A machine learning-based industrial equipment fault diagnosis method, device and medium

Abstract

This application discloses a machine learning-based method, device, and medium for industrial equipment fault diagnosis. The method includes: collecting operational data of the target equipment and preprocessing the data to obtain monitoring data; determining a pre-trained fault diagnosis model, which is an improved graph neural network model, with the input layer of the improved graph neural network model employing a fuzzy inference mechanism; and inputting the monitoring data into the fault diagnosis model to determine the equipment status and fault type of the target equipment. By using a graph neural network, the topological relationships between samples can be fully explored and utilized, making it particularly suitable for processing complex data from multiple sensors. It can capture the interdependencies between nodes, thereby improving the accuracy of diagnosis. The fuzzy inference mechanism employed in the input layer of the graph neural network is used to handle the uncertainty in the data, and can be adjusted according to the fuzziness of the data, improving the ability to handle uncertain data.

Description

Technical Field

[0001] This application relates to the field of machine learning, specifically to a machine learning-based method, equipment, and medium for diagnosing industrial equipment faults. Background Technology

[0002] In recent years, with the widespread application of industrial equipment, equipment fault diagnosis has become an important part of ensuring production safety, improving production efficiency, and reducing maintenance costs.

[0003] Traditional equipment fault diagnosis methods often rely on human experience or rule-based diagnostic models. These methods tend to suffer from low diagnostic accuracy, slow response, and poor adaptability to data patterns when dealing with complex, nonlinear, and multivariate equipment data. Especially when processing large amounts of real-time sensor data, traditional methods fail to fully exploit the correlations between various sensors and the uncertainties within the data, resulting in limited fault detection effectiveness. Summary of the Invention

[0004] To address the aforementioned problems, this application proposes a machine learning-based method, equipment, and medium for industrial equipment fault diagnosis, wherein the method includes:

[0005] The operation data of the target equipment is collected and preprocessed to obtain monitoring data; a pre-trained fault diagnosis model is determined, wherein the fault diagnosis model is an improved graph neural network model and the input layer of the improved graph neural network model adopts a fuzzy inference mechanism; the monitoring data is input into the fault diagnosis model to determine the equipment status and fault type of the target equipment.

[0006] In one example, the process of collecting operational data from the target device and preprocessing the operational data to obtain monitoring data specifically includes: storing the operational data in JSON format; cleaning the operational data to obtain first intermediate data; normalizing the first intermediate data to obtain second intermediate data; and discretizing the second intermediate data to obtain the monitoring data.

[0007] In one example, before determining the pre-trained fault diagnosis model, the method further includes: acquiring historical operating data of the target device and labeling the pre-processed historical operating data, the labeling content including at least the device status and fault type; inputting the labeled historical operating data into an initial fault diagnosis model, and training the initial fault diagnosis model by setting a dynamic damping coefficient to obtain the fault diagnosis model; determining the device information corresponding to the target device, and storing the fault diagnosis model based on the device information; the device information includes at least one of device name, device type, device location, and device parameters.

[0008] In one example, training the initial fault diagnosis model by setting a dynamic damping coefficient specifically includes: determining the initial damping coefficient, the decay rate, the current iteration of model training, and the loss change rate corresponding to the current iteration; and determining the damping coefficient corresponding to the current iteration based on the initial damping coefficient, the decay rate, the current iteration of model training, and the loss change rate corresponding to the current iteration.

[0009] In one example, inputting the monitoring data into the fault diagnosis model to determine the device status and fault type of the target device specifically includes: in the input layer of the fault diagnosis model, determining the target data with fuzzy labels corresponding to the target node; determining the fuzzy confidence factor corresponding to the target data based on the feature distribution dispersion of the target data; determining the input features corresponding to the target node based on a preset fuzzy inference function, the original fuzzy features of the target node, and the fuzzy confidence factor; and determining the device status and fault type of the target device based on the input features.

