Bag-type dust collector fault detection method and device, storage medium and computer equipment

Abstract

According to the bag-type dust collector fault detection method and device, the storage medium and the computer equipment provided by the invention, the detection data of the bag-type dust collector are firstly obtained, on one hand, the bag breaking detection model can perform feature extraction on the detection data based on the convolutional network which is good at feature extraction to obtain the feature sequence; and then a nonlinear relation in the feature sequence is learned through a multi-layer feedforward network, so that the bag breaking position and the bag breaking probability are accurately predicted. Therefore, the maintenance response can be triggered at the initial stage of failure, and the energy consumption of the dust remover is reduced. On the other hand, the ash blockage diagnosis model can respectively capture long-term time sequence characteristics and short-term time sequence characteristics in the detection data, the long-term time sequence characteristics can identify the mode and trend of ash blockage, the short-term time sequence characteristics can identify the initial sign of ash blockage, and then the long-term time sequence characteristics and the short-term time sequence characteristics are combined to predict the ash blockage degree; the perspectiveness and the accuracy of fault detection are improved, and the efficient operation of the bag-type dust collector is ensured.

Description

Technical Field

[0001] This application relates to the field of deep learning technology, and in particular to a method, apparatus, storage medium and computer equipment for fault detection of bag filters. Background Technology

[0002] In waste incineration, "bag filters" refers to baghouse dust collectors, a type of dry dust filtration device primarily used to capture fine, dry, non-fibrous dust. The working principle of a baghouse dust collector is to filter dust-laden gas using the filtration effect of fibrous fabric. When the dust-laden gas enters the baghouse dust collector, larger, heavier dust particles settle due to gravity and fall into the ash hopper, while the gas containing finer dust particles is purified by trapping the dust as it passes through the filter media.

[0003] In the actual application of baghouse dust collectors, if the baghouse dust collector has faults such as damaged filter bags, ineffective dust removal, or poor sealing, and these faults cannot be detected in a timely and accurate manner, the hidden faults will continue to accumulate, eventually leading to increased energy consumption and decreased processing efficiency of the baghouse dust collector system. Summary of the Invention

[0004] The purpose of this application is to at least solve one of the above-mentioned technical defects, especially the technical defect in the prior art where if a bag filter dust collector has problems such as filter bag damage, dust removal failure, or poor sealing, and these problems cannot be detected in a timely and accurate manner, the hidden operational faults will continue to accumulate, eventually leading to increased energy consumption and decreased processing efficiency of the bag filter dust collector system.

[0005] In a first aspect, this application provides a method for detecting faults in a baghouse dust collector, the method comprising:

[0006] Obtain test data from baghouse dust collectors;

[0007] The detection data is input into a pre-optimized bag breakage detection model to obtain the bag breakage location and probability of the bag filter. The bag breakage detection model extracts features from the detection data based on a convolutional network to obtain a feature sequence, and learns the nonlinear relationship in the feature sequence through a multi-layer feedforward network to predict the bag breakage location and probability.

[0008] The detection data is input into a pre-optimized dust blockage diagnosis model to obtain the degree of dust blockage of the bag filter. The dust blockage diagnosis model predicts the degree of dust blockage by capturing the long-term and short-term time-series features in the detection data respectively and combining the long-term and short-term time-series features.

[0009] The alarm level is determined based on the location of the broken bag, the probability of the broken bag, and the degree of ash blockage, and a detection report is generated based on the location of the broken bag, the probability of the broken bag, the degree of ash blockage, and the alarm level.

[0010] In one embodiment, acquiring the detection data of the bag filter includes:

[0011] Based on the various sensors installed in the bag filter, multidimensional sensing data of the bag filter is collected.

[0012] The multidimensional sensing data is cleaned and normalized to obtain the target sensing data;

[0013] The operating parameters of the bag filter are obtained, and the detection data of the bag filter is generated based on the target sensor data and the operating parameters.

[0014] In one embodiment, the bag breakage detection model includes a first convolutional network, a second convolutional network, and a multilayer feedforward network. The step of inputting the detection data into the pre-optimized bag breakage detection model to obtain the bag breakage location and probability of the bag filter dust collector includes:

[0015] Based on the first convolutional network, spatial features and local dependencies in the detection data are extracted to form a feature sequence;

[0016] Based on the second convolutional network, feature information in the detection data is further extracted to obtain deep features, and feature vectors for prediction are determined based on the feature sequence and the deep features.

[0017] The feature vector is nonlinearly transformed based on the multilayer feedforward network to predict the location and probability of bag breakage in the bag filter.

[0018] In one embodiment, the dust blockage diagnosis model includes a first time-series network, a second time-series network, and a classification network. The step of inputting the detection data into the pre-optimized dust blockage diagnosis model to obtain the dust blockage degree of the bag filter includes:

[0019] Based on the first temporal network, long-term dependencies in the detection data are captured to obtain long-term temporal features;

[0020] Based on the second temporal network, short-term dependencies in the detection data are captured to obtain short-term temporal features. The long-term temporal features and the short-term temporal features are then fused together to output comprehensive temporal features.

