Transformer management method and system based on Internet of Things

Abstract

The present invention relates to the field of transformer management technology, in particular to a transformer management method and system based on the Internet of Things, which comprises a data collection module, a transformer state analysis module, an environmental impact analysis module and a decision and adjustment module, wherein: the data collection module collects dynamic state data and transformer operating environment data, as well as static state data and transformer operating environment data; the transformer state analysis module establishes a logistic regression model and predicts the current transformer state according to the static state data and the dynamic state data; the environmental impact analysis module calculates the static environmental impact weight according to the static operating environment data using a linear regression algorithm; and the decision and adjustment module determines whether to adjust the environmental factors corresponding to the features in the dynamic operating environment data according to the current transformer state.

Description

Technical Field

[0001] The present invention relates to the technical field of mutual inductor management, and in particular to a mutual inductor management method and system based on the Internet of Things. Background Art

[0002] Other traditional transformer management systems have shown several obvious disadvantages. First, traditional systems often rely on relatively primitive data collection and processing methods, which not only limits the real-time update and accurate analysis of data, but also leads to slow and poor timeliness in judging the status of transformers. For example, these systems may only collect data periodically, such as daily or weekly inspections, and lack continuous monitoring of real-time data streams, making it impossible to detect and respond to small changes or abnormal conditions in transformer behavior in a timely manner. At the same time, such management systems often lack complex data analysis capabilities, such as the application of logistic regression and linear regression models, which makes it difficult for the system to predict and quantitatively analyze potential risks and faults, reducing the accuracy and efficiency of prevention and fault diagnosis.

[0003] In addition, traditional transformer management systems usually lack the ability to conduct in-depth analysis and weight priority sorting of the impact of environmental factors on equipment status. These systems may not be able to achieve real-time monitoring of environmental variables, let alone dynamically adjust environmental parameters based on complex data models. Without accurate environmental impact weight priority settings, even if the impact of environmental factors on transformer performance can be identified, traditional systems find it difficult to decide which parameters should be adjusted first and how to adjust them to most effectively restore or optimize the working state of the transformer. The lack of this processing mechanism means that when the transformer is in an abnormal state, the response of the traditional system is often delayed, and the processing method may appear rough or imprecise, thereby increasing the time to return to normal and reducing the operating efficiency and reliability of the equipment. Summary of the invention

[0004] The purpose of the present invention is to provide a mutual inductor management method and system based on the Internet of Things to solve the problems raised in the above background technology.

[0005] To achieve the above object, one of the objects of the present invention is to provide the following technical solution: a mutual inductor management method based on the Internet of Things, characterized in that it includes the following method steps:

[0006] S1. Acquire dynamic state data and dynamic operating environment data of the transformer, as well as static state data and static operating environment data of the transformer; wherein the static operating environment data is data corresponding to each environment type collected when the transformer is in an abnormal state;

[0007] S2. Establish a logistic regression model based on the static state data, and use a linear regression algorithm to calculate the weight corresponding to each environment type in the static operating environment data, and sort each environment type in descending order according to the weight to obtain a sequence table;

[0008] S3, the logistic regression model predicts the state of the transformer by inputting dynamic state data, wherein:

[0009] If the prediction result is normal, a notification that the current transformer status is normal is sent to the administrator;

[0010] If the prediction result is abnormal, proceed to S4;

[0011] Wherein, after all the environment types in the sequence table are deleted and the prediction result is still abnormal, a notification is sent to the administrator that the dynamic operating environment data does not affect the current transformer;

[0012] S4, adjusting the first environment type in the sequence list in S2 in the dynamic operating environment data, and deleting the environment type from the sequence list after each adjustment, and returning to S3 to re-predict the state of the mutual inductor.

[0013] As a further improvement of the present technical solution, the dynamic state data in S1 includes but is not limited to current, voltage and transformer temperature; the dynamic operating environment data includes but is not limited to humidity, ambient temperature and vibration intensity; the static state data includes the characteristics of the dynamic state data and the abnormal state of the corresponding transformer; the static operating environment data is the data when the transformer is in an abnormal state, including but not limited to humidity, ambient temperature and vibration intensity.

