relationship score change curves between key points of each pre-set filter cartridge are constructed, and these curves show the changes of key point relationship scores over time. Combining the gas concentration change curves and the key point relationship score change curves, we can have a more comprehensive understanding of the correlation between different filter cartridges and their changes in different time periods.

[0043] For example, assuming that the distance between filter cartridge 1 and filter cartridge 2 is relatively close, and the distance between filter cartridge 3 and them is relatively far, therefore, the first node scores of filter cartridge 1 and filter cartridge 2 are relatively high, and the first node score of filter cartridge 3 is relatively low. In time period 1, the difference in gas concentration between filter cartridge 1 and filter cartridge 2 is relatively small, and the difference between filter cartridge 3 and them is relatively large, therefore, the second node scores of filter cartridge 1 and filter cartridge 2 are relatively high, and the second node score of filter cartridge 3 is relatively low. Combining the first node score and the second node score, the key point relationship score between the key points of each filter cartridge in each time period is calculated. The gas concentration change curve and the key point relationship score change curve are constructed. Through the curve, it can be obtained that the gas concentration of filter cartridge 3 suddenly increases in time period 3, and its key point relationship score with other filter cartridges is also significantly reduced. Based on the analysis of these curves, it is determined that filter cartridge 3 has a fault in time period 3, and this fault may affect the performance of other filter cartridges.

[0044] S104, based on the distribution association relationship of the first characteristic data matrix set, performing data enhancement processing on the first characteristic data matrix set to obtain second characteristic data matrix sets corresponding to different filter cartridge failures.

[0045] In one embodiment of the present application, the first feature data matrix set is normalized to adjust the scale of the feature data. Multiple feature data matrices corresponding to the same fault filter cartridge are processed by feature polynomial combination to generate new features. The distribution association relationship of the first feature data matrix set and the new feature input are preset into a GANs model to generate a first new data sample. The distribution association relationship of the first feature data matrix set and the new feature input are preset into a VAEs model to generate a second new data sample. Based on the new features, the first new data sample and the second new data sample, data enhancement processing is performed on the first feature data matrix set.

[0046] Specifically, the first feature data matrix set is normalized, and each feature value is adjusted to a specific range. For multiple feature data matrices corresponding to the same faulty filter cartridge, two features (such as gas concentration and temperature) can be selected for polynomial combination processing. For example, a new feature "gas concentration*temperature" is generated. This new feature combines different features to better reflect the working status and fault conditions of the filter cartridge.

[0047] Furthermore, the normalized first feature data matrix set distribution association relationship and the new features are input into the preset GANs model. The GANs generator generates first new data samples based on these input information, which are similar to the original data in feature distribution, but have new details and changes. Similarly, this information is input into the preset VAEs model, and a second new data sample is generated through a decoding process. Based on the new features, the first new data sample and the second new data sample, the original first feature data matrix set can be data enhanced. For example, these new data samples can be mixed with the original data, or combined in some way (such as data fusion, feature splicing, etc.) to form a richer data set, so that the enhanced data set can be used to train a more accurate model to identify the failure mode of the filter cartridge.

[0048] The embodiments of the present application can improve the quality and richness of data through data preprocessing and enhancement steps, thereby improving the performance and generalization ability of the model.

[0049] S105. Construct a fault point detection model based on the second characteristic data matrix set, so as to perform fault point detection on the multi-cartridge filter based on the fault point detection model and the currently acquired gas output end data.

[0050] In one embodiment of the present application, a preset neural network model is trained based on a second feature data matrix set. Through a global self-attention mechanism, the derivative vector groups corresponding to each second feature data matrix set are determined, and the derivative vector groups are multiplied and summed to obtain the global most relevant information corresponding to each second feature data matrix set, so as to adjust the preset neural network model based on the global most relevant information. The weight of the adjusted preset neural network model is updated through a back propagation algorithm. When the training results meet the requirements, a fault point detection model is obtained.

[0051] Specifically, the embodiment of the present application trains a preset neural network model through the second feature data matrix set obtained after data enhancement. The preset neural network model in the embodiment of the present application can be a convolutional neural network.

[0052] Furthermore, an embodiment of the present application introduces a global self-attention mechanism to determine the most relevant information corresponding to each second feature data matrix set. The global self-attention mechanism is a mechanism that can capture the relationship between all elements in a data set. It can assign weights according to the correlation between elements to highlight the most important information. For each second feature data matrix set, it is converted into a derivative vector group. Then, the global self-attention mechanism is used to calculate the correlation between these vectors and obtain a weighted sum, which is the global most relevant information. Assuming that the second feature data matrix set contains gas concentration data at multiple time points, these data are converted into derivative vector groups, and the global self-attention mechanism is used to calculate the correlation between these vectors. Through calculation, a weighted sum is obtained, which represents the global most relevant information of gas concentration at all time points.