[0021] The severity of the comprehensive time-series features is assessed using the classification network to obtain the degree of dust clogging in the bag filter.

[0022] In one embodiment, fusing the long-term time series features and the short-term time series features to output a comprehensive time series feature includes:

[0023] Determine the learnable parameters of the ash blockage diagnostic model;

[0024] Obtain long-term feature weights and short-term feature weights from the learnable parameters;

[0025] The long-term and short-term time-series features are weighted and fused according to the long-term and short-term feature weights to obtain the comprehensive time-series features.

[0026] In one embodiment, the method further includes:

[0027] When scheduling maintenance tasks based on the aforementioned inspection report, the actual maintenance results are determined.

[0028] The actual inspection results are compared with the inspection report to generate a difference report. Based on the difference report, the parameters of the bag breakage detection model and the ash blockage diagnosis model are fine-tuned to update the bag breakage detection model and the ash blockage diagnosis model.

[0029] In one embodiment, after obtaining the degree of ash blockage in the bag filter, the method further includes:

[0030] Determine the preset solution library;

[0031] Search the solution library for a cleaning solution that matches the degree of ash blockage, and obtain a cleaning solution that matches the degree of ash blockage.

[0032] A dust removal command is generated according to the dust removal scheme, and the dust removal command is triggered.

[0033] Secondly, this application provides a fault detection device for a bag filter dust collector, the device comprising:

[0034] The data acquisition module is used to acquire the detection data of the bag filter dust collector;

[0035] The bag breakage detection module is used to input the detection data into a pre-optimized bag breakage detection model to obtain the bag breakage location and probability of the bag dust collector. The bag breakage detection model extracts features from the detection data based on a convolutional network to obtain a feature sequence, and learns the nonlinear relationship in the feature sequence through a multi-layer feedforward network to predict the bag breakage location and probability.

[0036] The dust blockage diagnosis module is used to input the detection data into a pre-optimized dust blockage diagnosis model to obtain the degree of dust blockage of the bag filter. The dust blockage diagnosis model predicts the degree of dust blockage by capturing the long-term time-series features and short-term time-series features in the detection data respectively, and combining the long-term time-series features and the short-term time-series features.

[0037] The report generation module is used to determine the alarm level based on the location of the broken bag, the probability of the broken bag, and the degree of ash blockage, and to generate a detection report based on the location of the broken bag, the probability of the broken bag, the degree of ash blockage, and the alarm level.

[0038] Thirdly, this application provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the bag filter fault detection method as described in any of the above embodiments.

[0039] Fourthly, this application provides a computer device, including: one or more processors, and a memory;

[0040] The memory stores computer-readable instructions, and when the one or more processors execute the computer-readable instructions, they perform the steps of the bag filter fault detection method as described in any of the above embodiments.

[0041] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0042] The bag filter dust collector fault detection method, apparatus, storage medium, and computer equipment provided in this application first acquire the detection data of the bag filter dust collector. On one hand, the detection data is input into a pre-optimized bag breakage detection model. This model can extract features from the detection data based on a convolutional network, which excels at feature extraction, to obtain a feature sequence. Then, it further learns the nonlinear relationship in the feature sequence through a multi-layer feedforward network, thereby accurately predicting the location and probability of bag breakage. This can trigger maintenance response in the early stage of fault occurrence, effectively curbing problems caused by filter bag damage and improving the dust collector's processing efficiency. On the other hand, the detection data is input into a pre-optimized ash blockage diagnosis model. This ash blockage diagnosis model can capture long-term and short-term time-series features from the detection data, respectively. The long-term time-series features can identify the pattern and trend of ash blockage, while the short-term time-series features can identify the initial signs of ash blockage. Then, the long-term and short-term time-series features are combined to predict the degree of ash blockage, improving the foresight and accuracy of fault detection. Finally, the alarm level is determined from two dimensions: bag breakage status and ash blockage degree, to generate a detection report. This enables timely and accurate early warning signals, ensuring the efficient operation of the bag filter. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A flowchart illustrating a method for detecting faults in a baghouse dust collector, provided in an embodiment of this application;

[0045] Figure 2 A flowchart illustrating the construction and deployment of a bag breakage detection model provided in this application embodiment;

[0046] Figure 3 A flowchart illustrating the construction and deployment of a gray blockage diagnostic model provided in this application embodiment;

[0047] Figure 4 An example architecture diagram of an online detection system provided in this application embodiment;

[0048] Figure 5 This is a schematic diagram of the structure of a bag filter fault detection device provided in an embodiment of this application;

[0049] Figure 6 This is an internal structural diagram of a computer device provided in an embodiment of this application. Detailed Implementation

[0050] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0051] In one embodiment, this application provides a method for detecting faults in a baghouse dust collector. The following embodiments illustrate the application of this method to a server. It is understood that the baghouse dust collector fault detection method can be executed by a single server or by a server cluster consisting of multiple servers, and this application does not impose any specific limitations on this.

[0052] like Figure 1 As shown, this application provides a method for fault detection in a baghouse dust collector, the method comprising:

[0053] S101: Obtain the test data of the bag filter.

