Back clip-based power equipment state information acquisition method and system

Abstract

The present application belongs to the technical field of power equipment monitoring, and particularly relates to a power equipment state information acquisition method and system based on a back clamp, which is characterized by the following steps: connecting the back clamp device with the interface of the power equipment to establish a data communication link; collecting the running state data of the power equipment in real time, including voltage, current and temperature parameters; pre-processing the collected data to filter out noise and standardize; transmitting the processed data to a cloud server through a wireless communication module; and analyzing the data based on a preset algorithm by the server to generate a device state report and feed back to the terminal. The present application adopts a back clamp design, is convenient to install, has strong compatibility, and can be widely applied to the state monitoring of various power equipment to improve the operation and maintenance efficiency and reduce the labor cost.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power equipment monitoring, and particularly relates to a power equipment state information acquisition method and system based on a back clamp. BACKGROUND

[0002] In the power system, ensuring the stable operation of power equipment is crucial for the reliability of the power grid. With the growth of power demand and the expansion of equipment scale, accurately acquiring power equipment state information has become a key link.

[0003] In practical applications, power equipment state information acquisition faces many challenges. On the one hand, traditional monitoring methods rely on manual inspection, which is not only inefficient but also prone to missed or false inspections, making it difficult to monitor the equipment operating state in real time. For example, in some remote substation areas, manual inspection cycles are long, and equipment failures cannot be discovered and handled in a timely manner, affecting power supply. On the other hand, existing monitoring systems have deficiencies in data collection, transmission, and analysis. Some equipment data collection interfaces are not unified, making it difficult to achieve interconnection between devices; data transmission is susceptible to interference, resulting in information loss or errors; and the analysis method is simple, making it impossible to accurately predict potential equipment failures.

[0004] Currently, although some technologies attempt to solve these problems, they all have limitations. Some technologies only focus on single device or single parameter monitoring, lacking systematicness; some monitoring devices are complex to install and costly, making it difficult to promote on a large scale. Therefore, there is an urgent need for an efficient and convenient power equipment state information acquisition method and system to meet the growing monitoring needs of the power system. SUMMARY

[0005] The main purpose of the present application is to provide a power equipment state information acquisition method and system based on a back clamp. By designing a convenient back clamp device, fast and stable connection with power equipment and real-time acquisition of multiple parameters are achieved; by using various data processing and communication technologies, accurate data transmission and secure interaction are ensured; by means of the powerful analysis capability of the cloud and a reasonable health evaluation mechanism, the operating state of the power equipment is accurately evaluated and trend predicted, ultimately establishing an efficient, intelligent, and low-cost power equipment state monitoring system, improving the reliability and stability of the power system operation, and meeting the demand for intelligent management of power operation and maintenance.

[0006] To this end, the technical solution adopted by the present application is as follows:

[0007] A power equipment state information acquisition method based on a back clamp, comprising the following steps:

[0008] S1, through the physical interface of the back clamp device and the standardized data interface of the power equipment, a bidirectional data communication link is established;

[0009] S2, collecting running state data of the power equipment in real time, the running state data at least including voltage, current, temperature, running frequency and equipment identification code;

[0010] S3, pre-processing the collected running state data, the pre-processing including signal filtering, AD conversion and data normalization;

[0011] S4, uploading the pre-processed running state data to a cloud server through a wireless communication module built in the back clamp, the wireless communication module supporting at least one LPWAN protocol;

[0012] S5, after the cloud server receives and stores the running state data, generating a power equipment health state evaluation report based on a time series analysis model and a preset threshold comparison algorithm; the power equipment state evaluation report including current power equipment health state and future power equipment health state;

[0013] S6, the cloud server pushes the power equipment state evaluation report and corresponding running state data to a user terminal.

[0014] Further, the physical interface in step S1 runs a multi-protocol adaptive interface engine, dynamically adapting communication protocols of different power equipment, including the following steps:

[0015] 1) Protocol sniffing: the back clamp device automatically scans the power equipment interface after power-on, detects the communication protocols supported by the current equipment, including RS-485, Modbus and IEC 61850;

[0016] 2) Protocol loading: matching the corresponding protocol analysis module from the protocol library built in the back clamp device, dynamically loading to the memory;

[0017] 3) Link keep-alive: the back clamp device periodically sends heartbeat frames and monitors the response delay, if timeout, triggers re-linking.