[0010] In one example, inputting the monitoring data into the fault diagnosis model to determine the device status and fault type of the target device specifically includes: determining the target node within the target layer and the adjacent nodes of the target node; obtaining a first initial feature representation of the target node, a second initial feature representation of the adjacent nodes, and an adjacency matrix between the target node and the adjacent nodes; and fusing the first initial feature representation and the second initial feature representation based on the adjacency matrix to obtain a fused feature representation of the target node.

[0011] In one example, after determining the device status and fault type of the target device, the method further includes: when the device status of the target device is a fault state, obtaining the fault type of the target device; and based on the fault type, triggering a corresponding instruction for the target device, wherein the corresponding instruction includes at least one of a maintenance instruction or a shutdown instruction.

[0012] In one example, each piece of operational data includes at least a timestamp, device number, data collection location, sensor type, and sensor measurement value; the nodes in the fault diagnosis model represent the data features corresponding to a piece of operational data of the target device, the edges connecting each node represent the similarity between the pieces of operational data, and the data features corresponding to the nodes are updated through graph convolutional layers.

[0013] This application also provides a machine learning-based industrial equipment fault diagnosis device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform: collecting operating data of a target device and preprocessing the operating data to obtain monitoring data; determining a pre-trained fault diagnosis model, the fault diagnosis model being an improved graph neural network model, the input layer of the improved graph neural network model employing a fuzzy inference mechanism; and inputting the monitoring data into the fault diagnosis model to determine the equipment status and fault type of the target device.

[0014] This application also provides a non-volatile computer storage medium storing computer-executable instructions, characterized in that the computer-executable instructions are configured to: collect operating data of a target device and preprocess the operating data to obtain monitoring data; determine a pre-trained fault diagnosis model, wherein the fault diagnosis model is an improved graph neural network model and the input layer of the improved graph neural network model adopts a fuzzy inference mechanism; and input the monitoring data into the fault diagnosis model to determine the device status and fault type of the target device.

[0015] The method proposed in this application can bring the following beneficial effects:

[0016] 1. Compared to traditional machine learning methods that are usually based on independent samples, this application uses graph neural networks to fully explore and utilize the topological relationships between samples. It is particularly suitable for processing complex data from multi-source sensors and can capture the interdependencies between nodes, thereby improving the accuracy of diagnosis.

[0017] 2. A fuzzy inference mechanism is employed in the input layer of the graph neural network to handle uncertainty in the data. Traditional machine learning methods typically assume that the data is deterministic, while real-world industrial data often exhibits fuzziness and uncertainty. The fuzzy inference mechanism can adjust according to the fuzziness of the data, improving its ability to handle uncertain data.

[0018] 3. By setting a mechanism to dynamically adjust the damping coefficient, the model can automatically adjust its learning strategy according to the needs of different stages during training, avoiding overfitting and improving training efficiency. This mechanism is different from static training methods, allowing the model to focus on learning the graph structure in the early stage and gradually enhance regularization in the later stage to avoid overfitting.

[0019] 4. A multi-level feature fusion module is adopted, which fuses local and global features through graph convolution operations, enabling the model to capture multi-dimensional information about the device status at different levels, thereby improving the model's ability to understand and classify data diversity. Attached Figure Description

[0020] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0021] Figure 1 This is a flowchart illustrating a machine learning-based industrial equipment fault diagnosis method in an embodiment of this application.

[0022] Figure 2 This is a schematic diagram of the structure of an industrial equipment fault diagnosis device based on machine learning, as described in an embodiment of this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0025] Figure 1 This diagram illustrates a machine learning-based fault diagnosis method for industrial equipment, provided for one or more embodiments of this specification. The method can be applied to different types of industrial equipment. The process can be executed by computing devices in the relevant field (e.g., cloud servers, risk control servers, or smart mobile terminals used to control industrial equipment). Certain input parameters or intermediate results in the process can be manually adjusted to help improve accuracy.

[0026] The analysis method involved in the embodiments of this application can be implemented by a terminal device or a server, and this application does not impose any special limitations on it. For ease of understanding and description, the following embodiments are all described in detail using a server as an example.