[0054] The detection data includes sensor data and operating parameters of the bag filter dust collector. The sensor data can include data on dimensions such as pressure, flow rate, temperature, and vibration.

[0055] In this step, when fault detection of a bag filter is required, the bag filter to be tested is first identified. Then, sensor data of the bag filter is acquired based on the sensors installed in it. Next, the recent operating parameters of the bag filter are obtained based on a preset monitoring system or historical monitoring records. These operating parameters include, but are not limited to, filtration velocity, cleaning pressure, equipment resistance, and air leakage rate. After determining the sensor data and operating parameters of the bag filter, they are combined to obtain the detection data of the bag filter.

[0056] S102: Input the detection data into the pre-optimized bag breakage detection model to obtain the bag breakage location and probability of the bag filter.

[0057] Among them, the location of the broken bag refers to the specific location in the bag filter where the damage has occurred, which can be represented by numbered zones.

[0058] In this step, the detection data is input into a pre-optimized bag breakage detection model, enabling the model to extract features from the detection data using a convolutional network, resulting in a feature sequence. Using a convolutional network for feature extraction efficiently captures features from the detection data, reducing data dimensionality. Then, a multi-layer feedforward network is used to learn the nonlinear relationships in the feature sequence, predicting the bag breakage location and probability. This allows for flexible learning of complex nonlinear relationships between features. By combining these two networks, their advantages are fully utilized, achieving efficient feature extraction and learning of complex nonlinear relationships, thereby reducing the possibility of false positives and false negatives. Furthermore, since the detection data contains multi-dimensional sensor data, multi-dimensional information can be fused during bag breakage detection, resulting in more accurate detection. It is understood that the multi-layer feedforward network includes multiple hidden layers, and nonlinearity can be introduced through activation functions to learn complex nonlinear relationships in the input data.

[0059] Specifically, the optimization process of the bag breakage detection model can include: constructing a model architecture containing convolutional networks and multi-layer feedforward networks to obtain a pre-trained model; using historical detection data of bag filters as training samples; using the bag breakage location and probability corresponding to the training samples as sample labels; and iteratively training the pre-trained model using a loss function. During iterative training, the server can input the training samples and their corresponding sample labels into the pre-trained model for forward propagation, and use the loss function to adjust the parameters of the pre-trained model during backpropagation until the pre-trained model meets certain iterative conditions. The trained pre-trained model can then be used as the bag breakage detection model. After obtaining the bag breakage detection model, its accuracy and generalization ability can be verified and evaluated in terms of precision, recall, and F1-score, allowing for fine-tuning based on the verification and evaluation results.

[0060] In one embodiment, the convolutional network can be a CNN neural network, and the multilayer feedforward network can be a BP (Back Propagation) neural network.

[0061] S103: Input the detection data into the pre-optimized dust blockage diagnosis model to obtain the degree of dust blockage of the bag filter.

[0062] Among them, the degree of dust blockage is used to quantitatively assess the thickness of the dust layer accumulated on the surface of the filter bags inside the baghouse dust collector or the severity of the blockage.

[0063] In this step, the detection data is input into a pre-optimized clogging diagnosis model. This model captures both long-term and short-term time-series features from the detection data and combines these features to predict the degree of clogging. Capturing long-term time-series features helps identify patterns and trends in clogging and analyzes dynamic changes during the clogging process. Capturing short-term time-series features helps identify initial signs of clogging and analyzes short-term and sudden changes in the data. Therefore, predicting the degree of clogging using both long-term and short-term time-series features allows for a comprehensive consideration of both aspects during clogging diagnosis, improving accuracy and predictability. Furthermore, long-term time-series features refer to patterns and trends spanning a relatively long time span, typically reflecting long-term dependencies in the data. Short-term time-series features refer to patterns and changes exhibited within a shorter time span, typically reflecting instantaneous changes, local fluctuations, or short-term trends in the data.

[0064] Specifically, the optimization process of the dust blockage diagnosis model may include: constructing a pre-trained model, using historical detection data of the bag filter as training samples, using the degree of dust blockage corresponding to the training samples as sample labels, and iteratively training the pre-trained model using a loss function. During the iterative training process, the server can input the training samples and their corresponding sample labels into the pre-trained model for forward propagation, and use the loss function to adjust the parameters of the pre-trained model during backpropagation until the pre-trained model meets certain iterative conditions. The completed pre-trained model can then be used as the dust blockage diagnosis model.

[0065] S104: Determine the alarm level based on the location of the broken bag, the probability of the broken bag, and the degree of ash blockage, and generate a detection report based on the location of the broken bag, the probability of the broken bag, the degree of ash blockage, and the alarm level.

[0066] The alarm level refers to the degree of danger classified according to information such as the location of bag breakage, the probability of bag breakage, and the degree of dust blockage. It can be used to assess the severity of the condition of a bag filter.

[0067] In this step, after determining the location of the bag breakage, the probability of bag breakage, and the degree of dust blockage, the impact of the location of the bag breakage, the probability of bag breakage, and the degree of dust blockage on the operating status of the bag filter is analyzed. Then, the alarm level is determined based on this impact. Finally, an inspection report of the bag filter is generated based on the location of the bag breakage, the probability of bag breakage, the degree of dust blockage, and the alarm level. At this time, the corresponding level of alarm can be triggered and the inspection report can be sent so that relevant personnel can quickly understand the details of the fault based on the inspection report when they receive the alarm.

