High-pressure air compressor fault alarm method and device, electronic terminal and storage medium

Abstract

The invention discloses a high-pressure air compressor fault alarm method and device, an electronic terminal and a storage medium in the technical field of power transformation and distribution. The method comprises the steps that time sequence features are obtained; inputting the time sequence feature into the prediction model, extracting a local feature of the time sequence feature through a convolutional layer, capturing a hidden state of the local feature in a current time step through a GRU layer, generating a comprehensive sequence representation, integrating the comprehensive sequence representation based on a full-connection layer to obtain an integrated feature, and outputting the integrated feature. Obtaining a predicted voltage at an output layer based on the integrated features; and triggering a monitoring alarm corresponding to a prediction result when the prediction voltage is detected to be greater than a preset upper limit voltage or less than a preset lower limit voltage in a preset detection period. According to the invention, the technical problem that the fault recovery time is prolonged due to the fact that maintenance personnel are difficult to quickly identify the fault and take corresponding measures because the upper monitoring system cannot provide clear alarm content can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power transformation and distribution, and in particular to a method, device, electronic terminal and storage medium for high-voltage air compressor fault alarm. Background Art

[0002] In the cigarette manufacturing industry, high-voltage motors play a vital role. Take the high-voltage air compressor as an example. It is responsible for converting mechanical energy into gas pressure energy and providing the necessary compressed air for each link of the cigarette production line. These links include cigarette making machines, packaging machines, silk making equipment, logistics equipment, and dust removal equipment. The stable operation of the high-voltage air compressor is crucial to ensure the efficiency of the entire production process and product quality.

[0003] In the event of power grid fluctuations, high-pressure air compressors may suddenly shut down, causing a sharp drop in pressure in the gas supply pipeline. Although there are air compressors in hot standby mode on site, due to the lack of effective remote control methods, it is difficult for operators to start the standby equipment in time to maintain or restore pipeline pressure.

[0004] When a fault occurs, maintenance personnel need to handle it according to the alarm information on the air compressor panel. However, the existing upper monitoring system does not provide clear alarm content, which makes it difficult for maintenance personnel to quickly identify the problem and take corresponding measures, thus prolonging the recovery time of the fault.

[0005] Therefore, there is an urgent need for a high-pressure air compressor fault alarm method, device, electronic terminal and storage medium to solve the above technical problems. Summary of the invention

[0006] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a high-pressure air compressor fault alarm method, which can solve the technical problem that the upper-level monitoring system used in the prior art cannot provide clear alarm content whether it is related to voltage changes, making it difficult for maintenance personnel to quickly identify faults and take corresponding measures, thereby prolonging the fault recovery time.

[0007] To achieve the above object, the present invention is implemented by adopting the following technical solutions: In a first aspect, the present invention provides a high-pressure air compressor fault alarm method, comprising: Obtain external line operation data of high-pressure air compressor; Reshape the external line operation data to obtain time series features that meet the prediction requirements of the pre-built prediction model; Input the time series features into the prediction model, extract local features of the time series features through the convolution layer of the prediction model, capture the hidden state of the local features at the current time step through the GRU layer of the prediction model, generate a comprehensive sequence representation, integrate the comprehensive sequence representation based on the fully connected layer of the prediction model to obtain integrated features, and obtain a predicted voltage based on the integrated features at the output layer; When it is detected within a preset detection period that the predicted voltages are all greater than a preset upper limit voltage, or are all less than a preset lower limit voltage, a monitoring alarm corresponding to the prediction result is triggered.

[0008] Furthermore, the training process of the prediction model includes: Acquire historical external line operation data of the high-pressure air compressor, reshape the historical external line operation data, and acquire time series features that meet the training requirements of the prediction model; Inputting the time series features into the convolution layer of the prediction model to extract local features; Inputting the local features into the GRU layer of the prediction model to capture the hidden state of the local features at the current time step and generate a comprehensive sequence representation; Integrating the comprehensive sequence representation based on the fully connected layer of the prediction model to obtain integrated features, and obtaining a predicted voltage based on the integrated features at the output layer; A combined loss function is calculated based on the predicted voltage and historical external line operation data, the gradient of the combined loss function with respect to the model parameters is calculated by a back propagation algorithm, and the model parameters and the learning rate are updated according to the gradient until the combined loss function converges or reaches a preset maximum number of training rounds.