[0027] It should be noted that the server can be a single device or a system composed of multiple devices, i.e., a distributed server. This application does not make any specific limitations in this regard.

[0028] like Figure 1 As shown in the figure, this application provides a machine learning-based industrial equipment fault diagnosis method, including:

[0029] S101: Collect the operating data of the target device and preprocess the operating data to obtain monitoring data.

[0030] Operational data of the target equipment is collected through sensors and monitoring devices. Data collection can be direct or indirect, specifically by configuring specific sensors such as temperature, pressure, and acceleration sensors to collect relevant physical parameters in real time. Sensors can be installed on equipment, objects, or in the environment, transmitting data to a data acquisition center via wireless or wired communication. The operational data is stored in JSON format for subsequent processing and analysis. Each data entry contains multiple fields, primarily including a timestamp, device ID, acquisition location, sensor type, sensor measurement value (e.g., temperature, pressure), and other supplementary information. This supplementary information refers to optional data types such as measurement point, equipment model, and equipment lifespan.

[0031] After obtaining the operational data, it needs to be preprocessed to obtain monitoring data that can be input into the model.

[0032] In one embodiment, during preprocessing, the running data needs to be stored in JSON format; then, the running data is cleaned to obtain first intermediate data; the first intermediate data is normalized to obtain second intermediate data; finally, the second intermediate data is discretized to obtain monitoring data. Here, both the first and second intermediate data are intermediate data, and they are used to distinguish the preprocessing steps.

[0033] Specifically, during data cleaning, since the original data may contain some noise and outliers, the goal of data cleaning is to ensure the accuracy and consistency of the data by removing outliers, filling in missing values, and correcting formatting errors. For each piece of collected data, an outlier detection mechanism based on statistical methods, such as standard deviation-based methods, is used to exclude data points that exceed the normal range; missing values ​​are filled in using interpolation methods.

[0034] When performing data normalization, all collected numerical data are normalized to meet the training requirements of machine learning models. Since the measurement units of different sensors may differ, to avoid the negative impact of inconsistent data scales, Min-Max normalization or Z-score normalization methods are used to map all measurement data to a uniform scale interval. For example, the data is mapped to the [0,1] interval, thus ensuring that different types of data have the same impact.

[0035] Since the collected data is generally continuous numerical data, it needs to be discretized to meet the requirements of the classification task. A discretization method based on the K-means clustering algorithm can be used. First, the data is grouped according to specific features (such as time, spatial location, measurement value, etc.), and the data values ​​within each group are replaced with the mean or median of that group, thus obtaining discretized features. Finally, the data after the above processing steps will have the following discretized properties:

[0036] Ra represents the timestamp (normalized value), Da represents the sensor ID (discreteized category), Ea represents the device status (discreteized status category), Fa represents the measurement location (e.g., the area or component number where the sensor is located, discretized category), Ga represents the temperature value (discreteized value), Ha represents the pressure value (discreteized value), Ia represents the vibration value (discreteized value), Ja represents the humidity value (discreteized value), Ka represents the device load (discreteized value), and La represents the operating frequency (discreteized value). It should be noted that this embodiment is only to illustrate one data format and type of the present invention. In practical applications, the data usually has more than 10 attributes, and the number of attributes may reach dozens or even hundreds.

[0037] S102: Determine the pre-trained fault diagnosis model.

[0038] After obtaining the monitoring data, it is necessary to determine the fault diagnosis model corresponding to the target equipment. The fault diagnosis model here is an improved graph neural network model, and the input layer of the graph neural network model adopts a fuzzy inference mechanism.