[0068] Specifically, when determining the alarm level based on the location of bag breakage, the probability of bag breakage, and the degree of dust blockage, a pre-built mapping rule can be used to determine the alarm level corresponding to the location of bag breakage, the probability of bag breakage, and the degree of dust blockage. Alternatively, based on preset evaluation indicators, the impact of the location of bag breakage, the probability of bag breakage, and the degree of dust blockage on the operating status of the bag filter can be evaluated, and then the corresponding alarm level can be determined based on this impact on the operating status. This application does not impose specific limitations in this regard.

[0069] In the above embodiments, the detection data of the bag filter is first acquired. On one hand, the detection data is input into a pre-optimized bag breakage detection model. This model can extract features from the detection data using a convolutional network, which excels at feature extraction, to obtain a feature sequence. Then, a multi-layer feedforward network is used to learn the nonlinear relationship in the feature sequence, thereby accurately predicting the location and probability of bag breakage. This can trigger a maintenance response in the early stages of a fault, effectively curbing problems caused by filter bag damage and improving the dust collector's processing efficiency. On the other hand, the detection data is input into a pre-optimized ash blockage diagnosis model. This model can capture long-term and short-term time-series features from the detection data. Long-term time-series features can identify the patterns and trends of ash blockage, while short-term time-series features can identify the initial signs of ash blockage. Then, the long-term and short-term time-series features are combined to predict the degree of ash blockage, improving the foresight and accuracy of fault detection. Finally, the alarm level is determined from two dimensions: bag breakage status and ash blockage degree, to generate a detection report. This enables timely and accurate early warning signals to ensure the efficient operation of the bag filter.

[0070] In one embodiment, acquiring the detection data of the bag filter includes:

[0071] S1: Collect multi-dimensional sensing data of the bag filter based on the various sensors installed in the bag filter.

[0072] S2: Perform data cleaning and normalization on the multidimensional sensing data to obtain the target sensing data.

[0073] S3: Obtain the operating parameters of the bag filter and generate the detection data of the bag filter based on the target sensor data and operating parameters.

[0074] In this embodiment, various sensors, such as pressure sensors, temperature sensors, vibration sensors, and flow sensors, are installed in the bag filter. These sensors collect multidimensional sensing data from the bag filter. This multidimensional sensing data is then cleaned, including removing outliers, duplicates, or missing data, and converting the data format to ensure data quality. The cleaned multidimensional sensing data is then normalized to obtain target sensing data, ensuring that data from different sensors can be compared on the same scale and avoiding the negative impact of different dimensions. Finally, the operating parameters of the bag filter are obtained, and these operating parameters and the obtained target sensing data are summarized to obtain the current detection data of the bag filter.

[0075] In one embodiment, the bag breakage detection model includes a first convolutional network, a second convolutional network, and a multilayer feedforward network. Detection data is input into the pre-optimized bag breakage detection model to obtain the bag breakage location and probability of the bag filter, including:

[0076] S1: Extract spatial features and local dependencies from the detection data based on the first convolutional network to form a feature sequence.

[0077] S2: Based on the second convolutional network, further extract feature information from the detection data to obtain deep features, and determine the feature vector for prediction based on the feature sequence and deep features.

[0078] S3: Based on a multi-layer feedforward network, a nonlinear transformation is performed on the feature vector to predict the location and probability of bag breakage in the bag filter.

[0079] The system comprises three layers: a first convolutional network for initial feature extraction, a second convolutional network for deep feature extraction, and a multi-layer feedforward network for analyzing nonlinear relationships in the data. Spatial features refer to features in the detection data that are related to spatial location. Local dependencies refer to the relationships between adjacent or nearby regions in the detection data.

[0080] In this embodiment, firstly, spatial features and local dependencies in the detection data are extracted based on a first convolutional network to form a feature sequence. This process uses convolutional and pooling layers in the first convolutional network to process the detection data, capturing local features and spatial structures. Subsequently, a second convolutional network further extracts feature information from the detection data to obtain deep features. Building upon the first convolutional network, the second convolutional network uses more convolutional and pooling layers to extract higher-level feature information; these deep features can capture complex patterns and structures in the data. After determining the feature sequence and deep features, a fully connected layer combines the shallow and deep features to form a comprehensive feature vector for subsequent fault prediction tasks. Finally, a multi-layer feedforward network is used to perform a nonlinear transformation on the feature vector to predict the bag breakage location and probability of the bag filter. This staged feature extraction and prediction method not only effectively captures multi-level features in the data but also improves the model's generalization ability and prediction accuracy, helping to promptly detect and address bag breakage problems and improve the operational stability and reliability of the dust collector.

[0081] In one embodiment, such as Figure 2 As shown, Figure 2 This is a flowchart illustrating the construction and deployment of a bag-breaking detection model, as provided in an embodiment of this application. Figure 2In this context, data segmentation refers to operations such as dividing the training set and validation set. Convolutional neural networks include a first convolutional network and a second convolutional network. Coupled BP neural networks are multi-layer feedforward networks.