[0009] Further, calculating the combined loss function based on the predicted voltage and historical external line operation data includes: , , , , , , in, is the combined loss function, is the mean square error of the predicted voltage, is the mean square logarithmic error of the predicted voltage, is the gradient smoothing error of the predicted voltage, is the mean square error of power prediction, are weight coefficients of loss terms, The historical external line operation data is i Voltage data, For the i The predicted voltage, N is the total number of voltage data, The historical external line operation data is j Power data, For the j The predicted power, M is the total number of power data.

[0010] Furthermore, it also includes: when the monitoring alarm corresponding to the prediction result is triggered and the high-voltage air compressor is in the open state, the lockout voltage protection is started; when the monitoring alarm corresponding to the prediction result is triggered and the high-voltage air compressor is in the closed state, the normal tripping procedure is executed.

[0011] Furthermore, it also includes: After detecting the trip, calculate the voltage difference in the external line operation data before and after the trip. When the absolute value of the voltage difference is greater than or equal to the preset trip threshold, judge that the cause of the trip is the external line voltage fluctuation. Record the cause of the trip and keep monitoring. When the monitoring result shows that the voltage fluctuation is normal, send a reset command to control the high-voltage air compressor to start normally.

[0012] Furthermore, reshaping the external line operation data includes: Establishing a voltage data set and a power data set according to the voltage and power data in the external line operation data; Standardizing the voltage dataset and the power dataset by using a Z score to obtain a standardized voltage dataset and a standardized power dataset; Creating a voltage sequence and a power sequence according to the standardized voltage data set and power data set, and assigning corresponding sequence labels to the voltage sequence and the power sequence; Based on the sequences and sequence labels, creating a serialized data training set, a serialized data validation set, and a serialized data test set; Reshape the sequence data in the serialized data training set, the serialized data validation set, and the serialized data test set to obtain a combined training set input, a combined validation set input, and a combined test set input; Reshape the label data in the serialized data training set, the serialized data validation set, and the serialized data test set to obtain a reshaped two-dimensional label array of the training set, a two-dimensional label array of the validation set, and a two-dimensional label array of the test set; Constructing a time series feature that meets the prediction model training requirements based on the combined training set input and the training set two-dimensional label array; Constructing a time series feature that meets the prediction model verification requirements based on the combined verification set input and the verification set two-dimensional label array; A time series feature that meets the prediction requirements of the prediction model is constructed based on the combined test set input and the test set two-dimensional label array.

[0013] In a second aspect, the present invention provides a high-pressure air compressor fault alarm device, comprising: The acquisition module is used to obtain the external line operation data of the high-pressure air compressor; A data reshaping module, used to reshape the external line operation data to obtain time series features that meet the prediction requirements of the pre-built prediction model; A prediction module, used for inputting the time series features into the prediction model, extracting local features of the time series features through the convolution layer of the prediction model, capturing the hidden state of the local features at the current time step through the GRU layer of the prediction model, generating a comprehensive sequence representation, integrating the comprehensive sequence representation based on the fully connected layer of the prediction model to obtain integrated features, and obtaining a predicted voltage based on the integrated features at the output layer; The alarm module is used to trigger a monitoring alarm corresponding to the prediction result when it is detected within a preset detection period that the predicted voltages are all greater than a preset upper limit voltage, or are all less than a preset lower limit voltage.