[0014] As a further improvement of the technical solution, it is characterized in that the logistic regression model is established based on the static state data in S2, specifically including:

[0015] By converting the abnormal state of the mutual inductor in the static state data into a numerical type, 1 represents abnormality and 0 represents normality, and using it as a label;

[0016] Labels are used to guide the optimization direction of the loss function during the training process of the logistic regression model. The logistic regression model estimates the logistic regression model parameters by learning the relationship between the input data and the output label;

[0017] A logistic regression model is built by minimizing the difference between the model's predictions and the actual labels by minimizing the loss function.

[0018] As a further improvement of the technical solution, S2 uses a linear regression algorithm to calculate the weight corresponding to each environment type in the static operating environment data, specifically including:

[0019] According to the mathematical formula of the linear regression algorithm, the least square method is used to solve the model parameters by minimizing the gap between the actual observed values ​​and the predicted values ​​of the linear regression algorithm;

[0020] Minimize the residual sum of squares between the linear regression model prediction value and the actual observation value and find the optimal parameter value;

[0021] The optimal parameter value is the weight corresponding to each environment type in the static operating environment data, and the environment types are sorted in descending order according to the weight to obtain a sequence table.

[0022] As a further improvement of the technical solution, the logistic regression model specifically defines a hypothesis function of the model by initializing model parameters, including a weight vector and a bias term;

[0023] Assume that the function linearly combines the input features with the parameters, adds the bias term, and uses the Sigmoid function to map this value to a probability value between 0 and 1;

[0024] Define the cost function of the model, a cross-entropy-based loss function that measures the difference between the model's predictions and the actual labels;

[0025] The parameter values ​​are adjusted by minimizing the cost function. The parameter optimization is done with the help of the stochastic gradient descent algorithm, which calculates the partial derivative of the cost function with respect to each parameter and updates the parameter values.

[0026] As a further improvement of the present technical solution, the mathematical formula of the linear regression algorithm is Y=β0+β1X1+β2X2+β3X3...+βnXn, and the label Y in the linear regression algorithm is set to 1; the independent variable X contains the data corresponding to each environmental type, β1, β2, β3...βn are the weights corresponding to the data of each environmental type, and β0 is the intercept.

[0027] The second object of the present invention is to provide a transformer management system based on the Internet of Things, including a data collection module, a transformer status analysis module, an environmental impact analysis module and a decision and adjustment module, wherein:

[0028] The data collection module is used to obtain the dynamic state data and dynamic operating environment data of the transformer, as well as the static state data and static operating environment data of the transformer; wherein the static operating environment data is the data collected under the abnormal state of the transformer

[0029] The transformer state analysis module establishes a logistic regression model based on the static state data; at the same time, the environmental impact analysis module uses a linear regression algorithm to calculate the weight corresponding to each environment type in the static operating environment data, and sorts each environment type in descending order according to the weight to obtain a sequence table;

[0030] The decision and adjustment module is used to send notifications to administrators and to adjust the environment type in the dynamic operating environment data;

[0031] The logistic regression model is used to predict the state of the transformer by inputting dynamic state data. If the prediction result is normal, the decision and adjustment module sends a notification to the administrator that the current transformer state is normal; if the prediction result is abnormal, the decision and adjustment module adjusts the first environment type in the corresponding sequence list in the dynamic operating environment data, and deletes the environment type from the sequence list after each adjustment, and re-predicts the state of the transformer using the logistic regression model;

[0032] Among them, after all the environment types in the sequence table are deleted and the prediction result is still abnormal, the decision and adjustment module sends a notification to the administrator that the dynamic operating environment data does not affect the current transformer

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] 1. The transformer management method and system based on the Internet of Things obtains and records the dynamic and static state data of the transformer in real time through the data collection module, such as voltage, current, temperature and environmental conditions; these data are transmitted to the transformer state analysis module and the environmental impact analysis module, where the data not only supports the establishment and training of logistic regression and linear regression models, but is also used to predict the working state of the transformer in real time and analyze the impact of environmental factors on abnormal states; in this process, the continuously optimized algorithm model can timely and accurately predict and identify the abnormal behavior of the transformer, providing data support for subsequent decision-making. In addition, through the quantitative evaluation of environmental factors, the system can point out which specific environmental variables have a significant impact on transformer abnormalities, providing a scientific basis for the formulation of adjustment measures.

