Method for automatically diagnosing and processing faults of nitrogen-saving device

Abstract

The invention discloses an automatic fault diagnosis processing method for a nitrogen saving device, and relates to the technical field of fault processing. Comprising the following steps of preprocessing acquired fault sample data, including abnormal data removal and dimension reduction, defining a loss function of an XGBoost algorithm, constructing a decision tree to construct a fault diagnosis model, and performing fault diagnosis on equipment in combination with feature vectors subjected to dimension reduction. According to the automatic fault diagnosis processing method for the nitrogen saving device, the fault is found in real time by obtaining the fault event, the fault finding time is effectively shortened, safe operation of the nitrogen saving device is protected, meanwhile, when nitrogen leakage occurs, the source of nitrogen leakage can be controlled according to reasons or leaked nitrogen can be extracted through the air extraction mechanism, and the fault diagnosis processing efficiency is improved. And the condition of large-scale diffusion of nitrogen is avoided.

Description

Technical Field

[0001] This invention relates to the field of fault handling technology, specifically to an automatic fault diagnosis and handling method for nitrogen-saving devices. Background Technology

[0002] A pressure swing adsorption (PSA) nitrogen generator is a nitrogen-generating device designed and manufactured based on pressure swing adsorption technology. It uses air as raw material and high-quality carbon molecular sieves as adsorbents. Utilizing the principle of pressure swing adsorption, it selectively adsorbs air through a microporous molecular sieve to achieve oxygen and nitrogen separation. Cryogenic nitrogen generation can produce both gaseous and liquid nitrogen, satisfying process requirements for liquid nitrogen. The liquid nitrogen can be stored in a storage tank. When there is a nitrogen supply interruption or minor maintenance of the air separation unit, the liquid nitrogen in the storage tank is heated in a vaporizer and then supplied to the product nitrogen pipeline to meet the nitrogen demand of the process unit. However, nitrogen leakage can occur during production and transportation.

[0003] The invention patent with patent number CN116251441A discloses a pressure alarm processing mechanism for a nitrogen generator, which includes: the pressure alarm processing mechanism of the nitrogen generator is configured by the cooperation between the rotating component, the detection component and the exhaust component, so that when there is a pressure failure and nitrogen leakage occurs, an alarm is triggered and the leaked gas is collected and processed in the same way, thereby preventing the large-scale diffusion of gas.

[0004] As shown in the invention above, existing nitrogen generators detect nitrogen leaks by using a detection component and then collect the leaked nitrogen through an exhaust component. However, this method still requires manual intervention after an alarm is triggered, resulting in a slow response to malfunctions, low automation, and a significant need for manpower. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an automatic diagnosis and handling method for nitrogen-saving device faults, thus solving the existing problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an automatic diagnosis and handling method for nitrogen-saving device faults, comprising the following steps: Step 1: Collect real-time operating data of the nitrogen-saving device through the information acquisition module; Step 2: Obtain multiple historical fault data of the nitrogen-saving device, and preprocess the fault sample data, dividing the preprocessed fault sample data into training data and test data; Step 3: Construct a fault diagnosis model using the XGBoost algorithm, train the fault diagnosis model using the training data, and verify the fault diagnosis model using the test data to obtain the optimal fault diagnosis model. Step 4: Analyze real-time equipment operation data using a fault diagnosis model. When a fault is detected, send an alarm to management personnel via a remote alarm module. Step 5: Analyze the cause of the fault based on the model. If the cause is nitrogen leakage from the pipeline, close the pipeline valve. If the cause is nitrogen leakage from the tank, collect the leaked gas using the extraction mechanism.

[0007] Preferably, the preprocessing steps for the fault sample data are as follows: S1: Obtain historical fault data of the equipment, and label the equipment data according to the historical fault data, and use the labeled equipment data as fault sample data; S2: Clear abnormal data from faulty sample data; S3: Perform feature extraction on the fault sample data after removing abnormal data to complete the preprocessing of the fault sample data.

[0008] Preferably, the step involves extracting time-domain features, frequency-domain features, and time-frequency-domain features from the fault sample data after removing abnormal data, normalizing the time-domain features, frequency-domain features, and time-frequency-domain features to obtain a feature vector, thereby completing the preprocessing of the fault sample data.

[0009] Preferably, the normalized expression is as follows: ; ;; in, Let x represent the normalized eigenvectors. i The eigenvector represents the first eigenvector. Eigenvalues Indicates the number of faulty samples. Represents variance. Represents a constant, taken as 10. -8 .