[0082] In one embodiment, the dust blockage diagnosis model includes a first time-series network, a second time-series network, and a classification network. Detection data is input into the pre-optimized dust blockage diagnosis model to obtain the degree of dust blockage in the bag filter, including:

[0083] S1: Based on the first temporal network, long-term dependencies in the detection data are captured to obtain long-term temporal features.

[0084] S2: Based on the second temporal network, capture the short-term dependencies in the detection data to obtain short-term temporal features, and fuse long-term and short-term temporal features to output comprehensive temporal features.

[0085] S3: Use a classification network to assess the severity of the comprehensive time-series characteristics to obtain the degree of dust blockage in the bag filter.

[0086] The system comprises two temporal networks: a first temporal network for extracting short-term dependencies and a second temporal network for extracting long-term dependencies. A classification network is used to classify the gray-blocking level based on the comprehensive temporal features, outputting the corresponding degree of gray-blocking. The comprehensive temporal features refer to the feature set that integrates long-term and short-term temporal features, comprehensively reflecting the dynamic changes and dependencies of the detection data across different time scales.

[0087] In this embodiment, long-term temporal features are obtained by capturing long-term dependencies in the detection data using a first temporal network. This first temporal network focuses on patterns and trends spanning a long time span, thus extracting features reflecting long-term dynamics. Next, short-term temporal features are obtained by capturing short-term dependencies in the detection data using a second temporal network. This stage of the network focuses on immediate changes and local fluctuations in the data, extracting features reflecting short-term dynamics. Then, the long-term and short-term temporal features are fused to form a comprehensive temporal feature. This fusion process can be achieved in various ways, such as feature concatenation or weighted summation, combining features from two different time scales to more comprehensively describe the dynamic characteristics of the data. Finally, a classification network is used to assess the severity of the comprehensive temporal features to obtain the degree of dust clogging in the baghouse dust collector. The classification network can learn the mapping relationship between the comprehensive temporal features and the degree of dust clogging to output the severity assessment result of the dust clogging, i.e., the degree of dust clogging.

[0088] Specifically, by capturing long-term dependencies through a first-time-series network, the clogging diagnostic model can identify the gradual accumulation of clogging that may occur in a baghouse dust collector during long-term operation. The second-time-series network, on the other hand, can capture sudden clogging events that may occur in the short term. By fusing these two features, the clogging diagnostic model can more comprehensively assess the degree of clogging, thereby ensuring the stable and efficient operation of the baghouse dust collector.

[0089] In one embodiment, the first temporal network can be an LSTM (Long Short-Term Memory) network, the second temporal network can be an RNN (Recurrent Neural Network), and the classification network can be a deep belief network. Deep belief networks can learn complex features from unlabeled data, reducing reliance on feature engineering. The specific choice can be based on the function and characteristics of different networks, and this application does not impose specific limitations in this regard.

[0090] In one embodiment, such as Figure 3 As shown, Figure 3 This is a flowchart illustrating the construction and deployment of a gray blockage diagnostic model, as provided in an embodiment of this application. Figure 3 In this context, LSTM stands for First Temporal Network, RNN stands for Second Temporal Network, and DBN stands for Classification Network.

[0091] In one embodiment, long-term and short-term time-series features are fused to output comprehensive time-series features, including:

[0092] S1: Determine the learnable parameters of the ash blockage diagnostic model.

[0093] S2: Obtain long-term feature weights and short-term feature weights from learnable parameters.

[0094] S3: Weighted fusion of long-term and short-term time series features is performed based on the weights of long-term and short-term features to obtain comprehensive time series features.

[0095] Learnable parameters refer to the parameters that the model automatically adjusts through optimization algorithms during training.

[0096] In this embodiment, learnable parameters of the gray blockage diagnosis model are obtained. Then, long-term feature weights and short-term feature weights are extracted from these parameters. Finally, based on the long-term and short-term feature weights, the long-term and short-term time-series features are weighted and fused to obtain comprehensive time-series features. In this process, the long-term and short-term feature weights are used as learnable parameters of the gray blockage diagnosis model. This allows the model to automatically adjust its focus on long-term and short-term time-series features according to the characteristics of the data, which is then reflected in the long-term and short-term feature weights. This obtains appropriate long-term and short-term feature weights, improving the accuracy of subsequent gray blockage diagnosis.

[0097] In some embodiments, long-term feature weights and short-term feature weights can be pre-set based on historical experience and adjusted periodically. When fusing long-term and short-term time-series features, the pre-set long-term and short-term feature weights are directly obtained to perform weighted fusion of the long-term and short-term time-series features. For example, if there is prior knowledge about the gray blockage judgment, such as believing that short-term fluctuations have a greater impact on the current gray blockage state, higher weights can be given to short-term time-series features; if the long-term trend is more critical (e.g., gray blockage is caused by slow accumulation), higher weights can be given to long-term time-series features.