[0053] Furthermore, the globally most relevant information is used to adjust the parameters of the preset neural network model. This can be achieved by using the globally most relevant information as an additional input or as a component of the loss function. By adjusting the model parameters, the model can be made to pay more attention to the globally most relevant information, thereby improving the performance of the model. For example, when training a CNN model, the globally most relevant information can be added as an additional input feature to the input layer of the model, or as a component of the loss function to optimize the model. In this way, the model will pay more attention to the globally most relevant information during the training process, thereby improving the detection capability of the fault point.

[0054] Furthermore, the back propagation algorithm is used to update the weights of the preset neural network model. The back propagation algorithm is an optimization algorithm for training neural networks, which updates the weights by calculating the gradient of the loss function with respect to the model parameters. When training the CNN model, the back propagation algorithm is used to calculate the gradient of the loss function with respect to the model weights, and the weights are updated according to these gradients. Through multiple iterations of training, the loss value of the model can be gradually reduced, thereby improving the performance of the model. If the performance of the model meets the predetermined standards (for example, indicators such as accuracy and recall rate reach a certain threshold), it can be considered that the training process has been completed and a fault point detection model is obtained.

[0055] In one embodiment of the present application, the currently acquired gas output end data is input into the first time series detection model to obtain the time feature sequence of the multi-filter cartridge filter to be tested in the current time period, and the frequency domain feature is extracted from the currently acquired gas output end data, and the frequency domain feature is input into the second time series prediction model to obtain the frequency domain feature sequence corresponding to the multi-filter cartridge filter to be tested in the current time period. The gas composition data corresponding to the air inlet end of the multi-filter cartridge filter to be tested in the current time period is obtained, and the warning threshold is determined based on the gas composition data. The total number of time features in the time feature sequence and the frequency domain features in the frequency domain feature sequence is obtained, and when the ratio between the number of abnormal features and the total number is greater than the warning threshold, a warning message is generated.

[0056] Specifically, the first time series detection model in the embodiment of the present application can be an LSTM model. The currently acquired gas output end data is input into the LSTM model, and the model outputs a time feature sequence, in which each feature represents the state change of the filter at a certain point in time, such as flow fluctuation, pressure change rate, etc.

[0057] Furthermore, the gas output data is transformed by FFT to extract frequency components, such as fundamental frequency, harmonics, etc. Then, these frequency domain features are input into another time series prediction model, such as ARIMA model or prediction model based on deep learning, and the model outputs a frequency domain feature sequence, which contains the changes of different frequency components over time.

[0058] Furthermore, the gas composition data corresponding to the air inlet end of the multi-cartridge filter to be tested in the current time period is obtained. Based on these gas composition data, an early warning threshold is determined for subsequent judgment on whether the number of abnormal characteristics reaches the early warning condition. Assume that the monitored gas components include oxygen content. According to historical data and experience, when the oxygen content is lower than a certain threshold (such as lower than 20%), the filter may face the risk of clogging or failure. Therefore, this oxygen content threshold is used as the early warning threshold.

[0059] Further, the total number of time features in the time feature sequence and the total number of frequency domain features in the frequency domain feature sequence are obtained. Then, the number of abnormal features (i.e., the number of features that are beyond the normal range or expected) is calculated. If the ratio between the number of abnormal features and the total number is greater than the warning threshold, a warning message is generated, indicating that the filter may have problems or need maintenance.

[0060] For example, 100 time features and 100 frequency domain features are obtained, a total of 200 features. After analysis, it is found that 30 of the features are beyond the normal range (such as abnormal flow fluctuations, abnormal frequency components, etc.). Therefore, the number of abnormal features is 30. If the set warning threshold is 10%, then since 30 / 200=15%, which is greater than the warning threshold, it is considered that the filter may be abnormal and a warning message is generated.

[0061] In one embodiment of the present application, the currently acquired gas output end data is input into the fault point detection model, so as to output multiple reference fault filter cartridge information based on the fault point detection model, and perform a fault alarm based on the priority of the multiple reference fault filter cartridge information.

[0062] Specifically, after obtaining the current gas output data, it is input into the fault point detection model, which will output multiple reference fault filter cartridge information based on the input gas output data. The importance of each fault point is determined based on the probability, degree, impact or urgency of the fault, so as to issue a fault alarm according to priority.