[0035] 2. The decision-making and adjustment module in the mutual inductor management method and system based on the Internet of Things is the core execution unit of the system. It dynamically adjusts the operating environment of the mutual inductor to optimize its working state according to the data and analysis results provided by the first two modules; when the system detects that the mutual inductor is in an abnormal state, the decision-making unit will start the adjustment unit to adjust key factors such as ambient temperature and humidity according to the preset environmental impact weight priority; after each adjustment, the system will re-evaluate the state of the mutual inductor until the state returns to normal or all influencing factors are adjusted; this means that the system can not only respond quickly when a problem is initially discovered, but also ensure that each step of adjustment is based on the latest data and analysis results through continuous monitoring and adjustment feedback; this highly dynamic and responsive processing flow significantly improves the system's response speed and processing accuracy to faults, greatly reduces downtime and maintenance costs caused by faults, and also extends the service life of the equipment, enhancing the operating efficiency and reliability of the entire Internet of Things system. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of the overall module of the present invention;

[0037] Figure 2 It is a schematic diagram of the decision-making and adjustment module unit of the present invention;

[0038] Figure 3 It is a schematic diagram of the overall method flow of the present invention.

[0039] In the figure: 100, data collection module; 200, transformer status analysis module; 300, environmental impact analysis module; 400, decision and adjustment module; 401, decision unit; 402, adjustment unit. DETAILED DESCRIPTION

[0040] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] See also Figure 1-3 One of the purposes of this embodiment is to provide a transformer management system based on the Internet of Things, including a data collection module 100, a transformer status analysis module 200, an environmental impact analysis module 300 and a decision and adjustment module 400.

[0042] The data collection module 100 collects dynamic state data and transformer operating environment data, as well as static state data and transformer operating environment data; dynamic data refers to instantly updated data for real-time analysis; and static data refers to historical data in the database for algorithm model building.

[0043] The dynamic state data in the data collection module 100 includes but is not limited to current, voltage and transformer temperature, etc., and the dynamic operating environment data includes but is not limited to humidity, ambient temperature and vibration intensity, etc.; the static state data includes the characteristics of the dynamic state data and the abnormal state of the corresponding transformer; the static operating environment data is the data when the transformer is in an abnormal state, including but not limited to humidity, ambient temperature and vibration intensity, etc.

[0044] The transformer state analysis module 200 establishes a logistic regression model based on the static state data, and uses the established logistic regression model to predict whether the transformer is in an abnormal state based on the dynamic state data, specifically including:

[0045] The abnormal state of the transformer in the static state data is converted into a numerical type, 1 for abnormality and 0 for normality, and used as a label. The label is used to guide the optimization direction of the loss function during the training process of the logistic regression model. The logistic regression model estimates the logistic regression model parameters by learning the relationship between the input data (features) and the output label (abnormal state). The difference between the model prediction value and the actual label is minimized by minimizing the loss function, thereby establishing a logistic regression model.

[0046] The specific construction of the logistic regression model is as follows:

[0047] Define the hypothesis function: initialize the regression coefficient vector w and the bias term b, define the hypothesis function h(x) of the logistic regression model, use the Sigmoid function to linearly combine the input feature x and the parameter w, and add the bias term b, and map it to a probability value between 0 and 1 through the Sigmoid function. The Sigmoid function maps any real input value to an output value between 0 and 1. The mathematical form is f(x) = 1 / (1+e^(-x)). The function form of the hypothesis function h(x) is: h(x) = Sigmoid(w^T*x+b), where T refers to the transpose of the vector w, w is a weight vector, x is the input vector, and b is the bias term;

[0048] Define the cost function: Define the cost function J(w,b) of the logistic regression model. The cost function is the cross entropy loss function, which measures the difference between the model prediction result and the label. The form of the cost function is: J(w,b) = -1 / m*sum(y*log(h(x))+(1-y)*log(1-h(x))), where sum is the summation operation and y is the true label;

[0049] Parameter optimization: The optimal parameters w and b are obtained by minimizing the cost function J(w,b), and the stochastic gradient descent algorithm is used to update the parameters to reduce the cost function. First, the predicted value h(x) is calculated, and then the partial derivatives of the cost function with respect to the parameters w and b are calculated. Finally, the parameters w2=w1-learning_rate*dw, b2=b1-learning_rate*db are updated; where learning_rate is the learning rate, dw and db are the partial derivatives of the cost function with respect to the parameters, w1 is the weight vector of the current iteration step, w2 is the weight vector of the next iteration step, b1 is the bias term of the current iteration step, and b2 is the bias term of the next iteration step.

