A method and system for monitoring the state of an electrical device

Abstract

The present invention discloses a method and system for monitoring the state of electrical equipment, which relates to the technical field of electrical equipment monitoring. The method includes collecting data according to the monitored electrical equipment collection items and removing outliers; setting an initial collection frequency and performing linear interpolation and spline interpolation respectively based on the collection items to fill the collection item data; and constructing a long short-term memory network (LSTM) model for time series prediction. The method of the present invention ensures the comprehensive monitoring of the state of electrical equipment and environmental factors through the comprehensive data collection of multi-functional sensors and smart meters. Through the interpolation operations performed by the linear interpolation method and the cubic spline interpolation method, the effect of performing interpolations with different complexities according to different data types is achieved, improving the accuracy and smoothness of the interpolation results, ensuring the smooth transition of electrical data in the time series. By calculating the comprehensive deviation value from the predicted value of the LSTM model, risk assessment calculations can be performed.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical equipment monitoring, and in particular to an electrical equipment state monitoring method and system. Background Art

[0002] Electrical equipment condition monitoring technology usually refers to recording multiple data of the monitored equipment. With the rapid development of information technology and sensor technology, modern electrical equipment condition monitoring systems have gradually introduced advanced technologies such as smart sensors, the Internet of Things (IoT) and big data analysis, making the condition monitoring of electrical equipment more intelligent and efficient, and able to achieve real-time monitoring and remote data collection. In recent years, the application of artificial intelligence (AI) and machine learning (ML) technologies in electrical equipment condition monitoring has begun to show great potential, which can not only improve monitoring accuracy and response speed, but also realize intelligent decision-making support, providing a strong guarantee for the safe and stable operation of the power system.

[0003] Although modern electrical equipment condition monitoring technology has made great progress, the existing monitoring system often faces problems of missing data and noise interference during data collection and processing, depending on the different collection item data and the collection frequency of the collection equipment. Data missing and noise will not only affect the quality of monitoring data, but also lead to a decrease in the accuracy of the prediction model. At the same time, with the increase of detection item data, the existing electrical equipment condition monitoring leads to a complex data volume and inconvenience in comprehensive dynamic adjustment based on the current monitoring risk. It is impossible to dynamically adjust the sampling frequency according to real-time monitoring data and risk scores, which reduces the monitoring efficiency of electrical equipment status. Summary of the invention

[0004] In view of the problems existing in the above-mentioned existing electrical equipment status monitoring methods and systems, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is that in the process of data collection and processing, the existing monitoring system often faces the problems of missing data and noise interference depending on the different collection item data and the collection frequency of the collection equipment. Missing data and noise will not only affect the quality of the monitoring data, but also lead to a decrease in the accuracy of the prediction model. At the same time, with the increase of detection item data, the existing electrical equipment status monitoring leads to a complicated data volume and inconvenience in comprehensive dynamic adjustment based on the current monitoring risk. The sampling frequency cannot be dynamically adjusted according to the real-time monitoring data and risk score, which reduces the monitoring efficiency of the electrical equipment status.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for monitoring the state of electrical equipment, comprising:

[0007] Collect data based on the monitored electrical equipment collection items and remove abnormal values;

[0008] Set the initial acquisition frequency and perform linear interpolation and spline interpolation respectively based on the acquisition items to fill the acquisition item data;

[0009] Construct a long short-term memory network (LSTM) model for time series prediction, comprehensively analyze the deviation between the actual acquisition value and the predicted value, calculate the risk score, set the sampling period, and dynamically adjust the sampling frequency based on the risk score to analyze the efficiency of equipment status monitoring;

[0010] Perform visual display based on the acquisition data and risk score data, and perform secure storage and management of the data.

[0011] As a preferred solution of the electrical equipment status monitoring method described in the present invention, wherein: data acquisition is performed according to the monitored electrical equipment acquisition items, and outliers are removed, including,

[0012] The electrical equipment performs data acquisition according to the acquisition items through a multi-functional sensor and a smart meter. The acquisition items include electrical data and environmental data, where the electrical data includes voltage deviation and frequency deviation, and the environmental data includes temperature, vibration, and noise, as well as the monitoring frequency of the corresponding acquisition items;

[0013] A low-pass filter is used to smooth the data, retain the low-frequency effective signal, and remove the high-frequency noise;

[0014] The three-sigma rule is used to remove outliers. Calculate the mean and standard deviation of the acquisition item data, and remove outliers, which is expressed as:

[0015]

[0016] Where X represents the acquisition item data, μ represents the acquisition item mean, and σ represents the acquisition item standard deviation;

[0017] Values outside the range are regarded as outliers and removed.

