Method and system for intelligently controlling noise of outer cover of transformer

Abstract

The invention provides a method and a system for intelligently controlling noise of a transformer outer cover, and the method comprises the steps: monitoring the operation state data of a transformer in real time through installing a plurality of sensors, carrying out the primary processing of the data on an edge calculation node, transmitting the data to a central server through a wireless sensor network, and extracting features through principal component analysis; an LSTM deep learning model is constructed based on historical data to predict a noise level, load distribution and current are dynamically adjusted according to a prediction result, an energy-saving mode is optimized, filter parameters are automatically adjusted by using an adaptive filter technology in combination with real-time monitoring data, and noise of different frequency bands is suppressed; the development of the user interface supports remote access and manual adjustment of working parameters, regularly evaluates the system performance and updates an intelligent control strategy, effectively reduces the noise of the transformer, improves the energy efficiency and the equipment reliability, and is suitable for transformer noise management in various industrial scenes.

Description

Technical Field

[0001] The invention relates to the field of transformer noise reduction, and in particular to a method and system for intelligently controlling transformer cover noise. Background Art

[0002] In industrial environments, transformers are one of the key equipment. The noise generated during their operation not only affects the quality of the working environment, but may also cause interference to the surrounding community. Traditional noise control methods mainly rely on physical isolation measures and manual adjustment of transformer loads to reduce noise levels.

[0003] Traditional noise control methods are mostly static, that is, they are rarely adjusted once installed. Faced with ever-changing working conditions and environmental factors, these fixed measures are difficult to adapt in real time, resulting in poor noise suppression effects. In addition, manual adjustment of loads to reduce noise requires professional operation and has a slow response speed, making it impossible to respond quickly to emergencies.

[0004] Due to the lack of precise data analysis and support, existing control strategies often fail to effectively optimize energy consumption. For example, maintaining high current output under low load conditions not only increases unnecessary energy consumption, but may also accelerate the aging process of the equipment.

[0005] Current noise control systems generally lack intelligent perception and adaptive adjustment capabilities, which means they cannot automatically adjust their behavior to achieve optimal performance based on various parameters monitored in real time. The lack of intelligent support also means that it is difficult for the system to achieve predictive maintenance, which may lead to additional noise problems caused by equipment failure. Summary of the invention

[0006] The purpose of the present invention is to provide a method and system for intelligently controlling transformer cover noise.

[0007] The problem to be solved by the present invention is: by integrating multiple sensors to monitor the operating status data, using edge computing for data preprocessing and feature extraction, combining deep learning algorithms to predict the noise level, and dynamically adjusting the load and current to optimize the energy-saving mode, using adaptive filter technology to suppress noise in different frequency bands, developing a user interface to support remote monitoring and manual adjustment, regularly evaluating system performance and updating intelligent control strategies to ensure continuous optimization, effectively reducing noise levels, improving energy efficiency, and enhancing system stability and maintenance convenience.

[0008] A method for intelligently controlling noise of transformer cover, the technical scheme adopted is as follows: S1: Install sensors to monitor the operating status data of the transformer cover, including acoustic data, vibration data, temperature data, current data, humidity data and air flow velocity data; S2: Perform preliminary processing on the data monitored by the sensor on the edge computing node, including filtering and denoising. Based on wireless sensor networks and Internet of Things technologies, the data after preliminary processing is transmitted in real time, and features are extracted from the transmitted data. S3: Based on the features extracted from historical data and combined with deep learning algorithms, a prediction model is constructed to predict the noise level under different working conditions, and the model parameters are continuously optimized based on the transformer cover operating status data obtained through continuous monitoring; S4: Establish a dynamic load adjustment model, combine the noise levels under different working conditions predicted in S3, dynamically start the energy-saving mode, and automatically adjust the current; S5: Based on adaptive filter technology and combined with the structural characteristics of the transformer housing, an adaptive filter function model is established according to real-time monitoring results to automatically adjust the filter parameters and suppress noise in different frequency bands; S6: Develop a user interface that provides real-time data display, trend analysis, and alarm functions, displays the transformer's operating status and supports manual adjustment of the transformer's operating parameters, and provides remote access via the Internet; S7: Regularly evaluate the dynamic load adjustment model and the adaptive filter function model to check whether the actual noise control effect meets expectations, establish a feedback loop and update the intelligent control strategy.