[0010] Preferably, the step of constructing the fault diagnosis model using the XGBoost algorithm is as follows: 1) Based on the preprocessed fault sample data, define a custom loss function for the XGBoost algorithm; 2) Initialize the predicted value for each fault sample data; 3) Calculate the derivative of the loss function with respect to the predicted value of each fault sample data; 4) Build an XGBoost decision tree based on the derivative information, and obtain a fault diagnosis model based on the XGBoost decision tree.

[0011] Preferably, the expression for the loss function is as follows: ; ; Among them, J(f t ) represents the loss function, n represents the total number of faulty samples, and y i Indicates the true category of the fault. Ω(f) represents the fault prediction value at iteration t-1. t ) represents the regularization term, C represents the constant term, and f t (x i Let ) represent the objective function at the t-th iteration, and γ and Both represent pre-designed hyperparameters, T t w represents the number of leaf nodes. j Let J represent the weight of the leaf node j, and T represent the number of leaf nodes.

[0012] Preferably, the training data is used to train the DRN fault diagnosis model, specifically as follows: The training data is input into the deep residual network, and the model output is calculated through forward propagation. The model output is compared with the true label, the value of the loss function is calculated, the gradient of the model parameters is calculated using the chain rule, the gradient of the loss function is passed from the output layer back to the input layer, and the model parameters are updated using gradient descent based on the calculated gradient and learning rate.

[0013] Preferably, the test data is used to verify the fault diagnosis model, specifically as follows: The test data is input into the model, and the output of the model is obtained through forward propagation. The output of the fault diagnosis model is compared with the true label of the test data to determine whether the performance index of the fault diagnosis model is qualified. If it is qualified, the optimal fault diagnosis model is output; otherwise, the model is retrained. Beneficial effects

[0014] This invention provides an automatic diagnosis and handling method for nitrogen-saving device malfunctions. Compared with existing technologies, it has the following advantages: 1. The automatic fault diagnosis and processing method for the nitrogen-saving device preprocesses the acquired fault sample data, including removing abnormal data and dimensionality reduction. At the same time, the present invention defines the loss function of its XGBoost algorithm, constructs a decision tree to build a fault diagnosis model, and combines the dimensionality-reduced feature vector to diagnose the equipment fault. It can solve the fault diagnosis of equipment with complex fault mechanisms, effectively reduce the safety hazards of the equipment, and has good diagnostic performance, improved fault diagnosis efficiency, and features automation and intelligence, which helps to protect the safe operation of the nitrogen-saving device.

[0015] 2. The automatic fault diagnosis and handling method for the nitrogen unit in this section improves the effectiveness of the sample data by clearing abnormal data from the fault sample data, which helps improve the accuracy of model training. By using a custom loss function of the XGBoost algorithm to construct the fault diagnosis model, the reliance on the experience of management personnel is reduced, realizing an automated and intelligent fault diagnosis process, improving the operability and sustainability of the system. By acquiring fault events in real time to detect faults, the fault detection time is effectively shortened. When nitrogen leakage occurs, the source of nitrogen leakage can be controlled according to the cause, or the leaked nitrogen can be extracted by the pumping mechanism to avoid large-scale nitrogen diffusion. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the fault sample data preprocessing process of the present invention; Figure 3 This is a schematic diagram of the process of constructing a fault diagnosis model using the XGBoost algorithm according to the present invention; Figure 4 This is a schematic diagram of the nitrogen leakage handling process of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] See Figure 1-4 This invention provides four technical solutions: First implementation method: An automatic diagnosis and handling method for nitrogen-saving device faults, comprising the following steps: Step 1: Collect real-time operating data of the nitrogen-saving device through the information acquisition module; Step 2: Obtain multiple historical fault data of the nitrogen-saving device, and preprocess the fault sample data, dividing the preprocessed fault sample data into training data and test data; Step 3: Construct a fault diagnosis model using the XGBoost algorithm, train the fault diagnosis model using the training data, and verify the fault diagnosis model using the test data to obtain the optimal fault diagnosis model. Step 4: Analyze real-time equipment operation data using a fault diagnosis model. When a fault is detected, send an alarm to management personnel via a remote alarm module. Step 5: Analyze the cause of the fault based on the model. If the cause is nitrogen leakage from the pipeline, close the pipeline valve. If the cause is nitrogen leakage from the tank, collect the leaked gas using the extraction mechanism.