[0098] In one embodiment, the bag filter fault detection method further includes:

[0099] S1: When scheduling maintenance tasks based on inspection reports, determine the actual maintenance results.

[0100] S2: Compare the actual inspection results with the inspection report to generate a difference report, and fine-tune the parameters of the bag breakage detection model and the ash blockage diagnosis model based on the difference report to update the bag breakage detection model and the ash blockage diagnosis model.

[0101] The actual maintenance results include the actual number of broken filters and the actual degree of ash blockage. Maintenance tasks include dust removal, filter bag repair or replacement, and airflow adjustment, among others.

[0102] In this embodiment, a feedback optimization mechanism can be set up. After the corresponding maintenance task is arranged based on the test report, the actual maintenance result is obtained. Then, the actual maintenance result is compared with the test report to determine the difference report reflecting the difference between the actual maintenance result and the test report. Finally, the parameters of the bag breakage detection model and the ash blockage diagnosis model can be fine-tuned according to the difference report to further optimize the bag breakage detection model and the ash blockage diagnosis model.

[0103] In one embodiment, after determining the degree of dust blockage in the bag filter, the bag filter fault detection method further includes:

[0104] S1: Determine the preset scheme library.

[0105] S2: Search the solution library for a cleaning solution that matches the degree of ash blockage, and obtain the cleaning solution that matches the degree of ash blockage.

[0106] S3: Generate a dust removal command based on the dust removal plan and trigger the dust removal command.

[0107] The solution library is used to store the mapping relationship between the degree of ash blockage and the ash removal solution.

[0108] In this embodiment, a cleaning solution matching the degree of ash blockage is queried from the solution library, and then the cleaning solution is obtained. Based on this solution, the cleaning operation is performed. For example, if the ash blockage is mild, the cleaning solution might be "suggest increasing the cleaning frequency or adjusting the cleaning method, such as appropriate reverse airflow." If the ash blockage is moderate, the cleaning solution might be "suggesting increasing the reverse airflow intensity or adjusting the airflow." If the ash blockage is severe, the cleaning solution might be "requiring manual inspection combined with powerful cleaning methods, such as cleaning agents, vibration cleaning, etc." By promptly recommending suitable cleaning solutions, time for manual judgment and decision-making can be saved, and the most suitable cleaning operation can be quickly determined, ensuring the efficient operation of the dust collector.

[0109] Specifically, the effectiveness of dust removal can be evaluated based on historical dust removal experience, thereby establishing a mapping relationship between the degree of dust blockage and the dust removal scheme, and recording it in the scheme library.

[0110] In one embodiment, such as Figure 4 As shown, Figure 4 This diagram illustrates the architecture of an online detection system provided in an embodiment of this application. The system can execute the baghouse dust collector fault detection method provided in this application. The system can be divided into a hardware layer, a data transmission layer, a data processing layer, a storage layer, an analysis and decision-making layer, and a user interface layer. Pre-optimized bag breakage detection models and dust blockage diagnosis models can be deployed in the analysis and decision-making layer.

[0111] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0112] The following describes the bag filter fault detection device provided in the embodiments of this application. The bag filter fault detection device described below can be referred to in correspondence with the bag filter fault detection method described above.

[0113] like Figure 5 As shown, this application provides a bag filter dust collector fault detection device 200, the device comprising:

[0114] Data acquisition module 201 is used to acquire the detection data of the bag filter dust collector;

[0115] The bag breakage detection module 202 is used to input the detection data into the pre-optimized bag breakage detection model to obtain the bag breakage location and probability of the bag dust collector. The bag breakage detection model extracts features from the detection data based on a convolutional network to obtain a feature sequence, and learns the nonlinear relationship in the feature sequence through a multi-layer feedforward network to predict the bag breakage location and probability.

[0116] The dust blockage diagnosis module 203 is used to input the detection data into the pre-optimized dust blockage diagnosis model to obtain the degree of dust blockage of the bag filter. The dust blockage diagnosis model predicts the degree of dust blockage by capturing the long-term time series features and short-term time series features in the detection data respectively and combining the long-term time series features and short-term time series features.

[0117] The report generation module 204 is used to determine the alarm level based on the location of the broken bag, the probability of the broken bag, and the degree of ash blockage, and to generate a detection report based on the location of the broken bag, the probability of the broken bag, the degree of ash blockage, and the alarm level.

[0118] In the above embodiments, the detection data of the bag filter is first acquired. On one hand, the detection data is input into a pre-optimized bag breakage detection model. This model can extract features from the detection data using a convolutional network, which excels at feature extraction, to obtain a feature sequence. Then, a multi-layer feedforward network is used to learn the nonlinear relationship in the feature sequence, thereby accurately predicting the location and probability of bag breakage. This can trigger a maintenance response in the early stages of a fault, effectively curbing problems caused by filter bag damage and improving the dust collector's processing efficiency. On the other hand, the detection data is input into a pre-optimized ash blockage diagnosis model. This model can capture long-term and short-term time-series features from the detection data. Long-term time-series features can identify the patterns and trends of ash blockage, while short-term time-series features can identify the initial signs of ash blockage. Then, the long-term and short-term time-series features are combined to predict the degree of ash blockage, improving the foresight and accuracy of fault detection. Finally, the alarm level is determined from two dimensions: bag breakage status and ash blockage degree, to generate a detection report. This enables timely and accurate early warning signals to ensure the efficient operation of the bag filter.