[0009] Furthermore, the sensors installed in S1 monitor the operating status data of the transformer cover, including: The installed sensors include acoustic sensors, vibration sensors, temperature sensors, current sensors, humidity sensors, and air velocity sensors; Acoustic sensors are installed at the location of noise sources inside the housing and at the locations of reflection and resonance around the outside of the housing to monitor sound signals of different frequencies; Vibration sensors are installed at the supporting points, bases and joints of the transformer where mechanical vibration and looseness occur to monitor vibration signals; The temperature sensor is installed at the heating parts of the transformer winding and core to monitor the ambient temperature data; Current sensors are installed at the input and output ports of the transformer to monitor current data; The temperature sensor is installed on the outer cover of the transformer to monitor the environmental humidity data; Air flow velocity sensors are installed at the air inlet and outlet of the transformer to monitor the air flow velocity data.

[0010] Furthermore, in S2, the data monitored by the sensor is preliminarily processed on the edge computing node, the preliminarily processed data is transmitted in real time, and feature extraction is performed on the transmitted data, including: Digital filters are used to remove high-frequency noise and low-frequency drift in the signal, wavelet transform is used to distinguish useful signals from background noise, and all data monitored by the sensor are converted into the same format; Compress the preprocessed data to reduce the data volume, and pack the compressed data into data packets, each of which includes a timestamp, a sensor ID, and data type information; Use encryption technology for data transmission and set up access control mechanisms to grant authorized users access rights; Principal component analysis is used to reduce data dimensions and extract features for deep learning.

[0011] Furthermore, the features extracted from the historical data in S3 are combined with a deep learning algorithm to build a prediction model, including: S31: dividing the features extracted from the historical acoustic data, vibration data, temperature data, current data, humidity data, and airflow velocity data into a time series data set according to the order of timestamps; S32: Use the LSTM long short-term memory network model, use the divided time series data set to train the model, and adjust the model parameters to minimize the prediction error; S33: A mapping function is constructed based on the LSTM long short-term memory network model, and the input multi-dimensional time series data features are converted into predicted noise levels through the mapping function; S34: The mapping function is ,in is the predicted noise level at time point t, , , , , and are the characteristic data of acoustics, vibration, temperature, current, humidity and air flow velocity at time point t, is the ReLU activation function, b is the bias term, , , , , and is the weight matrix corresponding to each type of sensor data; S35: Use online learning mechanism to continuously input new time series data sets to update the model.

[0012] Furthermore, in S4, a dynamic load adjustment model is established, and the energy-saving mode is dynamically turned on in combination with the noise levels under different working conditions predicted in S3, including: S41: Accumulate the load of each transformer at time point t to calculate the total load at time point t , ,in is the load of the i-th transformer at time point t; S42: Setting the noise threshold ,when and , turn on energy saving mode, ,in is the predicted noise level at time point t, is the low load threshold, k is the adjustment coefficient, and D is the load adjustment value; S43: Automatically adjust the current based on the load adjustment value calculated in S42, ,in is the voltage value at time t, is the power factor, is the current value at time t after adjustment.

[0013] Furthermore, in S5, an adaptive filter function model is established according to the real-time monitoring result to automatically adjust the filter parameters, including: S51: The real-time monitoring result is the real-time operating status data of the transformer cover, including acoustic data, vibration data, temperature data, current data, humidity data and air flow velocity data. The current data is the current value adjusted in S4. An adaptive filter function model is established according to the real-time monitoring result. S52: Adaptive filter function model , where n is the time index, N is the filter order, is the weight of the kth tap at time n, is the input transformer cover status data; S53: Update weights, ,in is the weight of the kth tap at time n+1, is the initial step size parameter, is the regulating factor, , , , They are real-time temperature data, real-time humidity data, adjusted current data and real-time air flow speed data. is the expected acoustic signal, is the actual output acoustic signal, is the correction function, It is real-time vibration data; S54: According to the real-time operating status data of the transformer cover and the adjusted current data, noises in different frequency bands are suppressed based on an adaptive filter function model.