[0018] Further, the voltage and current in step S2 are collected by high-precision voltage transformers and Hall effect current sensors built in the back clamp respectively;

[0019] The temperature is collected by a temperature sensor installed on the surface of the back clamp in contact with the power equipment;

[0020] The running frequency is obtained by zero-crossing detection on the voltage signal;

[0021] The equipment identification code is read by the RFID module of the back clamp to obtain the unique identification code of the power equipment;

[0022] The acquisition frequency of the back clamping device can be dynamically adjusted. The programmable sampling controller built-in the back clamping device automatically configures the acquisition frequency to the corresponding value in the range of 1Hz~10kHz according to the pre-stored device type and monitoring level mapping table.

[0023] Further, the signal filtering in step S3 includes:

[0024] Kalman filtering is applied to voltage and current signals to eliminate random noise;

[0025] Sliding average filtering is applied to temperature signals to smooth fluctuations;

[0026] Power frequency notch filtering is applied to operating frequency signals to suppress fundamental interference;

[0027] Further, the AD conversion is realized by the programmable ADC module built-in the back clamping device, which converts the operating state data from analog signal data to digital signal data, i.e. discrete data that can be processed by a computer;

[0028] The data normalization includes dividing the actually collected voltage data and current data by the rated voltage and rated current of the power equipment respectively to obtain unitless voltage and current standard values;

[0029] Linear scaling is applied to temperature data to map to the range of 0~1;

[0030] The operating frequency is divided by the rated operating frequency of the current equipment to obtain a standard frequency.

[0031] Further, the wireless communication module in step S4 adopts an abnormal priority upload mechanism during data transmission, which divides the operating state data into abnormal data and normal data, and the abnormal data is uploaded to the cloud server in priority;

[0032] At least one of the voltage data, current data or temperature data in the operating state data exceeds the pre-set threshold range, which is the abnormal data;

[0033] The wireless communication module adopts AES-128 encryption algorithm to encrypt the operating state data during wireless transmission, and verifies the identity legality of the back clamping device through a two-way authentication mechanism with the cloud server;

[0034] Further, the time series analysis model in step S5 adopts a pre-trained long short-term memory network (LSTM) to predict the operating state trend of the power equipment, and adjusts the model parameters of the LSTM according to the stored operating state data at a fixed period;

[0035] The model parameter adjustment of the LSTM includes the following steps:

[0036] The running state data stored in the cloud server is extracted to construct a data set, and the data in a sliding window with a time length of is used as input features, and the data in the next time steps is used as labels; let the current time be , the input features and the labels are represented as follows:

[0037]

[0038]

[0039] wherein, is the voltage data in the running state data, is the current data, is the temperature data, is the running frequency data; the input features and the labels form a data, and the above process is repeated until the number of data in the data set reaches .

[0040] The training loss is calculated: the loss function uses the mean square error to measure the prediction deviation:

[0041]

[0042] wherein, is the true value of the th data; is the predicted value output by the LSTM model according to the input features; is the number of data sets;

[0043] The model parameter is updated: the LSTM model parameter matrix is updated by using the stochastic gradient descent with momentum:

[0044]

[0045] wherein, is the learning rate, is the momentum coefficient; is the gradient of the loss function to , and is the training round.

[0046] Further, the health state of the power equipment is divided into normal, low risk, medium risk and high risk;

[0047] The threshold comparison algorithm includes voltage threshold, current threshold, temperature threshold and running frequency threshold;

[0048] The power equipment health state is normal, that is, the voltage, current, temperature and operating frequency in the operating state data are within the threshold range;

[0049] The power equipment health state is low risk, that is, one of the voltage, current, temperature and operating frequency in the operating state data is not within the threshold range;

[0050] The power equipment health state is medium risk, that is, two of the voltage, current, temperature and operating frequency in the operating state data are not within the threshold range;

[0051] The power equipment health state is high risk, that is, three or more of the voltage, current, temperature and operating frequency in the operating state data are not within the threshold range;

[0052] The current power equipment health state is determined by the operating state data at the current time according to a threshold comparison algorithm;

[0053] The future power equipment health state is determined by the future power equipment operating state output by the time series analysis model, and then according to a threshold comparison algorithm;

[0054] A back clamp-based power equipment state information acquisition system, comprising:

[0055] The back clamp acquisition module: integrated sensors and microcontrollers, can be connected with power equipment, collect operating state data and perform preliminary processing.