[0039] In one embodiment, different types of industrial equipment have different fault diagnosis model parameters. When training the fault diagnosis model parameters, it is first necessary to acquire the historical operating data of the target equipment and label the preprocessed historical operating data. The labeling content includes at least the equipment status and fault type. The equipment status can include normal or faulty states, and the fault type can include fuse blowout, short circuit, etc. Then, the labeled historical operating data needs to be input into the initial fault diagnosis model, and the initial fault diagnosis model is trained by setting a dynamic damping coefficient to obtain the fault diagnosis model. Simultaneously, after the fault diagnosis model is trained, it can be stored. When performing fault diagnosis on the target equipment again, the pre-stored fault diagnosis model can be directly retrieved and used, thus avoiding retraining the model. At this point, it is necessary to determine the equipment information corresponding to the target equipment and store the fault diagnosis model based on the equipment information; the equipment information includes at least one of the following: equipment name, equipment type, equipment location, and equipment parameters.

[0040] Furthermore, when training the initial fault diagnosis model by setting a dynamic damping coefficient, it is necessary to determine the initial damping coefficient, the decay rate, the current iteration of model training, and the loss change rate corresponding to the current iteration. Based on the initial damping coefficient, the decay rate, the current iteration of model training, and the loss change rate corresponding to the current iteration, the damping coefficient corresponding to the current iteration is determined.

[0041] The loss function of the fault diagnosis model can be expressed as:

[0042]

[0043] Where L(Θ,λ(t)) is the total loss function of the fault diagnosis model, and Θ is the set of all parameters in the fault diagnosis model (including the weight matrix W and the bias term b); Let be the final feature representation of node a at layer T; Sof() is the Softmax classifier function used to map node features to categories; y a Let be the true label of node a; LC() be the loss function (e.g., cross-entropy loss function); λ(t) be a dynamically adjusted damping coefficient that changes as training progresses; and |||| be the L2 norm. Compared with existing technologies, the fault diagnosis model provided in this application adds more elements to its loss function. This part can be used to dynamically set the damping coefficient.

[0044] Furthermore, the dynamically adjusted damping coefficient changes with the number of iterations. Through dynamic adjustment, the model can focus on learning the graph structure in the early stages, and gradually strengthen regularization in the later stages to avoid overfitting, as expressed in:

[0045]

[0046] In the formula, λ0 is the initial damping coefficient; tanh() is the hyperbolic tangent function; α is the decay rate; t is the current iteration of training; λ(t) is the damping coefficient in the t-th iteration; and Δ(t) is the rate of change of loss. Preferably, λ0 is set to 0.2.

[0047] Furthermore, the calculation method for the rate of change of loss is expressed as follows:

[0048]

[0049] In the formula, L(t) is the loss function of the graph neural network in the t-th iteration; L(t-1) is the loss function of the graph neural network in the (t-1)-th iteration.

[0050] By setting a mechanism to dynamically adjust the damping coefficient, the model can automatically adjust its learning strategy according to the needs of different stages during training, avoiding overfitting and improving training efficiency. This mechanism differs from static training methods, allowing the model to focus on learning the graph structure in the early stages and gradually enhance regularization in the later stages to avoid overfitting.

[0051] In one embodiment, when determining the fault diagnosis model corresponding to the target device, a fault diagnosis model previously used by the target device can be used as the fault diagnosis model for this fault diagnosis task. Alternatively, based on information such as the target device's device name, device type, device location, and device parameters, the similarity between each pre-stored fault diagnosis model and the target device can be determined in the database. Based on the similarity score, the device with the highest similarity to the target device is determined, and its pre-stored fault diagnosis model is used as the fault diagnosis model for the target device.

[0052] S103: Input the monitoring data into the fault diagnosis model to determine the equipment status and fault type of the target equipment.

[0053] After obtaining the monitoring data and determining the corresponding fault diagnosis model for the target device, the monitoring data can be input into the fault diagnosis model to determine the device status and fault type of the target device. The device status and fault type here are consistent with those mentioned earlier; the device status can include normal or faulty states, and the fault type can include faults such as blown fuses or short circuits.