[0014] Furthermore, the S7 establishes a feedback loop and updates the intelligent control strategy, including: The evaluation cycle is determined to be once a quarter based on the transformer working environment and change rate, and the historical acoustic data, vibration data, temperature data, current data, humidity data and air flow velocity data are selected for each quarter; Measure the noise level in different frequency bands in each quarter, compare the actual value with the target value, and calculate the total energy consumption of the transformer under different working conditions in each quarter; Based on the difference between the actual and target noise levels in different frequency bands of the transformer and the total energy consumption under different working conditions in each quarter, the intelligent control strategy is updated. The updated intelligent control strategy is tested in a simulation environment before deployment.

[0015] Furthermore, a system for intelligent control of transformer cover noise is used to implement any of the methods for intelligent control of transformer cover noise, and the system for intelligent control of transformer cover noise includes: a data acquisition module, an edge computing and preliminary processing module, a data analysis and feature extraction module, a noise prediction model building module, a dynamic load adjustment module, an adaptive filter module, a user interface and remote management module, and a feedback loop and strategy update module: Data acquisition module: Install sensors to monitor acoustics, vibration, temperature, current, humidity and air velocity data in real time, and use wireless sensor networks for data transmission; Edge computing and preliminary processing module: performs preliminary processing on the collected data, including filtering and denoising, converting all data into the same format, compressing the processed data, and adopting access control mechanisms; Data analysis and feature extraction module: further analyzes the transmitted data, uses principal component analysis to reduce the data dimension to extract features, processes chronological data sets, and identifies patterns and trends; Noise prediction model building module: Build a deep learning model based on historical data analysis to predict the noise level under different working conditions, use a long short-term memory network model to process time series data, adjust parameters to minimize prediction errors, and continuously input new data to update the model; Dynamic load adjustment module: dynamically adjusts load distribution according to the predicted noise level, optimizes energy-saving mode, accumulates the load of each transformer to get the total load, determines whether to turn on the energy-saving mode based on the noise threshold, and automatically adjusts the current based on the load adjustment value; Adaptive filter module: Establish an adaptive filter function model based on real-time monitoring results, automatically adjust filter parameters to suppress noise, dynamically adjust weights based on real-time data, and adjust step parameters based on temperature, humidity, current intensity, and airflow velocity data; User interface and remote management module: Provides real-time data display, trend analysis, alarm functions, supports remote access and manual adjustment of working parameters, develops a user-friendly graphical interface, displays indicators and status information, and allows authorized users to access the system via the Internet; Feedback loop and strategy update module: Regularly evaluate system performance, update intelligent control strategies based on feedback loops, set up comprehensive evaluations every quarter, compare actual noise levels with target values, and test updated strategies in a simulation environment.

[0016] The beneficial effects of the present invention are as follows: by combining multiple sensor data including acoustics, vibration, temperature, current, humidity, and airflow velocity data, a prediction model is constructed using a deep learning algorithm, which can accurately predict the noise level under different working conditions. The application of adaptive filter technology can dynamically adjust parameters according to real-time monitoring results, effectively suppress noise in different frequency bands, and improve the accuracy and efficiency of noise control; The dynamic load adjustment model automatically turns on or off the energy-saving mode according to the predicted noise level and adjusts the current value to reduce energy consumption, while ensuring that the noise level meets the requirements, reducing energy consumption, extending the service life of the equipment and reducing maintenance costs; Regularly evaluating system performance and updating intelligent control strategies based on feedback loops ensures the system's continuous optimization and ability to adapt to environmental changes. The introduction of an online learning mechanism allows the model to continuously improve itself as new data is input, thereby improving the stability and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flow chart of a method for intelligent control of transformer cover noise; Figure 2 The figure is a system module diagram for intelligent control of transformer cover noise. DETAILED DESCRIPTION

[0018] The present invention is further clearly and completely described below, but the protection scope of the present invention is not limited thereto.