[0018] As a preferred solution of the electrical equipment status monitoring method described in the present invention, wherein: the setting of the initial acquisition frequency and performing linear interpolation and spline interpolation respectively based on the acquisition items to fill the acquisition item data includes,

[0019] Take the maximum monitoring frequency of the acquisition item as the initial acquisition frequency;

[0020] Perform time series synchronization based on the acquisition item data, and use the linear interpolation method to fill the data for the environmental data, which is expressed as:

[0021]

[0022] Where X t and X t+nDenote the collected environmental data at events t and t + n, X t+i Denote the interpolated collected environmental data, i represents the interpolation position, and n represents the time step;

[0023] Use the spline interpolation method to fill in the data based on the electrical data of the collection item, expressed as:

[0024] S i (x) = a i + b i (x - x i ) + c i (x - x i ) 2 + d i (x - x i ) 3 ;

[0025] where a i , b i , c i and d i are the coefficients of the cubic spline function, x represents the independent variable time, x i represents the i-th electrical data collection time point, and S i (x) represents the interpolation result of the i-th segmented interval;

[0026] Determine the time points that need to be interpolated through the initial collection frequency, and determine the interval to which the time points that need to be interpolated belong based on the collection time of the electrical data, and calculate the interpolation result through the spline function.

[0027] As a preferred solution of the electrical equipment status monitoring method described in the present invention, where: constructing a long short-term memory network LSTM model for time series prediction, comprehensively analyzing the deviation between the collected actual value and the predicted value and calculating the risk score, including,

[0028] Arrange the interpolated data sequence in chronological order and perform standardization processing on the data;

[0029] Construct a long short-term memory network LSTM model for time series prediction, including an input layer, a hidden layer, and an output layer, where the input layer receives the collected data of the collection item, the hidden layer and the output layer identify the time series through LSTM units, and the output of the last LSTM unit is used as the predicted value for future time;

[0030] Based on the collected item data at time t as the collected actual value, compare it with the predicted value of the long short-term memory network LSTM model at time t, and calculate the absolute value of the difference between the collected actual value and the predicted value as the absolute deviation value;

[0031] Based on the absolute deviation value, comprehensively consider the relative deviation and trend deviation between the collected actual value and the predicted value, expressed as:

[0032]

[0033] Among them, CDI t represents the comprehensive deviation value, AE t represents the absolute deviation value at time t, Y t-1 represents the actual acquisition value at time t - 1, represents the LSTM model prediction value at time t - 1, Y t represents the actual acquisition value at time t, represents the LSTM model prediction value at time t;

[0034] Based on the comprehensive deviation index CDI t Use a linear regression model to judge the risk score, expressed as:

[0035] R t = α + β·CDI t ;

[0036] Among them, α represents the baseline value of the risk score, R t represents the risk score at time t, and β represents the weight of the comprehensive deviation index;

[0037] Analyze the data distribution of the risk scores in the historical data, and use the 95th percentile as the risk threshold;

[0038] If the calculated risk score is greater than or equal to the risk threshold, it is determined as high risk, and the maximum frequency of the monitoring frequency data of the acquisition item is set as the acquisition frequency;

[0039] If the calculated risk score is less than the risk threshold, it is determined as low risk, and the acquisition frequency is dynamically set.

[0040] As a preferred solution of the electrical equipment status monitoring method described in the present invention, wherein: the sampling period is set, and the sampling frequency is dynamically adjusted based on the risk score to analyze the efficiency of equipment status monitoring, including,

[0041] Set the sampling period based on the current acquisition frequency, expressed as:

[0042]

[0043] Among them, T t represents the sampling period of the risk score at time t, F' t represents the acquisition frequency at the current time t;

[0044] Within the sampling period of the risk score, dynamically adjust the sampling frequency based on the risk score R t expressed as:

[0045]

[0046] Among them, F t represents the dynamic sampling frequency, and R min represents the minimum risk score within the sampling period, and R 95 represents the 95th percentile of the risk score within the sampling period, and F max represents the maximum sampling frequency within the sampling period, and F min represents the minimum sampling frequency within the sampling period;

[0047] Based on the calculation of the dynamic sampling frequency to replace the initial sampling frequency, analyze the dynamic energy consumption, and calculate the interpolation of the dynamic energy and the reference energy consumption, which is expressed as:

[0048]

[0049] ΔE = E d - E d ;

[0050] Among them, E d represents the dynamic energy consumption within the sampling period, ΔE represents the energy consumption difference, and E b represents the reference energy consumption within the sampling period;

[0051] Based on the energy consumption difference of the sampling period, analyze the efficiency of the equipment status monitoring, which is expressed as:

[0052]

[0053] Among them, η is the efficiency of the equipment status monitoring, N represents the total number of sampling periods, ΔE i represents the energy consumption difference of the i-th sampling period, and E b,i represents the reference energy consumption of the i-th sampling period.