[0054] After model training is completed, the preprocessed real-time monitoring data is input into the trained graph neural network model for fault diagnosis. The model first models the uncertainty of the input features through the fuzzy inference module, optimizes the information transmission weights between nodes by combining the dynamic damping adjustment mechanism, and uses graph convolution operations to aggregate the topological relationships of multi-source sensor data (such as discrete features of temperature, pressure, vibration, etc.) of the equipment layer by layer to generate equipment status representation. The model outputs the fault probability distribution at the node level and uses a classifier to determine the current status (normal / abnormal) of the equipment and the specific fault type (such as overload, sensor failure, etc.).

[0055] In one embodiment, a fuzzy inference mechanism is integrated into the input layer of the fault diagnosis model to effectively model the fuzziness of the data from the initial stage. During fuzzy inference, the target data with fuzzy labels corresponding to the target nodes needs to be identified first in the input layer of the fault diagnosis model. Then, based on the feature distribution dispersion of the target data, the fuzzy confidence factor corresponding to the target data is determined. Next, based on a preset fuzzy inference function, the original fuzzy features of the target nodes, and the fuzzy confidence factor, the input features corresponding to the target nodes are determined. Finally, the input features obtained from fuzzy inference can be input into the fault diagnosis wear resistance system to determine the equipment status and fault type of the target device.

[0056] Here is an explanation of fuzzy labels: For example, the operating data of the target equipment contains operating data used to characterize the intensity of equipment vibration. However, whether the intensity of vibration is severe is subject to the subjective judgment of the staff. Even if it is characterized by 0 representing low intensity of vibration, 1 representing medium intensity of vibration, and 2 representing high intensity of vibration, the operating data will still contain too much human judgment. In this case, the operating data is considered to have fuzzy labels.

[0057] The fuzzy inference process is explained in detail below: The basic operation of a graph neural network is to update the representation of each node through graph convolution. The graph convolution operation of the fault diagnosis model can be represented as follows:

[0058]

[0059] In the formula, Let be the feature representation of node a at layer t; Let be the feature representation of node a at layer t+1; Let A be the set of neighboring nodes of node a; ab is an element in the adjacency matrix, representing the connection strength between node a and node b; W is the weight matrix of the t-th graph convolutional layer, representing the weighted sum of information from each neighboring node; b c is the bias term for the graph convolutional layer; Sig() is the Sigmoid activation function.

[0060] To address the uncertainty in the data, the fuzzy inference module uses fuzzy rules for reasoning and adjustment, deriving fuzzy inference results and using them as the initial input for nodes. Unlike traditional static fuzzy rules, the fuzzy inference rules, which are adaptive to data distribution, make the inference results more closely match the characteristics of the data; specifically, the fuzzy characteristics of node a are represented as follows: A new feature representation x′ can be obtained by calculating using fuzzy inference rules. a The fuzzy inference module fine-tunes the node features, enhancing the graph neural network's ability to handle fuzzy features. The inference process is represented as follows:

[0061]

[0062] In the formula, denoted as the original fuzzy feature of node a, representing the uncertain or fuzzy part of the data; Fe() is the fuzzy inference function; θ is the parameter of the fuzzy inference function.

[0063] Furthermore, the fusion ratio is controlled by a fuzzy confidence factor to fuse the original features and fuzzy inference results, utilizing the fuzzy confidence factor γ. a ∈[0,1] measures the level of certainty of node features, and is expressed as:

[0064]

[0065] In the formula, The fused feature representation serves as the input to the graph neural network during training. Furthermore, the fuzzy confidence factor is calculated from the feature distribution dispersion, expressed as follows:

[0066]

[0067] In the formula, x a For the original features of node a, σ(x) a ) represents the standard deviation of the features of node a, which can be obtained by calculating the standard deviation of the feature sequence values; τ is the temperature coefficient. Preferably, τ can be set to 0.3.

[0068] In one example, the fuzzy inference function can be calculated as follows:

[0069]

[0070] In the formula, μ i (x a ) = soft(W μ x a ) represents the activation weight of the i-th fuzzy rule; Soft() is the Softmax classifier function, which maps node features to categories; d is the membership projection matrix; k is the preset number of rules; d is the number of features; It is a learnable rule base matrix.