[0050] The dynamic state data is input into the established logistic regression model, and the result of whether the transformer is abnormal is output, where 1 indicates abnormality and 0 indicates normality, which is used as the current transformer state.

[0051] The transformer status analysis module 200 uses a logistic regression model to perform real-time analysis on the collected dynamic transformer status data, and can accurately predict whether the transformer is in an abnormal state. This capability not only enhances the efficiency of monitoring the operating status of the transformer, but also improves the ability to predict and prevent faults, thereby significantly reducing downtime and maintenance costs caused by faults; in addition, the transformer status analysis module 200 provides accurate transformer status information for subsequent use. By monitoring the status of the transformer in real time and accurately predicting its abnormal conditions, the module ensures that the decision and adjustment module 400 can respond in a timely manner, thereby initiating necessary adjustment measures, such as environmental control tuning and fault prevention measures, providing data support for the optimization of maintenance strategies, helping to achieve the optimal operating state of the transformer, extending the life of the equipment, and ensuring the stable and reliable operation of the system, thereby enhancing the efficiency and sustainable performance of the entire Internet of Things system.

[0052] The environmental impact analysis module 300 uses a linear regression algorithm to calculate the influence weight of the data features in the transformer operating environment data on the transformer being in an abnormal state (expressed by the characteristic coefficients in the linear regression algorithm: β1, β2, β3...), based on the static operating environment data, specifically including:

[0053] The mathematical formula of the linear regression algorithm is Y=β0+β1X1+β2X2+β3X3...+βnXn, and the label Y in the linear regression algorithm is set to 1; the independent variable X contains the data corresponding to each environmental type, β1,β2,β3...βn are the weights corresponding to the data of each environmental type, β0 is the intercept, and β0 is the intercept; the least squares method is used to solve the model parameters by minimizing the gap between the actual observed value and the predicted value of the linear regression model. During the training process of the linear regression model, the optimal parameter value is found so that the residual square sum between the predicted value of the linear regression model and the actual observed value is minimized; after the linear regression model training is completed, the influence weight of each environmental data feature on the abnormal state of the transformer will be obtained. These weights (β1,β2,β3...) represent the contribution degree (weight) of each environmental factor to the abnormal state of the transformer under the abnormal state of the transformer, and the environmental types are sorted in descending order according to the weight to obtain a sequence table.

[0054] Through real-time monitoring and in-depth analysis of various environmental factors (such as temperature, humidity, vibration, etc.), the impact of these external conditions on the performance of the transformer and the entire system can be accurately evaluated. This analysis is extremely important for identifying potential risks and predicting future state changes, so that the response to environmental changes is no longer passive, but can make predictive adjustments based on data; in addition, the output of the environmental impact analysis module 300 provides key information for the decision-making and adjustment module 400, enabling it to formulate more efficient and targeted adjustment measures to optimize system operating conditions and improve system stability and reliability.

[0055] The decision unit 401 in the decision and adjustment module 400 receives the current transformer state in the transformer state analysis module 200 and the static environmental impact weight in the environmental impact analysis module 300, and determines whether to adjust the environmental factors corresponding to the characteristics in the dynamic operating environment data according to the current transformer state, specifically including:

[0056] If the current transformer state is 0, it means that the current transformer is in normal state, and a notification indicating that the current transformer state is normal is sent to the administrator;