[0054] As a preferred solution of the electrical equipment status monitoring method described in the present invention, wherein: the visualization display based on the collected data and the risk score data includes,

[0055] Using the Seaborn visualization library in Python, collect and confirm the data to be displayed, including the risk score R t , the sampling frequency F t , the comprehensive deviation value CDI t , the energy consumption difference ΔE, the efficiency η of the status monitoring, the actual collected value Y t and the predicted value of the LSTM model

[0056] Use the Matplotlib tool to plot the change graph of the risk score and the sampling frequency. Display the data change through a line graph and mark the threshold line of the 95th percentile. Then, plot the comparison graph between the actual value and the predicted value, add a legend and annotations, and mark the key data points in the graph. Use the Matplotlib tool for annotation.

[0057] As a preferred solution of the electrical equipment status monitoring method described in the present invention, wherein: the secure storage and management of data refers to using the relational database PostgreSQL for structured data storage and management, using the ETL tool for data storage and management, and regularly storing the risk score R t , the efficiency η of status monitoring, the sampling frequency F t , the comprehensive deviation value CDI t , the energy consumption difference ΔE, the actual collected value Y t and the predicted value of the LSTM model into the database, use the database backup tool mysqldump for regular backup, and use the database security tool AWS RDS to protect the security and access control of the database.

[0058] Another object of the present invention is to provide a system for an electrical equipment status monitoring method, which includes

[0059] A data acquisition module that acquires the monitoring data of electrical equipment through an acquisition device, including electrical data and environmental data;

[0060] An outlier rejection module that calculates the mean and standard deviation of the data and rejects the outlier data that exceeds 3 standard deviation ranges;

[0061] A data interpolation module that fills in the missing data by linear interpolation and spline interpolation based on the initial acquisition frequency of the acquisition items;

[0062] An LSTM model prediction module that constructs a long short-term memory network LSTM model for time series prediction, analyzes the deviation between the actual collected value and the predicted value, and calculates the risk score;

[0063] A frequency adjustment module that sets the sampling period and dynamically adjusts the sampling frequency based on the risk score;

[0064] A visualization display module that performs visualization display based on the acquisition data and the risk score data;

[0065] A data management module that securely stores and manages the data.

[0066] A computer device includes: a memory and a processor; the memory stores a computer program, and when the processor executes the computer program, the steps of the above electrical equipment status monitoring method are implemented.

[0067] A computer-readable storage medium has a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned electrical equipment status monitoring method are implemented.

[0068] The beneficial effects of the present invention are as follows: Through the comprehensive data collection of the multi-functional sensor and the smart meter, the comprehensive monitoring of the electrical equipment status and environmental factors is ensured. Through the interpolation operations using the linear interpolation method and the cubic spline interpolation method, the effect of performing interpolations with different complexities according to different data types is achieved, improving the accuracy and smoothness of the interpolation results and ensuring the smooth transition of electrical data in the time series. By calculating the comprehensive deviation value from the predicted value of the LSTM model, risk assessment calculations can be performed and the sampling frequency can be dynamically adjusted, thereby adjusting the sampling frequency in a timely manner to achieve the effect of adjusting the sampling frequency according to the changes in the timely monitored data, further improving the utilization efficiency of the electrical equipment status monitoring data. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0070] Figure 1 It is a schematic flow chart of the electrical equipment status monitoring method.

[0071] Figure 2 It is a schematic structural diagram of the electrical equipment status monitoring system. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0072] In order to make the above-mentioned objects, features, and advantages of the present invention more obvious and understandable, the following will describe the specific embodiments of the present invention in detail with reference to the drawings in the specification.

[0073] Many specific details are set forth in the following description in order to fully understand the present invention, but the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0074] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation of the present invention. The phrase "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an embodiment that is separate or selectively mutually exclusive with other embodiments.

[0075] Example 1, refer to Figure 1 , this is the first example of the present invention. This example provides a method for monitoring the state of electrical equipment. The method for monitoring the state of electrical equipment includes

[0076] S1, collect data according to the monitored electrical equipment collection items and eliminate outliers;

[0077] Preferably, collect data according to the monitored electrical equipment collection items and eliminate outliers, including

[0078] The electrical equipment collects data according to the collection items through a multi-functional sensor and a smart meter. The collection items include electrical data and environmental data. Among them, the electrical data includes voltage deviation and frequency deviation, and the environmental data includes temperature, vibration, and noise, as well as the monitoring frequency of the corresponding collection items;

[0079] Use a low-pass filter to smooth the data, retain the low-frequency effective signal, and remove the high-frequency noise;

[0080] Use the three-sigma rule to eliminate outliers, calculate the mean and standard deviation of the collection item data, and eliminate outliers, expressed as:

[0081]

[0082] Where X represents the collection item data, μ represents the collection item mean, and σ represents the collection item standard deviation;

[0083] Values outside the range are regarded as outliers and eliminated.