[0014] In a third aspect, the present invention provides an electronic terminal, comprising a processor and a memory connected to the processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the steps of any of the above methods are performed.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of any of the above methods when executed by a processor.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention can timely predict and respond to possible abnormal situations by real-time monitoring of voltage fluctuations of external line voltages. When it is detected that the predicted voltages are all greater than a preset upper voltage limit, or are all less than a preset lower voltage limit within a preset detection period, a monitoring alarm corresponding to the prediction result is triggered, which can be considered to provide clear alarm content, which helps to reduce the shutdown of high-voltage air compressors caused by power grid fluctuations, thereby improving the stability of the power system. Ability to continuously optimize its prediction algorithm based on historical data, improve the accuracy of fault prediction and reduce the possibility of false alarms; Through timely alarms and automatic recovery mechanisms, the system can automatically release previous protection measures after detecting that the voltage has returned to normal, reducing the need for manual intervention, improving the system's self-healing ability, and better preventing equipment damage or power outages caused by voltage anomalies, thereby improving the safety of the entire power system, reducing production interruption time, and ensuring the production efficiency and product quality of the cigarette factory. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of a high-pressure air compressor fault alarm method provided by the first embodiment of the present invention; Figure 2 It is a complete execution diagram of a high-pressure air compressor fault alarm method provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0018] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0019] The term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship. Embodiment 1:

[0020] Figure 1 This is a flow chart of the high-pressure air compressor fault alarm method in the first embodiment of the present invention. This flow chart only shows the logical sequence of the method described in this embodiment. Under the premise of no conflict, in other possible embodiments of the present invention, different Figure 1 The steps shown or described are accomplished in the order shown.

[0021] The high-pressure air compressor fault alarm method provided in this embodiment can be applied to a terminal and can be executed by a high-pressure air compressor fault alarm device, which can be implemented by software and / or hardware, and can be integrated in a terminal, such as any smart phone, tablet computer or computer device with communication function. Figure 1 and Figure 2 As shown, the method of this embodiment specifically includes the following steps: Step 1: Obtaining the external line operation data of the high-pressure air compressor, the external line operation data includes real-time external line operation data and historical external line operation data, which are the same in data type except for the difference in acquisition time, and include: voltage data ( ): Three-phase voltage instantaneous value, power data ( ): Active power and reactive power, harmonic data ( ): The first 9 individual harmonics (including amplitude, phase and interharmonics), load data ( ): Three-phase load current. Before reshaping the data, the data can be cleaned in advance: missing values, outliers and duplicate data can be processed; Step 2: reshaping the external line operation data, including: A voltage data set is established based on the voltage and power data in the external line operation data. X voltage and power datasets X power ; The voltage dataset is analyzed by Z-score X voltage and power datasets X power Perform standardization to obtain a standardized voltage data set Z voltage and power datasets Z power : Compute the mean and standard deviation of the voltage data set: , , Compute the mean and standard deviation of the power data set: , , in, Refers to the voltage data set X voltage and power datasets X power The total number of data points in , and Z-score normalization is performed on each data point: For each data point in the voltage dataset : , For each data point in the power dataset : ; Combine the data points in the voltage dataset after Z-score standardization to obtain the standardized voltage dataset Z voltage , combine the data points in the power dataset after Z-score standardization to obtain the standardized and power dataset Z power ; According to the voltage data set after the standardization Z voltage and power datasets Z power Creating a Voltage Sequence S voltage，i and power sequence S power，i and for the voltage sequence S voltage，i and power sequence S power，i Assign corresponding sequence labels: Each collection contains data points, according to the preset training set ratio a, validation set ratio b and test set ratio c, where a+b+c=1, calculate the start and end indexes of the training set, validation set and test set. In this embodiment, a=60%, b=20%, c=20%, The starting index of the training set starts from 0 and the ending index is: , in, Indicates rounding down to ensure that the index is an integer. The starting index of the validation set is immediately followed by the ending index of the training set. The ending index of the validation set is: , The starting index of the test set is immediately followed by the ending index of the validation set. Since the test set usually occupies all the remaining data, its ending index is , whose starting index is: , Split the data according to the calculated index:

[0022] ,

[0023]

[0024]

[0025] , Set the sequence length T, and for each data point i (from T to N), create a voltage sequence and power sequence : , , Assign labels to the voltage sequences: , Assign labels to the power sequences: , Based on the sequences and sequence labels, a serialized data training set, a serialized data validation set, and a serialized data test set are created: Store each sequence and its label as a pair in a collection, and create serialized datasets for training, validation, and test sets, respectively. Among them, the training set serialization data is: The validation set serialized data is:

[0026] The test set serialized data is:

[0027] The sequence data in the serialized data training set, the serialized data validation set, and the serialized data test set are reshaped to obtain a merged training set input, a merged validation set input, and a merged test set input: The input data of the training set, validation set, and test set are three-dimensional arrays with a shape of (number of samples, time steps, number of features), where the number of samples determines the number of samples in the training set, validation set, and test set. , the time step is T, which is 360 in this embodiment, the number of features is F, which is 35 in this embodiment, Take the training set reshaping as an example: Voltage input data Power input data Taking voltage and power as two feature inputs of a sequence, the two data sets are merged, and the merged training set input is: Similar to the training set, the data of the validation set and the test set are reshaped and merged to finally obtain the merged validation set input and the merged test set input. Among them, the combined validation set input is: Combined test set input: The label data in the serialized data training set, the serialized data validation set, and the serialized data test set are reshaped to obtain the reshaped training set two-dimensional label array, the validation set two-dimensional label array, and the test set two-dimensional label array: For the label data, reshape it into a two-dimensional array. Since the dataset contains two features (voltage and power), we have merged them into one input dataset, so the label data should also be reshaped accordingly to match the output requirements of the model. Reshape the training set voltage labels: , in, is the original one-dimensional label array of the training set, is the number of samples in the training set, is the reshaped two-dimensional label array of the training set, whose shape is , Validation set voltage label reshape:

[0028] in, is the number of samples in the validation set, is the reshaped validation set two-dimensional label array, in, is the number of samples in the test set, is the reshaped test 2D label array; Constructing a time series feature that meets the prediction model training requirements based on the combined training set input and the training set two-dimensional label array; Constructing a time series feature that meets the prediction model verification requirements based on the combined verification set input and the verification set two-dimensional label array; Constructing a time series feature that meets the prediction requirements of the prediction model according to the combined test set input and the test set two-dimensional label array; Step 3: input the time series features that meet the prediction requirements of the prediction model into the prediction model, extract the local features of the time series features through the convolution layer of the prediction model, capture the hidden state of the local features at the current time step through the GRU layer of the prediction model, generate a comprehensive sequence representation, integrate the comprehensive sequence representation based on the fully connected layer of the prediction model to obtain integrated features, and obtain the predicted voltage based on the integrated features in the output layer; Step 4: When it is detected within a preset detection period that the predicted voltages are all greater than the preset upper limit voltage, or are all less than the preset lower limit voltage, a monitoring alarm corresponding to the prediction result is triggered. When it is detected within a preset detection period that the predicted voltages are all greater than the preset upper limit voltage, the system will consider that the output prediction value exceeds the upper limit and trigger an alarm of the upper-level monitoring system; when it is detected within a preset detection period that the predicted voltages are all less than the preset lower limit voltage, an alarm of the upper-level monitoring system is also triggered to realize the external line voltage fluctuation early warning function.

[0029] The training process of the prediction model includes: Acquire historical external line operation data of the high-pressure air compressor, reshape the historical external line operation data, and acquire time series features that meet the training requirements of the prediction model; Inputting the time series features into the convolution layer of the prediction model to extract local features; Inputting the local features into the GRU layer of the prediction model to capture the hidden state of the local features at the current time step and generate a comprehensive sequence representation; Integrating the comprehensive sequence representation based on the fully connected layer of the prediction model to obtain integrated features, and obtaining a predicted voltage based on the integrated features at the output layer; The combined loss function is calculated based on the predicted voltage and the historical external line operation data: , , , , , , in, is the combined loss function, is the mean square error of the predicted voltage, is the mean square logarithmic error of the predicted voltage, is the gradient smoothing error of the predicted voltage, is the mean square error of power prediction, are weight coefficients of loss terms, The historical external line operation data is i Voltage data, For the i The predicted voltage, N is the total number of voltage data, The historical external line operation data is j Power data, For the j The predicted power, M is the total number of power data. The predicted power is not used as the output of the prediction model in this embodiment, but is only added as a reference item for constructing the combined loss function. The correlation between voltage and power can improve the overall performance of the model. The gradient of the combined loss function with respect to the model parameters is calculated by the back propagation algorithm, and the model parameters and the learning rate are updated according to the gradient until the combined loss function converges or reaches a preset maximum number of training rounds.