[0056] The communication module: supports wired and LPWAN wireless communication, has data transmission and identity verification functions, and can safely upload the data processed by the back clamp acquisition module to the cloud server.

[0057] The cloud analysis module: provided with data storage, algorithm analysis and alarm generation units. Store the received data, evaluate the equipment health state with specific algorithms, and generate evaluation reports and alarm information.

[0058] The user terminal module: provides a visual interface, receives and displays the equipment state reports and alarm information pushed by the cloud analysis module, and facilitates users to master the equipment situation.

[0059] Compared with the prior art, the advantages of the present application are:

[0060] The back clamp device designed by the present application can quickly connect with power equipment and stably collect operating state data. Compared with traditional manual inspection, the efficiency of data collection is greatly improved, and errors that may occur in manual operation are avoided. The buckle structure design of the back clamp and the integration of multiple sensors realize the real-time synchronous collection of multiple parameters, providing a comprehensive and accurate data basis for the state monitoring of power equipment.

[0061] The application can accurately evaluate the health status of power equipment and effectively predict the future state by means of a cloud-based time sequence analysis model and a threshold comparison algorithm. Traditional methods can only make simple state judgments and are difficult to estimate potential risks of the equipment. The application can discover equipment abnormalities in advance through deep learning algorithm learning on a large amount of historical data, provide sufficient time for operation and maintenance personnel to handle, ensure stable operation of power equipment, and improve the reliability of the power system. BRIEF DESCRIPTION OF DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0063] Figure 1 The method flowchart of the present application is shown in the figure.

[0064] Figure 2 The step S2 flowchart of the present application is shown in the figure.

[0065] Figure 3 The step S5 flowchart of the present application is shown in the figure. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0067] Embodiment 1: Please refer to Figure 1 The back clamping-based power equipment state information acquisition method described in the embodiment includes the following steps:

[0068] S1, through the physical interface of the back clamping device and the standardized data interface of the power equipment, a bidirectional data communication link is established;

[0069] In the embodiment, the back clamping device is fixed to the power equipment through the buckle structure.

[0070] The buckle structure comprises: elastic clamping jaws: symmetrically arranged on both sides of the back clamp body, the end of the clamping jaw is provided with a barb for buckling the flange of the shell of the power equipment; locking mechanism: using a manual knob to adjust the clamping force of the clamping jaw to ensure the close docking of the back clamp and the power equipment interface; anti-displacement design: the inner side of the clamping jaw is provided with anti-slip lines and silica gel pads to prevent displacement caused by vibration;

[0071] In this embodiment, the back clamp device is built-in with a central processing unit, the physical interface runs a multi-protocol adaptive interface engine, dynamically adapts the communication protocols of different power equipment, and the specific process of the multi-protocol adaptive interface includes:

[0072] 1) Protocol sniffing: first, perform frequency band scanning control, the central processing unit sends instructions to the radio frequency module of the physical interface, so that it performs comprehensive scanning on the power equipment interface at different communication frequencies;

[0073] Secondly, signal feature extraction: the central processing unit uses fast Fourier transform to extract features from the collected power equipment interface signals;

[0074] Finally, fuzzy matching: the central processing unit uses a fuzzy matching algorithm to judge the protocol, the fuzzy matching algorithm uses an improved Levenshtein distance-weighted cosine similarity composite algorithm, and the specific steps are as follows:

[0075] Frame structure comparison: calculate the Levenshtein distance between the interface signal data frame and the template frame in the protocol feature library, which is represented as follows:

[0076]