[0084] Through the comprehensive data collection of the multi-functional sensor and the smart meter, the comprehensive monitoring of the electrical equipment state and environmental factors is ensured. The low-pass filter is used for data smoothing processing, effectively removing the high-frequency noise, retaining the useful signal, improving the accuracy of the data. After eliminating the outliers, the mean and standard deviation of the data can more accurately reflect the actual situation, providing a reliable data basis for subsequent data analysis and decision-making. Combining steps such as data collection, smoothing processing, and outlier elimination, the monitoring system can reflect the operating state of the electrical equipment and environmental changes in real time and accurately, improving the stability and reliability of the system.

[0085] S2, set the initial collection frequency and perform linear interpolation and spline interpolation respectively based on the collection items to fill the collection item data;

[0086] Preferably, set the initial collection frequency and perform linear interpolation and spline interpolation respectively based on the collection items to fill the collection item data, including

[0087] Take the maximum monitoring frequency of the collection item as the initial collection frequency;

[0088] Perform time series synchronization based on the collected item data, and use linear interpolation to fill in the environmental data, which is expressed as:

[0089]

[0090] Where X t and X t+n represent the collected environmental data at events t and t + n, X t+i represents the interpolated collected environmental data, i represents the interpolation position, and n represents the time step;

[0091] Use spline interpolation to fill in the electrical data based on the collected items, which is expressed as:

[0092] S i (x) = a i + b i (x - x i ) + c i (x - x i ) 2 + d i (x - x i ) 3 ;

[0093] Where a i , b i , c i and d i are the coefficients of the cubic spline function, x represents the independent variable time, x i represents the i-th electrical data collection time point, S i (x) represents the interpolation result of the i-th segmented interval, where the collection time x i and x i+1 are used as the interval;

[0094] Determine the spline function based on the collected electrical data, which is expressed as:

[0095] S i (x i ) = V i ;

[0096] S i (x i + 1) = V i+1 ;

[0097] Where V i and V i+1 respectively represent the electrical data collection values at collection times x i and x i+1 ;

[0098] Perform first-order derivative continuity and second-order derivative continuity respectively, and form a system of equations, which is expressed as:

[0099] S' i (x i+1 ) = S' i+1 (x i+1 );

[0100] b i + 2c i (x - x i ) + 3d i (x - x i ) 2 = S' i (x);

[0101] S”(x i+1 ) = S” i+1 (x i+1 );

[0102] 2c i + 6d i (x - x i ) = S” i (x);

[0103] where S' i (x) represents the first derivative of the i-th segmented interval, and S” i (x) represents the second derivative of the i-th segmented interval;

[0104] Set natural boundary conditions, expressed as:

[0105] S”0(x0) = 0;

[0106] S” n-1 (x n ) = 0;

[0107] where S”0(x0) represents the second derivative at the start point of the first segmented interval, and S” n-1 (x n ) represents the second derivative at the end point x n of the last segmented interval;

[0108] Use the linear algebra method to solve the system of equations and determine the coefficients of all spline functions;

[0109] Determine the time points to be interpolated based on the initial acquisition frequency, and determine the intervals to which the time points to be interpolated belong based on the acquisition time of the electrical data. Calculate the interpolation results through spline functions.

[0110] The initial acquisition frequency is set through the maximum frequency, achieving the time-series synchronization of the data of each acquisition item, ensuring the consistency of the data on the same time axis, facilitating subsequent data analysis and processing. The environmental data in the acquired data is adapted through the linear interpolation method, and interpolation filling is performed. The interpolation result can be calculated quickly, improving the continuity and integrity of the data. By using the cubic spline interpolation method to fill the electrical data, the accuracy and smoothness of the interpolation result are improved, ensuring the smooth transition of the electrical data in the time series, reducing the interpolation error, ensuring the smoothness and natural transition of the interpolation curve, and enhancing the credibility of the interpolation result. Through data synchronization, smoothing processing, and outlier removal, the data quality of the monitoring system is improved, providing a reliable data basis for subsequent trend analysis and dynamic risk assessment.