[0071] Here's a further explanation of the learnable rule base matrix: Suppose we have a device's vibration data as input features in a device fault diagnosis task. Assume the device's vibration signal exhibits different levels of ambiguity, potentially categorized as "minor fault," "moderate fault," or "severe fault." However, the boundaries between these labels are not very clear. Fuzzy inference can capture this ambiguity. The role of the rule base matrix in the fuzzy inference module is to assume three fuzzy rules: Rule 1: When the vibration signal is minor, the update ratio of node features is low; Rule 2: When the vibration signal is moderate, the update ratio of node features is moderate; Rule 3: When the vibration signal is severe, the update ratio of node features is high.

[0072] These rules are represented by a learnable rule base matrix, for example, the rule base matrix R might look like this:

[0073]

[0074] In this matrix, each column represents a rule, and the elements of the column represent the weight of that rule on different feature dimensions. During training, the graph neural network automatically adjusts the weights in matrix R so that fuzzy inference can better adapt to different vibration data characteristics. For example, in the case of "minor faults," the network may automatically adjust the rule base matrix, resulting in a lower update ratio for node features; in the case of "severe faults," the network may increase the weight of the rule, making the update of node features more obvious. The fuzzy inference mechanism can adjust according to the fuzziness of the data, improving the ability to handle uncertain data.

[0075] In one embodiment, in each round of training, the fault diagnosis wear resistance not only learns the direct relationships between nodes, but also extracts more complex and abstract features through a multi-level feature fusion module. This multi-level feature fusion module combines local node information with global information through a hierarchical graph convolution fusion strategy, enhancing the graph neural network's ability to understand data diversity. Specifically, it is necessary to determine the target node within the target layer and its adjacent nodes; then, obtain the first initial feature representation of the target node, the second initial feature representation of the adjacent nodes, and the adjacency matrix between the target node and its adjacent nodes. Based on the adjacency matrix, the first initial feature representation and the second initial feature representation can be fused to form the fused feature representation of the target node.

[0076] Assume that the node features of node a calculated at layer e are: This serves as the initial feature representation. The strategy of fusing local and global features allows the model to capture both local and global information at different levels, thereby improving the expressive power of the classifier, as shown below:

[0077]

[0078] In the formula, The feature fusion operation can be vector concatenation, addition, or weighted averaging. This represents the second feature of neighbor node b at layer e. In the fault diagnosis model, a node represents the data feature corresponding to a piece of operational data of the target device, and the edges connecting each node represent the similarity between the pieces of operational data. The data feature corresponding to the node is updated through a graph convolutional layer.

[0079] After several iterations, the representation vector of each node in the graph neural network is finally passed to the classifier layer for decision-making. The Softmax classifier function compares the feature vector of each node with the model centers of each category to assign a class label to each sample. By combining the learning results of the graph neural network with the adjustment of fuzzy inference, the classifier can achieve higher classification accuracy in complex data environments. The classification decision process is represented as follows:

[0080]

[0081] In the formula, This is the predicted class label for node a. `arg max()` is the function to retrieve the maximum output value.

[0082] In one embodiment, during the training of the fault diagnosis model, the learning rate can be adjusted in each round of training based on the current classifier error. This allows the model to adjust its training strategy at different stages, thereby improving training efficiency and classification accuracy, as expressed below:

[0083]

[0084] In the formula, ζ is the sensitivity coefficient; L(i) is the loss function of the i-th iteration of the graph neural network; L(i-1) is the loss function of the (i-1)-th iteration of the graph neural network; η(t) is the learning rate of the graph neural network in the t-th iteration; η0 is the initial learning rate of the graph neural network; β is the learning rate decay rate; E(t) is the cumulative error fluctuation; and t is the current iteration number of the training. Preferably, ζ is set to 0.01, β is set to 2, and η0 is set to 0.01.