[0019] In the above method, the training data is used to train the DRN fault diagnosis model, specifically as follows: The training data is input into the deep residual network, and the model output is calculated through forward propagation. The model output is compared with the true label, the value of the loss function is calculated, the gradient of the model parameters is calculated using the chain rule, the gradient of the loss function is passed from the output layer back to the input layer, and the model parameters are updated using gradient descent based on the calculated gradient and learning rate.

[0020] In the above method, the test data is used to verify the fault diagnosis model, specifically as follows: The test data is input into the model, and the output of the model is obtained through forward propagation. The output of the fault diagnosis model is compared with the true label of the test data to determine whether the performance index of the fault diagnosis model is qualified. If it is qualified, the optimal fault diagnosis model is output; otherwise, the model is retrained.

[0021] The second implementation method involves the following steps for preprocessing fault sample data: S1: Obtain historical fault data of the equipment, and label the equipment data according to the historical fault data, and use the labeled equipment data as fault sample data; S2: Clear abnormal data from faulty sample data; S3: Perform feature extraction on the fault sample data after removing abnormal data to complete the preprocessing of the fault sample data.

[0022] The process involves extracting time-domain features, frequency-domain features, and time-frequency-domain features from the fault sample data after removing abnormal data, normalizing the time-domain features, frequency-domain features, and time-frequency-domain features to obtain feature vectors, and completing the preprocessing of the fault sample data.

[0023] The normalization expression is as follows: ; ; in, Let x represent the normalized eigenvectors. i The eigenvector represents the first eigenvector. Eigenvalues Indicates the number of faulty samples. Represents variance. Represents a constant, taken as 10. -8 .

[0024] The third implementation method: The steps for constructing the fault diagnosis model using the XGBoost algorithm are as follows: 1) Based on the preprocessed fault sample data, define a custom loss function for the XGBoost algorithm; 2) Initialize the predicted value for each fault sample data; 3) Calculate the derivative of the loss function with respect to the predicted value of each fault sample data; 4) Build an XGBoost decision tree based on the derivative information, and obtain a fault diagnosis model based on the XGBoost decision tree.

[0025] The expression for the loss function is as follows: ; ; Among them, J(f t ) represents the loss function, n represents the total number of faulty samples, and y i Indicates the true category of the fault. Ω(f) represents the fault prediction value at iteration t-1. t ) represents the regularization term, C represents the constant term, and f t (x i Let ) represent the objective function at the t-th iteration, and γ and Both represent pre-designed hyperparameters, T t w represents the number of leaf nodes. j Let J represent the weight of the leaf node j, and T represent the number of leaf nodes.

[0026] Fourth implementation: The automatic fault diagnosis and processing method for nitrogen-saving device of the present invention provides a processing system, including an acquisition unit configured to acquire a target fault event; and a fault processing unit configured to perform the following fault processing steps: in response to acquiring the target fault event, confirming whether a fault has occurred, wherein different faults correspond to different fault events; if a fault is confirmed to have occurred, sending a fault notification corresponding to the target fault event to the terminal device of the target user; and performing a loss-stopping processing operation corresponding to the target fault event.

[0027] In a specific embodiment, the system is applied to the server and further includes: a configuration unit configured to configure different fault events for different faults of the server's service before performing the following fault handling steps; and a determination unit configured to determine the fault handling steps for each configured fault event.

[0028] In a specific embodiment, the fault handling unit is further configured to perform at least one of the following actions: sending a fault notification corresponding to the target fault event to the target user's terminal device; sending an email to the target user's terminal device; sending a short message to the target user's terminal device, wherein the content of the email and the content of the short message are both related to the fault corresponding to the target fault event, and the fault includes at least one of process fault and program fault.

[0029] In some embodiments, the system further includes a log writing unit configured to write execution information of each sub-step in the fault handling steps for the target fault event into a fault handling log, wherein the fault handling log is stored in a preset document-type database. In some embodiments, the fault handling steps further include: after performing the loss-stopping operation corresponding to the target fault event, outputting a manual intervention notification.

[0030] This application provides an electronic device, including: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement the method as described in any embodiment of the fault handling method.

[0031] This application provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method as described in any embodiment of the fault handling method.