[0111] S3. Construct a long short-term memory network (LSTM) model for time series prediction, comprehensively analyze the deviation between the actual acquired value and the predicted value, calculate the risk score, set the sampling period, and dynamically adjust the sampling frequency based on the risk score to analyze the efficiency of equipment status monitoring;

[0112] Preferably, construct a long short-term memory network (LSTM) model for time series prediction, comprehensively analyze the deviation between the actual acquired value and the predicted value, and calculate the risk score, including

[0113] Arrange the interpolated data sequence in chronological order and standardize the data;

[0114] Construct a long short-term memory network (LSTM) model for time series prediction, including an input layer, a hidden layer, and an output layer. The input layer receives the acquired data of the acquisition item, and the hidden layer and the output layer identify the time series through LSTM units, and the output of the last LSTM unit is used as the predicted value for future time;

[0115] Use the training set and adopt the mean square error loss function to calculate the loss, expressed as:

[0116]

[0117] where k represents the total number of samples, Y i represents the actual acquired value, represents the predicted value of the LSTM model;

[0118] Through the Adam optimizer and using the gradient descent method to iteratively optimize the model parameters, when the loss of the long short-term memory network (LSTM) model no longer decreases significantly during continuous iteration, stop the iteration, output the model parameters, and update the long short-term memory network (LSTM) model;

[0119] The data of the acquisition item at time t is used as the actual acquisition value, compared with the predicted value of the long short-term memory network LSTM model at time t, and the absolute value of the difference between the actual acquisition value and the predicted value is calculated as the absolute deviation value;

[0120] Based on the absolute deviation value, comprehensively considering the relative deviation and trend deviation between the actual acquisition value and the predicted value, it is expressed as:

[0121]

[0122] Among them, CDI t represents the comprehensive deviation value, AE t represents the absolute deviation value at time t, Y t-1 represents the actual acquisition value at time t-1, represents the predicted value of the LSTM model at time t-1, Y t represents the actual acquisition value at time t, represents the predicted value of the LSTM model at time t;

[0123] Based on the comprehensive deviation index CDI t Use a linear regression model to judge the risk score, which is expressed as:

[0124] R t =α + β·CDI t ;

[0125] Among them, α represents the baseline value of the risk score, R t represents the risk score at time t, and β represents the weight of the comprehensive deviation index;

[0126] Use the training set to optimize α and β using the least squares method, and define the optimization objective function, which is expressed as:

[0127]

[0128] Among them, G represents the output of the objective function, T represents the total number of samples, R' t represents the actual value;

[0129] Determine α and β based on minimizing the output of the objective function;

[0130] Analyze the data distribution of the risk scores in the historical data, and use the 95th percentile as the risk threshold;

[0131] If the calculated risk score is greater than or equal to the risk threshold, it is determined as high risk, and the maximum frequency of the acquisition item monitoring frequency data is set as the acquisition frequency;

[0132] If the calculated risk score is less than the risk threshold, it is determined as low risk, and the acquisition frequency is dynamically set.

[0133] By constructing and training an LSTM model, long-term dependencies in time series data can be effectively captured. Based on the interpolated acquisition data, the status of electrical equipment can be predicted. At the same time, based on the actual acquisition data of electrical equipment, by calculating the comprehensive deviation value from the predicted value of the LSTM model, the accuracy of the LSTM model prediction can be further analyzed, and risk assessment calculations can be performed. When the comprehensive deviation is large, it can be judged that the data acquisition risk of the acquisition equipment for electrical equipment increases, while when the comprehensive deviation is small, it can be judged that the risk decreases. Thus, the status of electrical equipment can be monitored only from the interpolated acquisition data and the LSTM model. Through linear regression models and historical data analysis, risk scoring thresholds are scientifically set to ensure the rationality and accuracy of risk assessment.

[0134] Furthermore, set the sampling period and dynamically adjust the sampling frequency based on the risk score to analyze the efficiency of equipment status monitoring, including

[0135] Set the sampling period based on the current acquisition frequency, expressed as:

[0136]

[0137] where T t represents the sampling period of the risk score at time t, and F' t represents the acquisition frequency at the current time t;

[0138] Within the sampling period of the risk score, dynamically adjust the sampling frequency based on the risk score R t , expressed as:

[0139]

[0140] where F t represents the dynamic sampling frequency, R min represents the minimum risk score within the sampling period, R 95 represents the 95th percentile of the risk score within the sampling period, F max represents the maximum sampling frequency within the sampling period, and F min represents the minimum sampling frequency within the sampling period;

[0141] Replace the initial sampling frequency based on the calculation of the dynamic sampling frequency, analyze the dynamic energy consumption, and calculate the interpolation between the dynamic energy and the reference energy consumption, expressed as:

[0142]

[0143] ΔE = E d -E d ;

[0144] where E drepresents the dynamic energy consumption within the sampling period, ΔE represents the energy consumption difference, and E b represents the reference energy consumption within the sampling period;

[0145] Analyze the efficiency of equipment condition monitoring based on the energy consumption difference within the sampling period, which is expressed as:

[0146]

[0147] where η is the efficiency of equipment condition monitoring, N represents the total number of sampling periods, and E d,i represents the dynamic energy consumption in the i-th sampling period, and E b,i represents the reference energy consumption in the i-th sampling period;

[0148] When calculating the dynamic energy consumption within the sampling period, in the calculation of the dynamic energy consumption formula, R t is a dynamically changing value that reflects the risk level of the equipment state at time t. To better capture the impact of the risk score on energy consumption, using a logarithmic function can compress a wide range of values into a smaller range, making the impact of high risk scores on energy consumption more significant, while the impact of low risk scores is appropriately weakened. This can more realistically reflect the energy consumption changes of the equipment at different risk levels, and at the same time avoid the mathematical problems caused by the logarithmic function at zero or negative values, thus ensuring the stability and accuracy of energy consumption calculation;

[0149] Then, by multiplying with F t to comprehensively consider the impact of the dynamic sampling frequency on energy consumption. Considering that the adjustment of the dynamic sampling frequency depends on the sampling period, integrating can accumulate the energy consumption at each time point within the entire sampling period, thereby obtaining a total energy consumption value. This method can better reflect the cumulative energy consumption situation in actual operation, so that the calculation of dynamic energy consumption not only considers the energy consumption at a certain moment, but also considers the change trend of energy consumption throughout the period;

[0150] Compared with the calculation of dynamic energy consumption, the reference energy consumption formula extracts the average value of the maximum and minimum dynamic sampling frequencies as the reference energy consumption within the sampling period, which can better reflect the trend change of dynamic energy consumption calculation in the process of calculating the energy consumption difference. In the calculation of the efficiency of equipment condition monitoring, by accumulating the ratio of the saved or increased value of energy consumption to the reference energy consumption, the efficiency of equipment condition monitoring is comprehensively evaluated.

[0151] By setting the sampling period based on the current acquisition frequency, the system can flexibly adapt to different changes in the acquisition frequency, improve the efficiency and flexibility of data acquisition, dynamically adjust the sampling period according to actual needs, optimize the use of data acquisition resources, avoid unnecessary high-frequency sampling, save system resources. By dynamically adjusting the sampling frequency, the system can respond in real time to changes in the risk score, increase the sampling frequency in high-risk situations to ensure timely capture of changes in the device status. When the risk score is high, increase the sampling frequency to improve the density of data acquisition, ensure the accuracy and timeliness of monitoring data, and help detect potential faults in a timely manner. When the risk score is low, reduce the sampling frequency, optimize the use of data acquisition resources, reduce data redundancy, and reduce the system burden. By dynamically adjusting the sampling frequency, ensure the stable operation of the system under different risk states, and improve the reliability and stability of the monitoring system.

[0152] S4. Based on the acquired data and risk score data, perform visual display, and securely store and manage the data;

[0153] Preferably, the visual display based on the acquired data and risk score data includes,

[0154] Using the Seaborn visualization library in Python, collect and confirm the data to be displayed, including the risk score R t , the sampling frequency F t , the comprehensive deviation value CDI t , the actual acquisition value Y t and the predicted value of the LSTM model

[0155] Use the Matplotlib tool to draw a graph of the risk score and the change in the sampling frequency, display the data change through a line chart, and mark the threshold line of the 95th percentile. Then draw a comparison chart of the actual value and the predicted value, add a legend and annotations to help understand the data change, mark the key data points in the chart, and use the Matplotlib tool for annotation.

[0156] By showing the change in the sampling frequency, the dynamic adjustment strategy of the system under different risk states can be intuitively understood. By marking the risk threshold line of the 95th percentile, it is convenient to identify and distinguish high-risk and low-risk situations, providing a basis for decision-making. By intuitively showing the comparison between the actual value and the predicted value, it is convenient to evaluate the prediction performance and accuracy of the LSTM model. Marking the key data points in the chart highlights the outliers or important change trends, helping to deeply analyze the data. By adding a legend and annotations, it helps to understand the content of the chart and the data change, and improves the readability and information transmission effect of the visualization chart.

[0157] Further, the secure storage and management of data refers to using the relational database PostgreSQL for structured data storage and management, using ETL tools for data storage and management, and regularly storing the risk score R t , sampling frequency F t , comprehensive deviation value CDI t , actual acquisition value Y t and the predicted value of the LSTM model into the database, using the database backup tool mysqldump for regular backup, and using the database security tool AWS RDS to protect the security and access control of the database.