[0019] A method for intelligently controlling noise of transformer cover, the technical scheme adopted is as follows: S1: Install sensors to monitor the operating status data of the transformer cover, including acoustic data, vibration data, temperature data, current data, humidity data and air flow velocity data; S2: Perform preliminary processing on the data monitored by the sensor on the edge computing node, including filtering and denoising. Based on wireless sensor networks and Internet of Things technologies, the data after preliminary processing is transmitted in real time, and features are extracted from the transmitted data. S3: Based on the features extracted from historical data and combined with deep learning algorithms, a prediction model is constructed to predict the noise level under different working conditions, and the model parameters are continuously optimized based on the transformer cover operating status data obtained through continuous monitoring; S4: Establish a dynamic load adjustment model, combine the noise levels under different working conditions predicted in S3, dynamically start the energy-saving mode, and automatically adjust the current; S5: Based on adaptive filter technology and combined with the structural characteristics of the transformer housing, an adaptive filter function model is established according to real-time monitoring results to automatically adjust the filter parameters and suppress noise in different frequency bands; S6: Develop a user interface that provides real-time data display, trend analysis, and alarm functions, displays the transformer's operating status and supports manual adjustment of the transformer's operating parameters, and provides remote access via the Internet; S7: Regularly evaluate the dynamic load adjustment model and the adaptive filter function model to check whether the actual noise control effect meets expectations, establish a feedback loop and update the intelligent control strategy.

[0020] refer to Figure 1 The figure shows a flow chart of a method for intelligently controlling transformer cover noise.

[0021] Furthermore, the sensors installed in S1 monitor the operating status data of the transformer cover, including: The installed sensors include acoustic sensors, vibration sensors, temperature sensors, current sensors, humidity sensors, and air velocity sensors; Acoustic sensors are installed at the location of noise sources inside the housing and at the locations of reflection and resonance around the outside of the housing to monitor sound signals of different frequencies; Vibration sensors are installed at the supporting points, bases and joints of the transformer where mechanical vibration and looseness occur to monitor vibration signals; The temperature sensor is installed at the heating parts of the transformer winding and core to monitor the ambient temperature data; Current sensors are installed at the input and output ports of the transformer to monitor current data; The temperature sensor is installed on the outer cover of the transformer to monitor the environmental humidity data; Air flow velocity sensors are installed at the air inlet and outlet of the transformer to monitor the air flow velocity data.

[0022] Furthermore, in S2, the data monitored by the sensor is preliminarily processed on the edge computing node, the preliminarily processed data is transmitted in real time, and feature extraction is performed on the transmitted data, including: Digital filters are used to remove high-frequency noise and low-frequency drift in the signal, wavelet transform is used to distinguish useful signals from background noise, and all data monitored by the sensor are converted into the same format; Compress the preprocessed data to reduce the data volume, and pack the compressed data into data packets, each of which includes a timestamp, a sensor ID, and data type information; Use encryption technology for data transmission and set up access control mechanisms to grant authorized users access rights; Principal component analysis is used to reduce data dimensions and extract features for deep learning.

[0023] Furthermore, the features extracted from the historical data in S3 are combined with a deep learning algorithm to build a prediction model, including: S31: dividing the features extracted from the historical acoustic data, vibration data, temperature data, current data, humidity data, and airflow velocity data into a time series data set according to the order of timestamps; S32: Use the LSTM long short-term memory network model, use the divided time series data set to train the model, and adjust the model parameters to minimize the prediction error; S33: A mapping function is constructed based on the LSTM long short-term memory network model, and the input multi-dimensional time series data features are converted into predicted noise levels through the mapping function; S34: The mapping function is ,in is the predicted noise level at time point t, , , , , and are the characteristic data of acoustics, vibration, temperature, current, humidity and air flow velocity at time point t, is the ReLU activation function, b is the bias term, , , , , and is the weight matrix corresponding to each type of sensor data; S35: Use online learning mechanism to continuously input new time series data sets to update the model.

[0024] Furthermore, in S4, a dynamic load adjustment model is established, and the energy-saving mode is dynamically turned on in combination with the noise levels under different working conditions predicted in S3, including: S41: Accumulate the load of each transformer at time point t to calculate the total load at time point t , ,in is the load of the i-th transformer at time point t; S42: Setting the noise threshold ,when and , turn on energy saving mode, ,in is the predicted noise level at time point t, is the low load threshold, k is the adjustment coefficient, and D is the load adjustment value; S43: Automatically adjust the current based on the load adjustment value calculated in S42, ,in is the voltage value at time t, is the power factor, is the current value at time t after adjustment.