[0085] Furthermore, based on the learning rate of the graph neural network in the t-th iteration, gradient descent is used for error backpropagation to achieve iterative training of the graph neural network. The above steps are repeated until a preset stopping iteration condition is met, indicating that the model training is complete. In one embodiment, the preset stopping iteration condition is reaching a preset maximum number of iterations; preferably, the preset maximum number of iterations is set to 500.

[0086] In one embodiment, after determining the device status and fault type of the target device, when the device status is faulty, the fault type of the target device can be obtained, and based on the fault type, the corresponding instruction of the target device can be triggered. This corresponding instruction includes at least one of a maintenance instruction or a shutdown instruction. Simultaneously, the diagnostic results are fed back to the monitoring system in a visual form. When a preset alarm threshold is triggered, maintenance suggestions or shutdown instructions are automatically generated, achieving a closed-loop response from data acquisition to fault decision-making, while also supporting historical diagnostic result retrospective analysis and root cause analysis of faults.

[0087] like Figure 2 As shown, this application embodiment also provides a machine learning-based industrial equipment fault diagnosis device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0088] The operation data of the target equipment is collected and preprocessed to obtain monitoring data; a pre-trained fault diagnosis model is determined, wherein the fault diagnosis model is an improved graph neural network model and the input layer of the improved graph neural network model adopts a fuzzy inference mechanism; the monitoring data is input into the fault diagnosis model to determine the equipment status and fault type of the target equipment.

[0089] This application embodiment also provides a non-volatile computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured as follows:

[0090] The operation data of the target equipment is collected and preprocessed to obtain monitoring data; a pre-trained fault diagnosis model is determined, wherein the fault diagnosis model is an improved graph neural network model and the input layer of the improved graph neural network model adopts a fuzzy inference mechanism; the monitoring data is input into the fault diagnosis model to determine the equipment status and fault type of the target equipment.

[0091] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device and medium embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the description of the method embodiments.

[0092] The devices and media provided in this application are one-to-one with the methods. Therefore, the devices and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0093] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0094] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0095] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0096] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0097] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0098] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0099] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0100] It should also be noted that 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 limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0101] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A machine learning-based industrial equipment failure diagnosis method, characterized by, include: Collect operational data from the target device and preprocess the operational data to obtain monitoring data; A pre-trained fault diagnosis model is determined, wherein the fault diagnosis model is an improved graph neural network model, and the input layer of the improved graph neural network model adopts a fuzzy inference mechanism. Before determining the pre-trained fault diagnosis model, the method further includes: Obtain historical operating data of the target device, and annotate the preprocessed historical operating data. The annotation content shall include at least the device status and fault type. The labeled historical operating data is input into the initial fault diagnosis model, and the initial fault diagnosis model is trained by setting the dynamic damping coefficient to obtain the fault diagnosis model, specifically including: Determine the initial damping coefficient, decay rate, current iteration of model training, and loss change rate corresponding to the current iteration; Based on the initial damping coefficient, the decay rate, the current iteration of model training, and the loss change rate corresponding to the current iteration, the damping coefficient corresponding to the current iteration is determined. The device information corresponding to the target device is determined, and the fault diagnosis model is stored based on the device information; the device information includes at least one of the following: device name, device type, device location, and device parameters. The monitoring data is input into the fault diagnosis model to determine the equipment status and fault type of the target equipment.

2. The method of claim 1, wherein, The process of collecting operational data from the target device and preprocessing the operational data to obtain monitoring data specifically includes: The runtime data is stored in JSON format; The operational data is cleaned to obtain first intermediate data; The first intermediate data is normalized to obtain the second intermediate data; The second intermediate data is discretized to obtain the monitoring data.