[0158] By using PostgreSQL for structured data storage and management, the organization and query efficiency of data are improved, facilitating multi-dimensional data analysis and mining. Through the ETL tool, the automated processing of data extraction, transformation, and loading is realized, reducing manual intervention, improving data processing efficiency and data quality. Through regular backup, the security and recoverability of data are ensured, and the database can be restored in a timely manner in case of failures or data loss, guaranteeing data security. Through the security management function of AWS RDS, the access permissions and data encryption of the database are set to protect data privacy and security, and improve the transparency and security of data management.

[0159] Example 2, referring to Figure 2 , is the second embodiment of the present invention. This embodiment is different from the previous one and provides a system for an electrical equipment status monitoring method, including

[0160] a data acquisition module that acquires the monitoring data of electrical equipment through acquisition devices, including electrical data and environmental data;

[0161] an outlier rejection module that calculates the mean and standard deviation of the data and rejects the outlier data exceeding 3 standard deviation ranges;

[0162] a data interpolation module that fills in the missing data with linear interpolation and spline interpolation based on the initial acquisition frequency of the acquisition items;

[0163] an LSTM model prediction module that constructs a long short-term memory network LSTM model for time series prediction, analyzes the deviation between the actual acquisition value and the predicted value, and calculates the risk score;

[0164] a frequency adjustment module that sets the sampling period and dynamically adjusts the sampling frequency based on the risk score;

[0165] a visualization display module that performs visualization display based on the acquisition data and risk score data;

[0166] a data management module that securely stores and manages the data.

[0167] When the above-mentioned function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0168] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in conjunction with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0169] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts (electronic devices) having one or more wirings, portable computer disk cartridges (magnetic devices), random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memories), fiber optic devices, and portable compact disc read-only memories (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or otherwise processing it as appropriate, and then storing it in a computer memory.

[0170] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following techniques known in the art or a combination thereof can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0171] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for monitoring the state of an electrical device, characterized in that: including collecting data according to the collected items of the monitored electrical equipment and removing outliers setting an initial sampling frequency and performing linear interpolation and spline interpolation respectively based on the collected items to fill in the collected item data constructing a long short-term memory network (LSTM) model for time series prediction, comprehensively analyzing the deviation between the actual collected value and the predicted value and calculating a risk score, setting a sampling period, and dynamically adjusting the sampling frequency based on the risk score to analyze the efficiency of equipment status monitoring performing visual display based on the collected data and the risk score data, and securely storing and managing the data the setting of the sampling period and the dynamic adjustment of the sampling frequency based on the risk score to analyze the efficiency of equipment status monitoring including setting a sampling period based on the current sampling frequency, expressed as where T t represents the sampling period of the risk score at time t, and F' t represents the acquisition frequency at the current time t; During the sampling period of the risk score, based on the risk score R t for dynamically adjusting the sampling frequency, expressed as: where F t represents the dynamic sampling frequency, R min represents the minimum risk score within the sampling period, R 95 represents the 95th percentile of the risk score within the sampling period, F max represents the maximum sampling frequency within the sampling period, F min represents the minimum sampling frequency within the sampling period; replacing the initial sampling frequency based on the calculation of the dynamic sampling frequency, analyzing the dynamic energy consumption, and calculating the interpolation between the dynamic energy and the baseline energy consumption, expressed as ΔE = E d - E d ; Among them, E d represents the dynamic energy consumption within the sampling period, ΔE represents the energy consumption difference, and E b represents the reference energy consumption within the sampling period; analyzing the efficiency of equipment status monitoring based on the energy consumption difference of the sampling period, expressed as where η is the efficiency of device status monitoring, N represents the total number of sampling periods, and ΔE i represents the energy consumption difference in the i-th sampling period, and E b,i represents the reference energy consumption in the i-th sampling period.