[0032] The aforementioned devices can be connected to I / O interfaces, including input devices such as touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, and gyroscopes; output devices such as liquid crystal displays (LCDs), speakers, and vibrators; storage devices such as magnetic tapes and hard disks; and communication devices. Communication devices allow electronic devices to communicate wirelessly or wiredly with other devices to exchange data. While electronic devices with various devices are shown, it should be understood that it is not required to implement or possess all of the shown devices. More or fewer devices may be implemented or possessed alternatively.

[0033] 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 process, method, article, or apparatus.

[0034] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An automatic diagnosis and handling method for nitrogen-saving device faults, characterized in that, Includes the following steps: Step 1: Collect real-time operating data of the nitrogen-saving device through the information acquisition module; Step 2: Obtain multiple historical fault data of the nitrogen-saving device, and preprocess the fault sample data, dividing the preprocessed fault sample data into training data and test data; Step 3: Construct a fault diagnosis model using the XGBoost algorithm, train the fault diagnosis model using the training data, and verify the fault diagnosis model using the test data to obtain the optimal fault diagnosis model. The DRN fault diagnosis model is trained using the training data, specifically as follows: The training data is input into the deep residual network, the model output is calculated through forward propagation, the model output is compared with the true label, the value of the loss function is calculated, the gradient with respect to the model parameters is calculated using the chain rule, the gradient of the loss function is passed from the output layer back to the input layer, and the model parameters are updated using gradient descent based on the calculated gradient and learning rate. The test data was used to validate the fault diagnosis model, specifically as follows: The test data is input into the model, and the output of the model is obtained through forward propagation. The output of the fault diagnosis model is compared with the true label of the test data to determine whether the performance index of the fault diagnosis model is qualified. If it is qualified, the optimal fault diagnosis model is output; otherwise, the model is retrained. Step 4: Analyze real-time equipment operation data using a fault diagnosis model. When a fault is detected, send an alarm to management personnel via a remote alarm module. Step 5: Analyze the cause of the fault based on the model. If the cause is nitrogen leakage from the pipeline, close the pipeline valve. If the cause is nitrogen leakage from the tank, collect the leaked gas using the extraction mechanism.

2. The automatic diagnosis and handling method for nitrogen-saving device faults according to claim 1, characterized in that: The steps for preprocessing the fault sample data are as follows: S1: Obtain historical fault data of the equipment, and label the equipment data according to the historical fault data, and use the labeled equipment data as fault sample data; S2: Clear abnormal data from faulty sample data; S3: Perform feature extraction on the fault sample data after removing abnormal data to complete the preprocessing of the fault sample data.

3. The automatic diagnosis and handling method for nitrogen-saving device faults according to claim 2, characterized in that: The process involves extracting time-domain features, frequency-domain features, and time-frequency-domain features from the fault sample data after removing abnormal data, normalizing the time-domain features, frequency-domain features, and time-frequency-domain features to obtain feature vectors, and completing the preprocessing of the fault sample data.

4. The automatic fault diagnosis and processing method for a nitrogen-saving device according to claim 3, characterized in that: The normalization expression is as follows: ； ； in, Let x represent the normalized eigenvectors. i The eigenvector represents the first eigenvector. Eigenvalues Indicates the number of faulty samples. Represents variance. Represents a constant, taken as 10. -8 .

5. The automatic diagnosis and handling method for nitrogen-saving device faults according to claim 1, characterized in that: The steps for constructing a fault diagnosis model using the XGBoost algorithm are as follows: 1) Based on the preprocessed fault sample data, define a custom loss function for the XGBoost algorithm; 2) Initialize the predicted value for each fault sample data; 3) Calculate the derivative of the loss function with respect to the predicted value of each fault sample data; 4) Build an XGBoost decision tree based on the derivative information, and obtain a fault diagnosis model based on the XGBoost decision tree.

6. The automatic diagnosis and handling method for nitrogen-saving device faults according to claim 5, characterized in that: The expression for the loss function is as follows: ； ； Among them, J(f t ) represents the loss function, n represents the total number of faulty samples, and y i Indicates the true category of the fault. Ω(f) represents the fault prediction value at iteration t-1. t ) represents the regularization term, C represents the constant term, and f t (x i Let ) represent the objective function at the t-th iteration, and γ and Both represent pre-designed hyperparameters, T t w represents the number of leaf nodes. j Let J represent the weight of the leaf node j, and T represent the number of leaf nodes.

Images