[0063] Figure 2 This is a schematic diagram of a structure of an analysis and judgment device for monitoring filter failure points provided in an embodiment of the present application. Figure 2As shown, the analysis and judgment device 200 for monitoring the failure point of the filter includes: at least one processor 201; and a memory 202 which is in communication with the at least one processor 201; wherein the memory 202 stores instructions which can be executed by the at least one processor 201, and the instructions are executed by the at least one processor 201 so that the at least one processor 201 can: obtain the gas output end data samples corresponding to the case where the multi-filter cartridge filter to be tested has failures in different filter cartridges; wherein the gas output end is evenly spaced and provided with a plurality of gas detection devices; the plurality of gas detection devices correspond to detecting different filter cartridge areas respectively; and the gas output end data samples are processed by the at least one processor 201; The method extracts feature data, and constructs first feature data matrix sets corresponding to different filter cartridge failures based on the extracted feature data; inputs the first feature data matrix set into a neural network diagram, and obtains distribution correlation relationships of the first feature data matrix sets corresponding to different filter cartridge failures through the neural network diagram; based on the distribution correlation relationship of the first feature data matrix set, performs data enhancement processing on the first feature data matrix set to obtain second feature data matrix sets corresponding to different filter cartridge failures; and constructs a fault point detection model based on the second feature data matrix set, so as to perform fault point detection on a multi-filter cartridge filter based on the fault point detection model and the currently acquired gas output end data.

[0064] A non-volatile computer storage medium provided in an embodiment of the present application stores computer executable instructions, wherein the computer executable instructions are configured to: obtain gas output end data samples corresponding to different filter cartridge failures of a multi-filter cartridge filter to be tested; wherein a plurality of gas detection devices are evenly spaced at the gas output end; and the plurality of gas detection devices correspond to detecting different filter cartridge areas, respectively; extract characteristic data from the gas output end data samples, and construct first characteristic data matrix sets corresponding to different filter cartridge failures based on the extracted characteristic data; input the first characteristic data matrix set into a neural network graph, and obtain distribution association relationships of the first characteristic data matrix sets corresponding to different filter cartridge failures through the neural network graph; perform data enhancement processing on the first characteristic data matrix set based on the distribution association relationship of the first characteristic data matrix set, and obtain second characteristic data matrix sets corresponding to different filter cartridge failures; and construct a fault point detection model based on the second characteristic data matrix set, so as to perform fault point detection on the multi-filter cartridge filter based on the fault point detection model and the gas output end data currently obtained.

[0065] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0066] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the embodiments of the present application may have various changes and variations. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application.

Claims

1. A method for analyzing and determining the failure point of a monitoring filter, characterized in that: The method comprises: Obtaining data samples of the gas output ends corresponding to the multi-filter cartridge filter to be tested when different filter cartridges fail; wherein the gas output ends are evenly spaced with multiple gas detection devices; the multiple gas detection devices correspond to detecting different filter cartridge areas respectively; Extracting characteristic data from the gas output end data sample, and constructing first characteristic data matrix sets corresponding to different filter cartridge failures based on the extracted characteristic data; Inputting the first characteristic data matrix set into a neural network graph, and obtaining distribution association relationships of the first characteristic data matrix sets corresponding to different filter cartridge failures through the neural network graph; Based on the distribution association relationship of the first characteristic data matrix set, the first characteristic data matrix set is subjected to data enhancement processing to obtain second characteristic data matrix sets corresponding to different filter cartridge failures respectively; A fault point detection model is constructed based on the second characteristic data matrix set, so as to perform fault point detection on the multi-cartridge filter based on the fault point detection model and the currently acquired gas output end data.

2. The method for analyzing and determining the failure point of a monitoring filter according to claim 1 is characterized in that: The extracting characteristic data from the gas output end data sample and constructing first characteristic data matrix sets corresponding to different filter cartridge failures based on the extracted characteristic data specifically includes: Performing time domain feature extraction on the gas output end data sample; and Extracting gas concentration features from the gas output end data sample by Hilbert transform; Based on the feature type, the time domain feature and the gas concentration feature are divided; wherein the time domain feature includes at least one of the peak time, the root mean square value and the waveform factor; the gas concentration feature includes at least the instantaneous amplitude extraction value and the gas concentration frequency extraction value; When the current filter cartridge to be tested fails, different types of features corresponding to the multiple gas detection devices are determined, and based on the different types of features, multiple feature data matrices are constructed to obtain the first feature data matrix set corresponding to the current failed filter cartridge to be tested.