[0119] In one embodiment, the data acquisition module includes:

[0120] The data acquisition submodule is used to collect multi-dimensional sensing data of the bag filter based on the various sensors installed in the bag filter.

[0121] The preprocessing submodule is used to perform data cleaning and normalization on multidimensional sensor data to obtain target sensor data.

[0122] The data determination submodule is used to acquire the operating parameters of the bag filter and generate the detection data of the bag filter based on the target sensor data and the operating parameters.

[0123] In one embodiment, the bag breakage detection model includes a first convolutional network, a second convolutional network, and a multi-layer feedforward network, and the bag breakage detection module includes:

[0124] The first extraction submodule is used to extract spatial features and local dependencies in the detection data based on the first convolutional network to form a feature sequence;

[0125] The second extraction submodule is used to further extract feature information from the detection data based on the second convolutional network to obtain deep features, and to determine the feature vector for prediction based on the feature sequence and deep features.

[0126] The prediction submodule is used to perform nonlinear transformation on the feature vector based on a multi-layer feedforward network to predict the location and probability of bag breakage in the bag filter.

[0127] In one embodiment, the gray blockage diagnosis model includes a first time-series network, a second time-series network, and a classification network, and the gray blockage diagnosis module includes:

[0128] The first capture submodule is used to capture long-term dependencies in the detection data based on the first temporal network to obtain long-term temporal features;

[0129] The second capture submodule is used to capture short-term dependencies in the detection data based on the second temporal network, obtain short-term temporal features, and fuse long-term and short-term temporal features to output comprehensive temporal features;

[0130] The evaluation submodule is used to assess the severity of comprehensive time-series characteristics using a classification network to obtain the degree of dust blockage in the bag filter.

[0131] In one embodiment, the second capture submodule includes:

[0132] The parameter determination unit is used to determine the learnable parameters of the ash blockage diagnostic model;

[0133] The weight acquisition unit is used to obtain long-term feature weights and short-term feature weights from learnable parameters;

[0134] The feature fusion unit is used to perform weighted fusion of long-term and short-term time-series features based on long-term and short-term feature weights to obtain comprehensive time-series features.

[0135] In one embodiment, the bag filter fault detection device further includes:

[0136] The result determination module is used to determine the actual maintenance result when maintenance tasks are arranged based on the inspection report;

[0137] The model update module is used to compare the actual maintenance results with the inspection report, generate a difference report, and fine-tune the parameters of the bag breakage detection model and the ash blockage diagnosis model based on the difference report to update the bag breakage detection model and the ash blockage diagnosis model.

[0138] In one embodiment, the bag filter fault detection device further includes:

[0139] The solution library determination module is used to determine the preset solution library;

[0140] The solution acquisition module is used to query the solution library for cleaning solutions that match the degree of ash blockage, and to acquire cleaning solutions that match the degree of ash blockage.

[0141] The scheme execution module is used to generate and trigger cleaning instructions based on the cleaning scheme.

[0142] The division of modules in the above-described bag filter fault detection device is merely illustrative. In other embodiments, the bag filter fault detection device can be divided into different modules as needed to complete all or part of its functions. Each module in the above-described bag filter fault detection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0143] In one embodiment, this application also provides a storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the bag filter fault detection method as described in any of the above embodiments.

[0144] In one embodiment, this application also provides a computer device storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the bag filter fault detection method as described in any of the above embodiments.

[0145] Indicatively, such as Figure 6 As shown, Figure 6 This is a schematic diagram of the internal structure of a computer device 300 provided in an embodiment of this application. The computer device 300 can be provided as a server. (Refer to...) Figure 6 The computer device 300 includes a processing component 302, which further includes one or more processors, and memory resources represented by memory 301 for storing instructions, such as application programs, that can be executed by the processing component 302. The application programs stored in memory 301 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 302 is configured to execute instructions to perform the baghouse dust collector fault detection method of any of the above embodiments.

[0146] The computer device 300 may also include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 may operate on an operating system stored in memory 301, such as Windows Server™, Mac OS X™, Unix™, Linux™, Free BSD™, or similar.

[0147] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0148] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a…" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. In this document, the singular forms "a," "an," and "the" may also include the plural forms unless the context clearly indicates otherwise. It should also be understood that the terms “comprising / including” or “having” specify the presence of the stated features, wholes, steps, operations, components, parts or combinations thereof, but do not exclude the possibility of the presence or addition of one or more other features, wholes, steps, operations, components, parts or combinations thereof. Meanwhile, the term “and / or” as used in this specification includes any and all combinations of the associated listed items.