[0025] Furthermore, in S5, an adaptive filter function model is established according to the real-time monitoring result to automatically adjust the filter parameters, including: S51: The real-time monitoring result is the real-time operating status data of the transformer cover, including acoustic data, vibration data, temperature data, current data, humidity data and air flow velocity data. The current data is the current value adjusted in S4. An adaptive filter function model is established according to the real-time monitoring result. S52: Adaptive filter function model , where n is the time index, N is the filter order, is the weight of the kth tap at time n, is the input transformer cover status data; S53: Update weights, ,in is the weight of the kth tap at time n+1, is the initial step size parameter, is the regulating factor, , , , They are real-time temperature data, real-time humidity data, adjusted current data and real-time air flow speed data. is the expected acoustic signal, is the actual output acoustic signal, is the correction function, It is real-time vibration data; S54: According to the real-time operating status data of the transformer cover and the adjusted current data, noises in different frequency bands are suppressed based on an adaptive filter function model.

[0026] Furthermore, the S7 establishes a feedback loop and updates the intelligent control strategy, including: The evaluation cycle is determined to be once a quarter based on the transformer working environment and change rate, and the historical acoustic data, vibration data, temperature data, current data, humidity data and air flow velocity data are selected for each quarter; Measure the noise level in different frequency bands in each quarter, compare the actual value with the target value, and calculate the total energy consumption of the transformer under different working conditions in each quarter; Based on the difference between the actual and target noise levels in different frequency bands of the transformer and the total energy consumption under different working conditions in each quarter, the intelligent control strategy is updated. The updated intelligent control strategy is tested in a simulation environment before deployment.

[0027] Furthermore, a system for intelligent control of transformer cover noise is used to implement any of the methods for intelligent control of transformer cover noise, and the system for intelligent control of transformer cover noise includes: a data acquisition module, an edge computing and preliminary processing module, a data analysis and feature extraction module, a noise prediction model building module, a dynamic load adjustment module, an adaptive filter module, a user interface and remote management module, and a feedback loop and strategy update module: Data acquisition module: Install sensors to monitor acoustics, vibration, temperature, current, humidity and air velocity data in real time, and use wireless sensor networks for data transmission; Edge computing and preliminary processing module: performs preliminary processing on the collected data, including filtering and denoising, converting all data into the same format, compressing the processed data, and adopting access control mechanisms; Data analysis and feature extraction module: further analyzes the transmitted data, uses principal component analysis to reduce the data dimension to extract features, processes chronological data sets, and identifies patterns and trends; Noise prediction model building module: Build a deep learning model based on historical data analysis to predict the noise level under different working conditions, use a long short-term memory network model to process time series data, adjust parameters to minimize prediction errors, and continuously input new data to update the model; Dynamic load adjustment module: dynamically adjusts load distribution according to the predicted noise level, optimizes energy-saving mode, accumulates the load of each transformer to get the total load, determines whether to turn on the energy-saving mode based on the noise threshold, and automatically adjusts the current based on the load adjustment value; Adaptive filter module: Establish an adaptive filter function model based on real-time monitoring results, automatically adjust filter parameters to suppress noise, dynamically adjust weights based on real-time data, and adjust step parameters based on temperature, humidity, current intensity, and airflow velocity data; User interface and remote management module: Provides real-time data display, trend analysis, alarm functions, supports remote access and manual adjustment of working parameters, develops a user-friendly graphical interface, displays indicators and status information, and allows authorized users to access the system via the Internet; Feedback loop and strategy update module: Regularly evaluate system performance, update intelligent control strategies based on feedback loops, set up comprehensive evaluations every quarter, compare actual noise levels with target values, and test updated strategies in a simulation environment.

[0028] refer to Figure 2 The figure shows a system module diagram for intelligent control of transformer cover noise.

[0029] The present invention provides a method and system for intelligent control of transformer cover noise. The method and system install multiple sensors to monitor transformer operating status data in real time, perform preliminary processing on the data on the edge computing node, and transmit the data to the central server through a wireless sensor network. Principal component analysis is used to extract features, and an LSTM deep learning model is constructed based on historical data to predict the noise level. The load distribution and current are dynamically adjusted according to the prediction results, the energy-saving mode is optimized, and adaptive filter technology is used in combination with real-time monitoring data to automatically adjust filter parameters and suppress noise in different frequency bands. A user interface is developed to support remote access and manual adjustment of working parameters, and system performance is regularly evaluated and the intelligent control strategy is updated. The method effectively reduces transformer noise, improves energy efficiency and equipment reliability, and is suitable for transformer noise management in various industrial scenarios.