3. The method for analyzing and determining the failure point of a monitoring filter according to claim 1 is characterized in that: The step of inputting the first characteristic data matrix set into a neural network diagram and obtaining the distribution association relationship of the first characteristic data matrix sets corresponding to different filter cartridge failures through the neural network diagram specifically includes: Inputting the first feature data matrix set corresponding to the first time period into the graph neural network; In the graph neural network, the gas concentration difference between the key points of each pre-set filter cartridge corresponding to the first time period is determined; wherein the key points of the pre-set filter cartridge are evenly arranged in the graph neural network, and the positions of the key points of the pre-set filter cartridge are related to the actual positions of each filter cartridge; Inputting a first feature data matrix set corresponding to a next time period into the graph neural network; In the graph neural network, the gas concentration difference between the key points of each preset filter cartridge corresponding to the next time period is determined; Until the first feature data matrix set is input into the graph neural network in the order of the extraction time of the gas output end data samples, and the gas concentration difference values ​​corresponding to multiple time periods are obtained; The key point relationship score is determined based on the gas concentration difference, and the first characteristic data matrix set distribution association relationship corresponding to different filter cartridge failures is determined based on the changes in the gas concentration differences corresponding to the multiple time periods and the key point relationship score.

4. A method for analyzing and determining failure points of monitoring filters according to claim 3, characterized in that: The determining of the distribution association relationship of the first characteristic data matrix sets corresponding to different filter cartridge failures based on the changes in the gas concentration differences corresponding to the multiple time periods and the key point relationship scores specifically includes: Determining the node distances between the key points of each of the preset filter cartridges, and obtaining a first node score based on the node distances; Determining a second node score according to the difference in gas concentration between the key points of each of the pre-installed filter cartridges; Obtaining the key point relationship score based on the first node score and the second node score; Based on the gas concentration differences corresponding to the multiple time periods, construct a gas concentration change curve between the key points of each pre-set filter cartridge; and Based on the key point relationship scores corresponding to the multiple time periods, a key point relationship score change curve between the key points of each preset filter cartridge is constructed; Based on the gas concentration variation curve and the relationship score variation curve, the distribution association relationship of the first characteristic data matrix sets corresponding to different filter cartridge failures is determined.

5. The method for analyzing and determining the failure point of a monitoring filter according to claim 1 is characterized in that: The performing data enhancement processing on the first feature data matrix set based on the distribution association relationship of the first feature data matrix set specifically includes: Normalizing the first feature data matrix set to adjust the scale of the feature data; Perform characteristic polynomial combination processing on multiple characteristic data matrices corresponding to the same faulty filter cartridge to generate new characteristics; Input the first feature data matrix set distribution association relationship and the new feature into a pre-set GANs model to generate a first new data sample; The distribution association relationship of the first feature data matrix set and the new feature input are pre-set to train the VAEs model to generate a second new data sample; Based on the new features, the first new data samples and the second new data samples, data enhancement processing is performed on the first feature data matrix set.

6. The method for analyzing and determining the failure point of a monitoring filter according to claim 1 is characterized in that: The constructing of the fault point detection model based on the second feature data matrix set specifically includes: Training a preset neural network model based on the second feature data matrix set; Determine the derivative vector groups corresponding to each of the second feature data matrix sets through a global self-attention mechanism, perform multiplication and summation processing on the derivative vector groups, and obtain the global most relevant information corresponding to each of the second feature data matrix sets, so as to adjust the preset neural network model based on the global most relevant information; Updating the weights of the adjusted preset neural network model through a back propagation algorithm; When the training result meets the requirements, the fault point detection model is obtained.

7. The method for analyzing and determining the failure point of a monitoring filter according to claim 1 is characterized in that: Before performing fault point detection on the multi-cartridge filter based on the fault point detection model and the currently acquired gas output end data, the method further includes: Inputting the currently acquired gas output end data into a first time series detection model to obtain a time characteristic sequence of the multi-cartridge filter to be tested within the current time period; and Extracting frequency domain features from the currently acquired gas output end data, inputting the frequency domain features into a second time series prediction model, and obtaining a frequency domain feature sequence corresponding to the multi-cartridge filter to be tested in the current time period; Acquire gas composition data corresponding to the air inlet end of the multi-filter cartridge filter to be tested in the current time period, and determine a warning threshold based on the gas composition data; The total number of time features in the time feature sequence and the total number of frequency domain features in the frequency domain feature sequence is obtained, and when the ratio between the number of abnormal features and the total number is greater than the warning threshold, warning information is generated.

8. The method for analyzing and determining the failure point of a monitoring filter according to claim 1 is characterized in that: The method of performing fault point detection on the multi-cartridge filter based on the fault point detection model and the currently acquired gas output end data specifically includes: Inputting the currently acquired gas output end data into the fault point detection model to output a plurality of reference fault filter cartridge information based on the fault point detection model; Based on the priorities of the multiple reference fault filter cartridge information, a fault alarm is issued.

9. An analysis and judgment device for monitoring filter failure points, characterized in that: The device comprises a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to execute the method according to any one of claims 1 to 8.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions can execute the method according to any one of claims 1 to 8.

Images