[0030] A Dropout layer can be added after the fully connected layer and the GRU layer, with the dropout rate set to 0.5. L2 regularization is to add an L2 regularization term to the weight parameters of the model to constrain the size of the parameters. Based on the Early Stopping mechanism, if the loss of the validation set does not decrease within 15 consecutive epochs, training is stopped to help prevent overfitting.

[0031] When the monitoring alarm corresponding to the predicted result is triggered and the high-voltage air compressor is in the open state, the lockout voltage protection is activated. The lockout voltage protection is an important safety feature. It ensures that the high-voltage air compressor will not start accidentally in the open state through multiple mechanisms, thereby protecting the safety of equipment and operators; when the monitoring alarm corresponding to the predicted result is triggered and the high-voltage air compressor is in the closed state, the normal tripping procedure is executed.

[0032] After the trip is detected, the voltage difference in the external line operation data before and after the trip is calculated. When the absolute value of the voltage difference is greater than or equal to the preset trip threshold, the trip cause is determined to be external line voltage fluctuation. The trip cause is recorded and continuously monitored. When the monitoring result shows that the voltage fluctuation is normal, a reset command is sent to control the high-voltage air compressor to start normally. When the absolute value of the voltage difference is less than the preset tripping threshold, it is determined that the tripping cause is not caused by external line voltage fluctuation, and manual processing is performed. After manual processing and confirmation, a reset command is sent to control the high-voltage air compressor to start normally.

[0033] Embodiment 2: Embodiment 2 of the present invention provides a high-pressure air compressor fault alarm device, comprising: The acquisition module is used to obtain the external line operation data of the high-pressure air compressor; A data reshaping module, used to reshape the external line operation data to obtain time series features that meet the prediction requirements of the pre-built prediction model; A prediction module, used for inputting the time series features into the prediction model, extracting local features of the time series features through the convolution layer of the prediction model, capturing the hidden state of the local features at the current time step through the GRU layer of the prediction model, generating a comprehensive sequence representation, integrating the comprehensive sequence representation based on the fully connected layer of the prediction model to obtain integrated features, and obtaining a predicted voltage based on the integrated features at the output layer; The alarm module is used to trigger a monitoring alarm corresponding to the prediction result when it is detected within a preset detection period that the predicted voltages are all greater than a preset upper limit voltage, or are all less than a preset lower limit voltage.

[0034] The high-pressure air compressor fault alarm device provided in the second embodiment of the present invention can execute the high-pressure air compressor fault alarm method provided in the first embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0035] Embodiment three: Embodiment 3 of the present invention further provides an electronic terminal, including a processor and a memory connected to the processor, wherein a computer program is stored in the memory, and the processor is configured to operate according to the instructions to execute the steps of the method described in embodiment 1.

[0036] The electronic terminal provided in the third embodiment of the present invention can execute the high-pressure air compressor fault alarm method provided in the first embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0037] Embodiment 4: Embodiment 4 of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in Embodiment 1 are implemented, and the computer program has functional modules and beneficial effects corresponding to the execution method.

[0038] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, devices, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

Claims

1. A high-pressure air compressor fault alarm method, characterized in that: include: Obtain external line operation data of high-pressure air compressor; Reshape the external line operation data to obtain time series features that meet the prediction requirements of the pre-built prediction model; Input the time series features into the prediction model, extract local features of the time series features through the convolution layer of the prediction model, capture the hidden state of the local features at the current time step through the GRU layer of the prediction model, generate a comprehensive sequence representation, integrate the comprehensive sequence representation based on the fully connected layer of the prediction model to obtain integrated features, and obtain a predicted voltage based on the integrated features at the output layer; When it is detected within a preset detection period that the predicted voltages are all greater than a preset upper limit voltage, or are all less than a preset lower limit voltage, a monitoring alarm corresponding to the prediction result is triggered.