Claims

1. A method for intelligent control of transformer cover noise, characterized in that: include: S1: Install sensors to monitor the operating status data of the transformer cover, including acoustic data, vibration data, temperature data, current data, humidity data and air flow velocity data; S2: Perform preliminary processing on the data monitored by the sensor on the edge computing node, including filtering and denoising. Based on wireless sensor networks and Internet of Things technologies, the data after preliminary processing is transmitted in real time, and features are extracted from the transmitted data. S3: Based on the features extracted from historical data and combined with deep learning algorithms, a prediction model is constructed to predict the noise level under different working conditions, and the model parameters are continuously optimized based on the transformer cover operating status data obtained through continuous monitoring; S4: Establish a dynamic load adjustment model, combine the noise levels under different working conditions predicted in S3, dynamically start the energy-saving mode, and automatically adjust the current; S5: Based on adaptive filter technology and combined with the structural characteristics of the transformer housing, an adaptive filter function model is established according to real-time monitoring results to automatically adjust the filter parameters and suppress noise in different frequency bands; S6: Develop a user interface that provides real-time data display, trend analysis, and alarm functions, displays the transformer's operating status and supports manual adjustment of the transformer's operating parameters, and provides remote access via the Internet; S7: Regularly evaluate the dynamic load adjustment model and the adaptive filter function model to check whether the actual noise control effect meets expectations, establish a feedback loop and update the intelligent control strategy.

2. A method for intelligently controlling transformer cover noise according to claim 1, characterized in that: The sensors installed in S1 monitor the operating status data of the transformer cover, including: The installed sensors include acoustic sensors, vibration sensors, temperature sensors, current sensors, humidity sensors, and air velocity sensors; Acoustic sensors are installed at the location of noise sources inside the housing and at the locations of reflection and resonance around the outside of the housing to monitor sound signals of different frequencies; Vibration sensors are installed at the supporting points, bases and joints of the transformer where mechanical vibration and looseness occur to monitor vibration signals; The temperature sensor is installed at the heating parts of the transformer winding and core to monitor the ambient temperature data; Current sensors are installed at the input and output ports of the transformer to monitor current data; The temperature sensor is installed on the outer cover of the transformer to monitor the environmental humidity data; Air flow velocity sensors are installed at the air inlet and outlet of the transformer to monitor the air flow velocity data.

3. A method for intelligently controlling transformer cover noise according to claim 1, characterized in that: In S2, the data monitored by the sensor is preliminarily processed on the edge computing node, the preliminarily processed data is transmitted in real time, and features are extracted from the transmitted data, including: Digital filters are used to remove high-frequency noise and low-frequency drift in the signal, wavelet transform is used to distinguish useful signals from background noise, and all data monitored by the sensor are converted into the same format; Compress the preprocessed data to reduce the data volume, and pack the compressed data into data packets, each of which includes a timestamp, a sensor ID, and data type information; Use encryption technology for data transmission and set up access control mechanisms to grant authorized users access rights; Principal component analysis is used to reduce data dimensions and extract features for deep learning.

4. A method for intelligently controlling transformer cover noise according to claim 1, characterized in that: In S3, the features extracted from historical data are combined with a deep learning algorithm to build a prediction model, including: S31: dividing the features extracted from the historical acoustic data, vibration data, temperature data, current data, humidity data, and airflow velocity data into a time series data set according to the order of timestamps; S32: Use the LSTM long short-term memory network model, use the divided time series data set to train the model, and adjust the model parameters to minimize the prediction error; S33: A mapping function is constructed based on the LSTM long short-term memory network model, and the input multi-dimensional time series data features are converted into predicted noise levels through the mapping function; S34: The mapping function is ,in is the predicted noise level at time point t, , , , , and are the characteristic data of acoustics, vibration, temperature, current, humidity and air flow velocity at time point t, is the ReLU activation function, b is the bias term, , , , , and is the weight matrix corresponding to each type of sensor data; S35: Use online learning mechanism to continuously input new time series data sets to update the model.