[0077] Wherein, is the Levenshtein distance function, represents the string corresponding to the interface signal data frame, and B represents the string corresponding to the template frame in the protocol feature library; represents the string from the first character to the character substring, represents the string from the first character to the character substring; is an indicator function;

[0078] b) Feature vector similarity: extract the following features of the interface signal to construct a vector: preamble length, byte interval standard deviation and CRC polynomial; calculate the cosine similarity with the protocol library:

[0079]

[0080] Wherein, is a vector constructed from the features of the interface signal, is a vector constructed from the corresponding protocol features in the protocol library;

[0081] c) Comprehensive score:

[0082]

[0083] wherein, is The maximum value is obtained in the protocol feature library, and the protocol feature library is selected as the final result of the match. The highest protocol is the final result of the match.

[0084] 2) Protocol loading

[0085] Match the corresponding protocol analysis module from the protocol library built in the back clamp device, and dynamically load it into the memory.

[0086] 3) Link keep-alive

[0087] After the back clamp device establishes a bidirectional data communication link, the central processing unit adopts an adaptive link keep-alive mechanism to ensure the stability of the communication link:

[0088] The central processing unit monitors the quality of the communication link and the working state of the power equipment in real time; when the central processing unit detects that the heartbeat frame timeout or response delay is abnormal, it will immediately trigger the fault diagnosis mechanism; the fault diagnosis mechanism is to comprehensively check the communication link, including the connection state of the physical interface, the working state of the protocol analysis module; according to the inspection results including module loading failure and module running abnormally, the central processing unit will take the strategy of reloading the protocol analysis module, adjusting the communication parameters and re-establishing the connection.

[0089] S2, real-time acquisition of power equipment running state data, the running state data at least includes voltage, current, temperature, running frequency and equipment identification code, reference Figure 2 as shown;

[0090] In this embodiment, the voltage and current are collected by the high-precision voltage transformer and Hall effect current sensor built in the back clamp, respectively;

[0091] The temperature is collected by a temperature sensor installed on the surface of the back clamp in contact with the power equipment;

[0092] The running frequency is obtained by zero-crossing detection of the voltage signal;

[0093] The equipment identification code is read by the RFID module of the back clamp to obtain the unique identification code of the power equipment;

[0094] The acquisition frequency of the back clamp device can be dynamically adjusted. According to the pre-stored device type and monitoring level mapping table, the programmable sampling controller built in the back clamp device automatically configures the acquisition frequency to a corresponding value in the range of 1 Hz to 10 kHz.

[0095] S3, preprocessing the collected operating state data, the preprocessing including signal filtering, AD conversion and data normalization;

[0096] In this embodiment, the signal filtering includes:

[0097] Kalman filtering is applied to voltage and current signals to eliminate random noise;

[0098] Sliding average filtering is used for temperature signals to smooth fluctuations;

[0099] Power frequency notch filtering is performed on the operating frequency signal to suppress fundamental interference;

[0100] In this embodiment, Kalman filtering, sliding average filtering and power frequency notch filtering are all prior art, and the technical details are not described;

[0101] The AD conversion is realized by a programmable ADC module built in the back clamp device, which converts the operating state data from analog signal data to digital signal data, i.e. discrete data that can be processed by a computer;

[0102] The ADC module can be one of successive approximation ADC, sigma-delta ADC and parallel comparison ADC;

[0103] The data normalization includes dividing the actually collected voltage data and current data by the rated voltage and rated current of the power equipment respectively to obtain unitless voltage and current unit values;

[0104] Linear scaling is performed on the temperature data to map to 0-1; the linear scaling formula is as follows:

[0105]

[0106] Wherein, W is the temperature data, is the minimum value of the temperature data, is the maximum value of the temperature data;

[0107] The operating frequency is divided by the rated operating frequency of the current equipment to obtain a unit frequency.