2. The electrical equipment status monitoring method according to claim 1, characterized in that: the collecting of data according to the collected items of the monitored electrical equipment and the removal of outliers, including the electrical equipment collects data according to the collected items through a multi-functional sensor and a smart meter. The collected items include electrical data and environmental data. The electrical data includes voltage deviation and frequency deviation, and the environmental data includes temperature, vibration, and noise, as well as the monitoring frequency of the corresponding collected items using a low-pass filter to smooth the data, retaining the low-frequency effective signal and removing the high-frequency noise using the three-sigma rule to remove outliers, calculating the mean and standard deviation of the collected item data, and removing outliers, expressed as where X represents the collected item data, μ represents the collected item mean, and σ represents the collected item standard deviation regarding the values outside the range as outliers and removing them 3. The electrical equipment status monitoring method according to claim 2, characterized in that: the setting of the initial sampling frequency and the performing of linear interpolation and spline interpolation respectively based on the collected items to fill in the collected item data, including taking the maximum monitoring frequency of the collected items as the initial sampling frequency performing time series synchronization based on the collected item data and using linear interpolation to fill in the environmental data, expressed as where X t and X t+n represent the acquired environmental data at events t and t + n, and X t+i represents the interpolated acquired environmental data, i represents the interpolation position, and n represents the time step; using spline interpolation to fill in the electrical data based on the collected items, expressed as S i f(x) = a i + b i (x - x i ) + c i (x - x i ) 2 + d i (x - x i ) 3 ; where a i , b i , c i and d i are the coefficients of the cubic spline function, x represents the independent variable time, and x i represents the i-th electrical data acquisition time point, and S i (x) represents the interpolation result of the i-th segmented interval; determining the time points that need interpolation through the initial sampling frequency, and determining the interval to which the time points that need interpolation belong based on the collection time of the electrical data, and calculating the interpolation result through the spline function 4. The electrical equipment status monitoring method according to claim 3, characterized in that: the constructing of a long short-term memory network (LSTM) model for time series prediction, comprehensively analyzing the deviation between the actual collected value and the predicted value and calculating a risk score, including arranging the interpolated data sequence in chronological order and performing standardization processing on the data constructing a long short-term memory network (LSTM) model for time series prediction, including an input layer, a hidden layer, and an output layer. The input layer receives the collected data of the collected items, and the hidden layer and the output layer identify the time series through LSTM units, and the output of the last LSTM unit is used as the predicted value for future time using the collected item data at time t as the actual collected value, comparing it with the predicted value of the long short-term memory network (LSTM) model at time t, and calculating the absolute value of the difference between the actual collected value and the predicted value as the absolute deviation value Based on the absolute deviation value, comprehensively considering the relative deviation and trend deviation between the collected actual value and the predicted value, it is expressed as: Among them, CDI t represents the comprehensive deviation value, AE t represents the absolute deviation value at time t, Y t-1 represents the actual acquisition value at time t - 1, represents the LSTM model prediction value at time t - 1, Y t represents the actual acquisition value at time t, represents the LSTM model prediction value at time t; Based on the comprehensive deviation index CDI t Use a linear regression model to judge the risk score, expressed as: R t = α + β·CDI t ; where α represents the baseline value of the risk score, R t represents the risk score at time t, and β represents the weight of the comprehensive deviation index; Analyze the data distribution of the risk scores in the historical data, and use the 95th percentile as the risk threshold; If the calculated risk score is greater than or equal to the risk threshold, it is determined as high risk, and the maximum frequency of the monitoring frequency data of the collection item is set as the collection frequency; If the calculated risk score is less than the risk threshold, it is determined as low risk, and the collection frequency is dynamically set.

5. The electrical equipment status monitoring method according to claim 4, characterized in that: The visualization display based on the collected data and the risk score data includes: Using the Seaborn visualization library in Python, collect and confirm that the data to be displayed includes the risk score R t , sampling frequency F t , comprehensive deviation value CDI t , energy consumption difference ΔE, efficiency η of status monitoring, actual collected value Y t and the predicted value of the LSTM model Use the Matplotlib tool to draw the change graph of the risk score and the sampling frequency, display the data change through a line graph, and mark the threshold line of the 95th percentile. Then draw the comparison graph between the actual value and the predicted value, add legends and annotations, mark the key data points in the chart, and use the Matplotlib tool for annotation.

6. The electrical equipment status monitoring method according to claim 5, characterized in that: The secure storage and management of data refer to using the relational database PostgreSQL for structured data storage and management, using ETL tools for data storage and management, and regularly storing the risk score R t , the efficiency η of status monitoring, and the sampling frequency F t , the comprehensive deviation value CDI t , the energy consumption difference ΔE, the actual acquisition value Y t , and the prediction value of the LSTM model into the database, using the database backup tool mysqldump for regular backups, and using the database security tool AWS RDS to protect the security and access control of the database.

7. A system based on the electrical equipment status monitoring method according to any one of claims 1-6, characterized in that: Including: A data collection module that collects the monitoring data of electrical equipment through a collection device, including electrical data and environmental data; An outlier removal module that calculates the mean and standard deviation of the data and removes the abnormal data exceeding 3 standard deviation ranges; A data interpolation module that fills the missing data by linear interpolation and spline interpolation based on the initial collection frequency of the collection item; An LSTM model prediction module that constructs a long short-term memory network LSTM model for time series prediction, analyzes the deviation between the collected actual value and the predicted value, and calculates the risk score; A frequency adjustment module that sets the sampling period and dynamically adjusts the sampling frequency based on the risk score; A visualization display module that performs visualization display based on the collected data and the risk score data; A data management module that securely stores and manages the data.

8. A computer device, comprising: A memory and a processor; The memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of the electrical equipment status monitoring method according to any one of claims 1 to 6 are implemented.

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

Images