[0149] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0150] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting a fault in a baghouse, the method comprising: The method comprises: obtaining detection data of a bag filter; inputting the detection data into a pre-optimized bag breakage detection model to obtain a bag breakage position and a bag breakage probability of the bag filter, the bag breakage detection model performing feature extraction on the detection data based on a convolution network to obtain a feature sequence, and learning a nonlinear relationship in the feature sequence through a multi-layer feedforward network to predict the bag breakage position and the bag breakage probability; inputting the detection data into a pre-optimized ash plugging diagnosis model to obtain an ash plugging degree of the bag filter, the ash plugging diagnosis model capturing long-term time sequence features and short-term time sequence features in the detection data respectively, and combining the long-term time sequence features and the short-term time sequence features to predict the ash plugging degree; determining an alarm level according to the bag breakage position, the bag breakage probability and the ash plugging degree, and generating a detection report according to the bag breakage position, the bag breakage probability, the ash plugging degree and the alarm level.

2. The fabric filter failure detection method according to claim 1, characterized in that, The obtaining of the detection data of the bag filter comprises: collecting multi-dimensional sensing data of the bag filter based on various sensors arranged in the bag filter; performing data cleaning and normalization processing on the multi-dimensional sensing data to obtain target sensing data; obtaining operating parameters of the bag filter, and generating the detection data of the bag filter according to the target sensing data and the operating parameters.

3. The fabric filter bag failure detection method of claim 1, wherein, The bag breakage detection model comprises a first convolution network, a second convolution network and a multi-layer feedforward network, and the inputting of the detection data into the pre-optimized bag breakage detection model to obtain the bag breakage position and the bag breakage probability comprises: extracting spatial features and local dependency relationships in the detection data based on the first convolution network to form a feature sequence; further extracting feature information in the detection data according to the second convolution network to obtain deep features, and determining a feature vector for prediction according to the feature sequence and the deep features; performing nonlinear transformation on the feature vector based on the multi-layer feedforward network to predict the bag breakage position and the bag breakage probability of the bag filter.

4. The fabric filter bag failure detection method of claim 1, wherein The ash plugging diagnosis model comprises a first time sequence network, a second time sequence network and a classification network, and the inputting of the detection data into the pre-optimized ash plugging diagnosis model to obtain the ash plugging degree of the bag filter comprises: capturing long-term dependency relationships in the detection data based on the first time sequence network to obtain long-term time sequence features; capturing short-term dependency relationships in the detection data based on the second time sequence network to obtain short-term time sequence features, and fusing the long-term time sequence features and the short-term time sequence features to output comprehensive time sequence features; performing severity evaluation on the comprehensive time sequence features by using the classification network to obtain the ash plugging degree of the bag filter.

5. The fabric filter bag failure detection method of claim 4, wherein, The fusing of the long-term time sequence features and the short-term time sequence features to output comprehensive time sequence features comprises: determining learnable parameters of the ash plugging diagnosis model; obtaining long-term feature weights and short-term feature weights from the learnable parameters; The long-term time sequence feature and the short-term time sequence feature are fused by weighting according to the long-term feature weight and the short-term feature weight, to obtain a comprehensive time sequence feature.

6. The fabric filter bag failure detection method of claim 1, wherein, The method further comprises: When arranging a maintenance task based on the detection report, determining an actual maintenance result; Comparing the actual maintenance result with the detection report to generate a difference report, and performing parameter fine-tuning on the broken bag detection model and the ash plugging diagnosis model according to the difference report, to update the broken bag detection model and the ash plugging diagnosis model.

7. The baghouse fault detection method of any of claims 1-6, wherein, After obtaining the ash plugging degree of the bag filter, the method further comprises: Determining a preset scheme library; Querying a dust cleaning scheme matching the ash plugging degree in the scheme library, and obtaining the dust cleaning scheme matching the ash plugging degree; Generating a dust cleaning instruction according to the dust cleaning scheme, and triggering the dust cleaning instruction.

8. A baghouse fault detection apparatus, comprising: The device comprises: A data acquisition module configured to acquire detection data of a bag filter; A broken bag detection module configured to input the detection data into a pre-optimized broken bag detection model to obtain a broken bag position and a broken bag probability of the bag filter, wherein the broken bag detection model is based on a convolution network to extract features from the detection data to obtain a feature sequence, and learns a non-linear relationship in the feature sequence through a multi-layer feedforward network to predict the broken bag position and the broken bag probability; An ash plugging diagnosis module configured to input the detection data into a pre-optimized ash plugging diagnosis model to obtain an ash plugging degree of the bag filter, wherein the ash plugging diagnosis model captures long-term time sequence features and short-term time sequence features in the detection data respectively, and combines the long-term time sequence features and the short-term time sequence features to predict the ash plugging degree; A report generation module configured to determine an alarm level according to the broken bag position, the broken bag probability, and the ash plugging degree, and generate a detection report according to the broken bag position, the broken bag probability, the ash plugging degree, and the alarm level.

9. A storage medium characterized by: The storage medium has computer readable instructions stored therein, and the computer readable instructions, when executed by one or more processors, cause the one or more processors to perform the steps of the bag filter fault detection method according to any one of claims 1 to 7.

10. A computer device, comprising: Comprise: One or more processors, and a memory; The memory has computer readable instructions stored therein, and the computer readable instructions, when executed by the one or more processors, perform the steps of the bag filter fault detection method according to any one of claims 1 to 7.

Images