[0108] S4, uploading the preprocessed operating state data to the cloud server through the wireless communication module built in the back clamp, the wireless communication module supporting at least one LPWAN protocol;

[0109] In the embodiment, the wireless communication module adopts an abnormal priority uploading mechanism during data transmission, divides the running state data into abnormal data and normal data, and uploads the abnormal data to the cloud server in priority;

[0110] At least one of the voltage data, the current data or the temperature data in the running state data exceeds a preset threshold range, which is the abnormal data;

[0111] The wireless communication module adopts an AES-128 encryption algorithm to encrypt the running state data during wireless transmission, and verifies the identity legality of the back clamping device through a two-way authentication mechanism with the cloud server;

[0112] The two-way authentication mechanism comprises:

[0113] 1) The back clamping device confirms the authenticity by verifying the cloud server certificate chain;

[0114] 2) The cloud server confirms the authenticity by comparing the hardware fingerprint and dynamic signature of the back clamping device;

[0115] 3) After authentication, an AES-128 encryption channel is established.

[0116] S5, after the cloud server receives and stores the running state data, generates a power equipment health state evaluation report based on a time sequence analysis model and a preset threshold comparison algorithm; the power equipment state evaluation report includes the current power equipment health state and the future power equipment health state, as shown in Figure 3 ;

[0117] In the embodiment, the time sequence analysis model adopts a pre-trained long short-term memory network (LSTM) to predict the running state trend of the power equipment, and adjusts the model parameters of the LSTM according to the stored running state data at a fixed period;

[0118] The steps of adjusting the model parameters of the LSTM include:

[0119] Constructing a data set: from the running state data stored in the cloud server, extracting sliding window data with a time length of as input features, and taking the data of time steps after the sliding window as labels; let the current time be , then the input features and the labels are represented as follows:

[0120]

[0121]

[0122] wherein, V is voltage data in the operating state data, I is current data, T is temperature data, F is operating frequency data; the above input features and labels constitute a piece of data, and the above process is repeated until the number of data in the data set reaches ;

[0123] Calculate the training loss: loss function The mean square error is used to measure the prediction deviation:

[0124]

[0125] wherein, Y is the true value of the th data; Y is the predicted value output by the LSTM model according to the input features; N is the number of data sets;

[0126] Model parameter update: update the LSTM model parameter matrix by the stochastic gradient descent with momentum:

[0127]

[0128] wherein, is the learning rate, is the momentum coefficient; is the gradient of the loss function to , and is the training round.

[0129] In the embodiment, the health state of the power equipment is divided into normal, low risk, medium risk and high risk;

[0130] The threshold comparison algorithm includes voltage threshold, current threshold, temperature threshold and operating frequency threshold;

[0131] The health state of the power equipment is normal, that is, the voltage, current, temperature and operating frequency in the operating state data are all within the threshold range;

[0132] The health state of the power equipment is low risk, that is, among the operating state data, one of the voltage, current, temperature and operating frequency is not within the threshold range;

[0133] The health state of the power equipment is medium risk, that is, among the operating state data, two of the voltage, current, temperature and operating frequency are not within the threshold range;

[0134] The power equipment health state is high risk, that is, three or more of voltage, current, temperature and operating frequency in the operating state data are not in the threshold range;

[0135] The current power equipment health state is determined by the operating state data at the current time according to a threshold comparison algorithm;

[0136] The future power equipment health state is determined by the future power equipment operating state output by the time series analysis model and according to a threshold comparison algorithm;

[0137] S6, the cloud server pushes the power equipment state evaluation report and the corresponding operating state data to the user terminal;

[0138] In this embodiment, after the cloud server completes the generation of the power equipment state evaluation report, the report and the corresponding operating state data are pushed to the user terminal, and the specific process is as follows:

[0139] The cloud server uses the message queue telemetry transfer (MQTT) protocol for data pushing;

[0140] When the power equipment state evaluation report is generated, the cloud server encapsulates the evaluation report and the related operating state data; the data encapsulation format uses the JSON (JavaScript Object Notation) format;

[0141] The user terminal includes but is not limited to a smart phone, a tablet computer, a smart bracelet and a computer;

[0142] The user terminal is pre-installed with a special power equipment monitoring application program; after the application program is started, the MQTT connection with the cloud server is established; after the connection is successfully established, the cloud server sends the encapsulated power equipment state evaluation report and the operating state data to the corresponding topic (Topic) through the publish (Publish) function of MQTT;