3. The method of claim 1, wherein, The step of inputting the monitoring data into the fault diagnosis model to determine the equipment status and fault type of the target equipment specifically includes: In the input layer of the fault diagnosis model, the target data with fuzzy labels corresponding to the target node is determined; Based on the feature distribution dispersion of the target data, determine the fuzzy confidence factor corresponding to the target data; Based on the preset fuzzy inference function, the original fuzzy features of the target node, and the fuzzy confidence factor, the input features corresponding to the target node are determined; Based on the input features, the device status and fault type of the target device are determined.

4. The method of claim 1, wherein, The step of inputting the monitoring data into the fault diagnosis model to determine the equipment status and fault type of the target equipment specifically includes: Identify the target node within the target layer, as well as the adjacent nodes of the target node; Obtain the first initial feature representation of the target node, the second initial feature representation of the adjacent nodes, and the adjacency matrix between the target node and the adjacent nodes; Based on the adjacency matrix, the first initial feature representation and the second initial feature representation are fused together to form the fused feature representation of the target node.

5. The method of claim 1, wherein, After determining the device status and fault type of the target device, the method further includes: When the device status of the target device is a fault state, the fault type of the target device is obtained; Based on the fault type, a corresponding instruction is triggered on the target device, the corresponding instruction including at least one of a maintenance instruction or a shutdown instruction.

6. The method of claim 1, wherein, Each piece of operational data includes at least a timestamp, device number, operational data at the collection location, sensor type, and sensor measurement value; In the fault diagnosis model, nodes represent data features corresponding to a line of operational data of the target device, and edges connecting nodes represent the similarity between lines of operational data. The data features corresponding to nodes are updated through graph convolutional layers.

7. A machine learning-based industrial equipment failure diagnosis device characterized by comprising: include: At least one processor; And, a memory communicatively connected to the at least one processor; 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 to enable the at least one processor to perform: Collect operational data from the target device and preprocess the operational data to obtain monitoring data; A pre-trained fault diagnosis model is determined, wherein the fault diagnosis model is an improved graph neural network model, and the input layer of the improved graph neural network model adopts a fuzzy inference mechanism. Before determining the pre-trained fault diagnosis model, the method further includes: Obtain historical operating data of the target device, and annotate the preprocessed historical operating data. The annotation content shall include at least the device status and fault type. The labeled historical operating data is input into the initial fault diagnosis model, and the initial fault diagnosis model is trained by setting the dynamic damping coefficient to obtain the fault diagnosis model, specifically including: Determine the initial damping coefficient, decay rate, current iteration of model training, and loss change rate corresponding to the current iteration; Based on the initial damping coefficient, the decay rate, the current iteration of model training, and the loss change rate corresponding to the current iteration, the damping coefficient corresponding to the current iteration is determined. The device information corresponding to the target device is determined, and the fault diagnosis model is stored based on the device information; the device information includes at least one of the following: device name, device type, device location, and device parameters. The monitoring data is input into the fault diagnosis model to determine the equipment status and fault type of the target equipment.

8. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions are set as follows: Collect operational data from the target device and preprocess the operational data to obtain monitoring data; A pre-trained fault diagnosis model is determined, wherein the fault diagnosis model is an improved graph neural network model, and the input layer of the improved graph neural network model adopts a fuzzy inference mechanism. Before determining the pre-trained fault diagnosis model, the method further includes: Obtain historical operating data of the target device, and annotate the preprocessed historical operating data. The annotation content shall include at least the device status and fault type. The labeled historical operating data is input into the initial fault diagnosis model, and the initial fault diagnosis model is trained by setting the dynamic damping coefficient to obtain the fault diagnosis model, specifically including: Determine the initial damping coefficient, decay rate, current iteration of model training, and loss change rate corresponding to the current iteration; Based on the initial damping coefficient, the decay rate, the current iteration of model training, and the loss change rate corresponding to the current iteration, the damping coefficient corresponding to the current iteration is determined. The device information corresponding to the target device is determined, and the fault diagnosis model is stored based on the device information; the device information includes at least one of the following: device name, device type, device location, and device parameters. The monitoring data is input into the fault diagnosis model to determine the equipment status and fault type of the target equipment.

Images