5. The method for intelligently controlling transformer cover noise according to claim 1, characterized in that: In S4, a dynamic load adjustment model is established, and the energy-saving mode is dynamically turned on in combination with the noise levels under different working conditions predicted in S3, including: S41: Accumulate the load of each transformer at time point t to calculate the total load at time point t , ,in is the load of the i-th transformer at time point t; S42: Setting the noise threshold ,when and , turn on energy saving mode, ,in is the predicted noise level at time point t, is the low load threshold, k is the adjustment coefficient, and D is the load adjustment value; S43: Automatically adjust the current based on the load adjustment value calculated in S42, ,in is the voltage value at time t, is the power factor, is the current value at time t after adjustment.

6. A method for intelligently controlling transformer cover noise according to claim 1, characterized in that: In S5, an adaptive filter function model is established according to the real-time monitoring result to automatically adjust the filter parameters, including: S51: The real-time monitoring result is the real-time operating status data of the transformer cover, including acoustic data, vibration data, temperature data, current data, humidity data and air flow velocity data. The current data is the current value adjusted in S4. An adaptive filter function model is established according to the real-time monitoring result. S52: Adaptive filter function model , where n is the time index, N is the filter order, is the weight of the kth tap at time n, is the input transformer cover status data; S53: Update weights, ,in is the weight of the kth tap at time n+1, is the initial step size parameter, is the regulating factor, , , , They are real-time temperature data, real-time humidity data, adjusted current data and real-time air flow speed data. is the expected acoustic signal, is the actual output acoustic signal, is the correction function, It is real-time vibration data; S54: According to the real-time operating status data of the transformer cover and the adjusted current data, noises in different frequency bands are suppressed based on an adaptive filter function model.

7. A method for intelligently controlling transformer cover noise according to claim 1, characterized in that: The S7 establishes a feedback loop and updates the intelligent control strategy, including: The evaluation cycle is determined to be once a quarter based on the transformer working environment and change rate, and the historical acoustic data, vibration data, temperature data, current data, humidity data and air flow velocity data are selected for each quarter; Measure the noise level in different frequency bands in each quarter, compare the actual value with the target value, and calculate the total energy consumption of the transformer under different working conditions in each quarter; Based on the difference between the actual and target noise levels in different frequency bands of the transformer and the total energy consumption under different working conditions in each quarter, the intelligent control strategy is updated. The updated intelligent control strategy is tested in a simulation environment before deployment.

8. A system for intelligent control of transformer cover noise, characterized in that: A method for intelligently controlling the noise of a transformer cover is used to implement any of the above, and the system for intelligently controlling the noise of a transformer cover comprises: a data acquisition module, an edge computing and preliminary processing module, a data analysis and feature extraction module, a noise prediction model building module, a dynamic load adjustment module, an adaptive filter module, a user interface and remote management module, and a feedback loop and strategy update module: Data acquisition module: Install sensors to monitor acoustics, vibration, temperature, current, humidity and air velocity data in real time, and use wireless sensor networks for data transmission; Edge computing and preliminary processing module: performs preliminary processing on the collected data, including filtering and denoising, converting all data into the same format, compressing the processed data, and adopting access control mechanisms; Data analysis and feature extraction module: further analyzes the transmitted data, uses principal component analysis to reduce the data dimension to extract features, processes chronological data sets, and identifies patterns and trends; Noise prediction model building module: Build a deep learning model based on historical data analysis to predict the noise level under different working conditions, use a long short-term memory network model to process time series data, adjust parameters to minimize prediction errors, and continuously input new data to update the model; Dynamic load adjustment module: dynamically adjusts load distribution according to the predicted noise level, optimizes energy-saving mode, accumulates the load of each transformer to get the total load, determines whether to turn on the energy-saving mode based on the noise threshold, and automatically adjusts the current based on the load adjustment value; Adaptive filter module: Establish an adaptive filter function model based on real-time monitoring results, automatically adjust filter parameters to suppress noise, dynamically adjust weights based on real-time data, and adjust step parameters based on temperature, humidity, current intensity, and airflow velocity data; User interface and remote management module: Provides real-time data display, trend analysis, alarm functions, supports remote access and manual adjustment of working parameters, develops a user-friendly graphical interface, displays indicators and status information, and allows authorized users to access the system via the Internet; Feedback loop and strategy update module: Regularly evaluate system performance, update intelligent control strategies based on feedback loops, set up comprehensive evaluations every quarter, compare actual noise levels with target values, and test updated strategies in a simulation environment.

Images