[0143] The user terminal subscribes to the topic to receive the pushed data in real time; after receiving the data, the application program analyzes the JSON format data; for the power equipment state evaluation report, the application program presents the report to the user in an intuitive visual interface, including but not limited to a chart of different colors representing the health state of the power equipment (green represents normal, yellow represents low risk, orange represents medium risk, and red represents high risk), and detailed text description is attached; for the operating state data, the application program displays the change trend of voltage, current, temperature and other parameters in the form of a list, so that the user can check the real-time operating condition and historical data of the power equipment at any time, so as to make decisions and arrange equipment maintenance in time.

[0144] A back clamp-based power equipment state information acquisition system, comprising:

[0145] Back clamp collection module: integrated sensor and microcontroller, can be connected with power equipment, collect running state data and perform preliminary processing.

[0146] Communication module: support wired and LPWAN wireless communication, with data transmission and identity verification function, can safely upload the data processed by the back clamp collection module to the cloud server.

[0147] Cloud analysis module: with data storage, algorithm analysis and alarm generation unit. Store the received data, evaluate the equipment health status with specific algorithm, generate evaluation report and alarm information.

[0148] User terminal module: provide visual interface, receive and display the equipment state report and alarm information pushed by the cloud analysis module, facilitate user to master the equipment situation.

[0149] The above formulas are dimensionless, only numerical values are used for calculation. These formulas are based on a large amount of data and obtained through software simulation, aiming to be as close to the actual situation as possible. The preset parameters in the formula can be adjusted by those skilled in the art according to specific needs.

[0150] In the description of the present specification, the description of the terms "one embodiment", "example", "specific example" and the like means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0151] The preferred embodiments of the application disclosed above are only used to help explain the application. The preferred embodiments do not describe all the details and do not limit the application to the specific embodiments. Obviously, many modifications and changes can be made according to the content of the present specification. The present specification selects and describes these embodiments in order to better explain the principles and practical applications of the application, so that those skilled in the art can well understand and utilize the application. The application is limited by the claims and their entire scope and equivalents.

Claims

1. A back clip-based power equipment state information acquisition method, characterized by, The method comprises the following steps: S1, establishing a bidirectional data communication link by connecting the physical interface of the back clamp device with the standardized data interface of the power equipment; S2, collecting the running state data of the power equipment in real time, wherein the running state data at least includes voltage, current, temperature, running frequency and equipment identification code; S3, preprocessing the collected running state data, wherein the preprocessing includes signal filtering, AD conversion and data normalization; S4, uploading the preprocessed running state data to the cloud server through the wireless communication module built in the back clamp, wherein the wireless communication module supports at least one LPWAN protocol; S5, after the cloud server receives and stores the running state data, generating a power equipment health state evaluation report based on a time series analysis model and a preset threshold comparison algorithm, wherein the power equipment state evaluation report includes the current power equipment health state and the future power equipment health state; S6, pushing the power equipment state evaluation report and the corresponding running state data to the user terminal by the cloud server; In step S1, the physical interface runs a multi-protocol adaptive interface engine to dynamically adapt the communication protocols of different power equipment, including the following steps: 1) Protocol sniffing: automatically scanning the power equipment interface after the back clamp device is powered on, detecting the communication protocols supported by the current equipment, including RS-485, Modbus and IEC 61850; 2) Protocol loading: matching the corresponding protocol analysis module from the protocol library built in the back clamp device and dynamically loading it into the memory; 3) Link keep-alive: periodically sending heartbeat frames and monitoring the response delay, and triggering re-linking if the timeout occurs; In step S5, the time series analysis model uses a pre-trained long short-term memory network LSTM to measure the running state trend of the power equipment, and adjusts the model parameters of the LSTM according to the stored running state data at a fixed period; The model parameter adjustment of the LSTM includes the following steps: Constructing dataset: from the running state data stored in the cloud server side, extract the sliding window data with a time length of to form the input features, and use the data after time steps of the sliding window as the labels; let the current time be , then the input features and the labels are represented as follows: wherein, is voltage data in the operating state data, is current data, is temperature data, is operating frequency data; the above input features and labels form a piece of data, and the above process is repeated until the number of data in the data set reaches ; Compute training loss: loss function Measure prediction bias using mean squared error: wherein, is the first true value of the i-th data; is the predicted value output by the LSTM model according to the input features; is the number of data sets; Model parameter update: update the LSTM model parameter matrix by momentum stochastic gradient descent : wherein, is a learning rate, is a momentum coefficient; is a gradient of the loss function with respect to is a gradient of the loss function with respect to is a training epoch.