2. A high-pressure air compressor fault alarm method according to claim 1, characterized in that: Also includes: The training process of the prediction model includes: Acquire historical external line operation data of the high-pressure air compressor, reshape the historical external line operation data, and acquire time series features that meet the training requirements of the prediction model; Inputting the time series features into the convolution layer of the prediction model to extract local features; Inputting the local features into the GRU layer of the prediction model to capture the hidden state of the local features at the current time step and generate a comprehensive sequence representation; Integrating the comprehensive sequence representation based on the fully connected layer of the prediction model to obtain integrated features, and obtaining a predicted voltage based on the integrated features at the output layer; A combined loss function is calculated based on the predicted voltage and historical external line operation data, the gradient of the combined loss function with respect to the model parameters is calculated by a back propagation algorithm, and the model parameters and the learning rate are updated according to the gradient until the combined loss function converges or reaches a preset maximum number of training rounds.

3. A high-pressure air compressor fault alarm method according to claim 2, characterized in that: Calculating the combined loss function based on the predicted voltage and historical external line operation data includes: ， ， ， ， ， ， in, is the combined loss function, is the mean square error of the predicted voltage, is the mean square logarithmic error of the predicted voltage, is the gradient smoothing error of the predicted voltage, is the mean square error of power prediction, are weight coefficients of loss terms, The historical external line operation data is i Voltage data, For the i The predicted voltage, N is the total number of voltage data, The historical external line operation data is j Power data, For the j The predicted power, M is the total number of power data.

4. A high-pressure air compressor fault alarm method according to claim 1, characterized in that: Also includes: When the monitoring alarm corresponding to the predicted result is triggered and the high-voltage air compressor is in the open state, the lockout voltage protection is started; when the monitoring alarm corresponding to the predicted result is triggered and the high-voltage air compressor is in the closed state, the normal tripping procedure is executed.

5. A high-pressure air compressor fault alarm method according to claim 4, characterized in that: Also includes: After detecting the trip, calculate the voltage difference in the external line operation data before and after the trip. When the absolute value of the voltage difference is greater than or equal to the preset trip threshold, judge that the cause of the trip is the external line voltage fluctuation. Record the cause of the trip and keep monitoring. When the monitoring result shows that the voltage fluctuation is normal, send a reset command to control the high-voltage air compressor to start normally.

6. A high-pressure air compressor fault alarm method according to claim 2, characterized in that: The data reshaping of the external line operation data includes: Establishing a voltage data set and a power data set according to the voltage and power data in the external line operation data; Standardizing the voltage dataset and the power dataset by using a Z score to obtain a standardized voltage dataset and a standardized power dataset; Creating a voltage sequence and a power sequence according to the standardized voltage data set and the power data set, and assigning corresponding sequence labels to the voltage sequence and the power sequence; Based on the sequences and sequence labels, creating a serialized data training set, a serialized data validation set, and a serialized data test set; Reshape the sequence data in the serialized data training set, the serialized data validation set, and the serialized data test set to obtain a combined training set input, a combined validation set input, and a combined test set input; Reshape the label data in the serialized data training set, the serialized data validation set, and the serialized data test set to obtain a reshaped two-dimensional label array of the training set, a two-dimensional label array of the validation set, and a two-dimensional label array of the test set; Constructing a time series feature that meets the prediction model training requirements based on the combined training set input and the training set two-dimensional label array; Constructing a time series feature that meets the prediction model verification requirements based on the combined verification set input and the verification set two-dimensional label array; A time series feature that meets the prediction requirements of the prediction model is constructed based on the combined test set input and the test set two-dimensional label array.

7. A high-pressure air compressor fault alarm device, characterized in that: include: The acquisition module is used to obtain the external line operation data of the high-pressure air compressor; A data reshaping module, used to reshape the external line operation data to obtain time series features that meet the prediction requirements of the pre-built prediction model; A prediction module, used for inputting the time series features into the prediction model, extracting local features of the time series features through the convolution layer of the prediction model, capturing the hidden state of the local features at the current time step through the GRU layer of the prediction model, generating a comprehensive sequence representation, integrating the comprehensive sequence representation based on the fully connected layer of the prediction model to obtain integrated features, and obtaining a predicted voltage based on the integrated features at the output layer; The alarm module is used to trigger a monitoring alarm corresponding to the prediction result when it is detected within a preset detection period that the predicted voltages are all greater than a preset upper limit voltage, or are all less than a preset lower limit voltage.

8. An electronic terminal, characterized in that: The method comprises a processor and a memory connected to the processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 6 are executed.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images