2. The method of claim 1, wherein, In step S2, the voltage and current are collected by the high-precision voltage transformer and the Hall effect current sensor built in the back clamp, respectively; The temperature is collected by a temperature sensor installed on the surface of the back clamp in contact with the power equipment; The running frequency is obtained by zero-crossing detection of the voltage signal; The equipment identification code is read by the RFID module of the back clamp; The collection frequency of the back clamp device can be dynamically adjusted, and the programmable sampling controller built in the back clamp device automatically configures the collection frequency to a corresponding value in the range of 1Hz~10kHz according to the pre-stored device type and monitoring level mapping table.

3. The method of claim 1, wherein, In step S3, the signal filtering includes: Applying Kalman filtering to voltage and current signals to eliminate random noise; Using sliding average filtering to smooth the temperature signal; Performing power frequency notch filtering on the running frequency signal to suppress the fundamental wave interference.

4. The method of claim 3, wherein, The AD conversion is realized by the programmable ADC module built in the back clamp device, which converts the running state data from analog signal data to digital signal data, i.e. discrete data that can be processed by a computer; The data normalization includes: dividing the actually collected voltage data and current data by the rated voltage and rated current of the power equipment respectively to obtain unitless voltage and current units; Linearly scaling the temperature data to map to the range of 0~1; Dividing the operating frequency by the rated operating frequency of the current equipment to obtain a unitless frequency.

5. The method of claim 1, wherein, The wireless communication module in step S4 adopts an abnormal priority upload mechanism during data transmission, divides the operating state data into abnormal data and normal data, and uploads the abnormal data to the cloud server in priority; If at least one of the voltage data, current data, or temperature data in the operating state data exceeds the preset threshold range, it is the abnormal data; The wireless communication module adopts an AES-128 encryption algorithm to encrypt the operating state data during wireless transmission, and verifies the identity legality of the back clamping device through a two-way authentication mechanism with the cloud server.

6. The method of claim 1, wherein, The health status of the power equipment is divided into normal, low risk, medium risk, and high risk; The threshold comparison algorithm includes voltage threshold, current threshold, temperature threshold, and operating frequency threshold; When the health status of the power equipment is normal, the voltage, current, temperature, and operating frequency in the operating state data are all within the threshold range; When the health status of the power equipment is low risk, one of the voltage, current, temperature, and operating frequency in the operating state data is not within the threshold range; When the health status of the power equipment is medium risk, two of the voltage, current, temperature, and operating frequency in the operating state data are not within the threshold range; When the health status of the power equipment is high risk, three or more of the voltage, current, temperature, and operating frequency in the operating state data are not within the threshold range; The current health status of the power equipment is determined by the operating state data at the current time according to the threshold comparison algorithm; The future health status of the power equipment is determined by the future operating state of the power equipment output by the time series analysis model and according to the threshold comparison algorithm.

7. A back clip based power equipment status information acquisition system for implementing the method of claims 1-6, characterized by, It includes: Back clamping collection module: integrated sensor and microcontroller, can be connected with power equipment, collect operating state data and perform preliminary processing; Communication module: supports wired and LPWAN wireless communication, has data transmission and identity verification functions, and can safely upload the data processed by the back clamping collection module to the cloud server; Cloud analysis module: with data storage, algorithm analysis and alarm generation unit; store the received data, evaluate the equipment health status with specific algorithm, generate evaluation report and alarm information; User terminal module: provides a visual interface, receives and displays the equipment state report and alarm information pushed by the cloud analysis module, and facilitates users to master the equipment situation.

Images