[0057] If the current transformer state is 1, it means that the current transformer is in an abnormal state, then the start adjustment unit 402 adjusts the environmental factors corresponding to the characteristics in the dynamic operating environment data in turn according to the size of the static environment impact weight (for example, the ambient temperature has the highest weight, so the ambient temperature is adjusted first, and the adjustment method includes but is not limited to starting the air-conditioning system), and generates new dynamic operating environment data (obtained again by the data collection module 100). After each adjustment, the decision unit 401 will make a judgment again and predict the current transformer state of the adjusted dynamic operating environment data (predicted again by the transformer state analysis module 200). If the new current transformer state is 0, it means that the current state has been Resolve the exception and send a relevant notification that the current transformer state is normal to the administrator; otherwise, continue to start the adjustment unit 402, and adjust the environmental factors corresponding to the features in the next dynamic operating environment data according to the size of the static environment impact weight until all the dynamic operating environment data features are adjusted, so as to determine whether the current transformer is in an abnormal state due to environmental factors; when the environmental factors corresponding to all the features in the dynamic operating environment data are adjusted, the decision unit 401 determines that the new current transformer state is 1 at this time, indicating that the current transformer is in an abnormal state not caused by environmental factors, and sends a notification to the management personnel that the current transformer state abnormality is not related to environmental factors.

[0058] The design and implementation of the decision and adjustment module 400 adopts a highly dynamic and responsive method process to ensure that the system can effectively monitor and optimize the operating status of the transformer in real time; first, the decision unit 401 comprehensively considers the current transformer status provided by the transformer status analysis module 200 and the environmental factor weight information provided by the environmental impact analysis module 300 to effectively identify whether the transformer is in an abnormal state and the contribution of environmental factors to the abnormal state; this process not only relies on accurate data analysis, but also combines complex algorithm models, such as logistic regression and linear regression, to predict and evaluate the impact of different factors;

[0059] When the transformer is in an abnormal state, the adjustment unit 402 adjusts the environmental variables that are assessed to be most likely to cause the abnormality, such as ambient temperature, humidity, etc., one by one according to the preset priority relationship, and adopts a dynamic adjustment strategy, such as turning on the air conditioner to lower the temperature, in conjunction with real-time status monitoring and feedback; this strategy ensures that after a comprehensive assessment of all possible influencing factors, adjustment measures can be implemented in a targeted manner, rather than blindly trial and error.

[0060] Specifically, when the transformer is in an abnormal state, the decision and adjustment module 400 can not only quickly identify and send abnormal notifications, but also prioritize the adjustment of those environmental variables that are most likely to affect the normal operation of the transformer according to the preset environmental factor influence weights, such as adjusting the ambient temperature first. After each adjustment, the decision unit 401 will re-evaluate the transformer status. If the status returns to normal, a normal status notification will be sent; if it is still abnormal, the module will continue to adjust other environmental factors in sequence according to the weights; this series of operations is not only based on the specific guidance obtained from data analysis, but also through real-time monitoring and dynamic adjustment feedback mechanism, it ensures that each adjustment step is targeted, thereby improving the efficiency and accuracy of handling abnormalities.

[0061] The second purpose of this embodiment is to provide a mutual inductor management method based on the Internet of Things, including the following method steps:

[0062] S1. Acquire dynamic state data and dynamic operating environment data of the transformer, as well as static state data and static operating environment data of the transformer; wherein the static operating environment data is data corresponding to each environment type collected when the transformer is in an abnormal state;

[0063] S2. Establish a logistic regression model based on the static state data, and use a linear regression algorithm to calculate the weight corresponding to each environment type in the static operating environment data, and sort each environment type in descending order according to the weight to obtain a sequence table;

[0064] S3, the logistic regression model predicts the state of the transformer by inputting dynamic state data, wherein:

[0065] If the prediction result is normal, a notification that the current transformer status is normal is sent to the administrator;

[0066] If the prediction result is abnormal, proceed to S4;

[0067] Wherein, after all the environment types in the sequence table are deleted and the prediction result is still abnormal, a notification is sent to the administrator that the dynamic operating environment data does not affect the current transformer;

[0068] S4, adjusting the first environment type in the sequence list in S2 in the dynamic operating environment data, and deleting the environment type from the sequence list after each adjustment, and returning to S3 to re-predict the state of the mutual inductor.

[0069] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A mutual inductor management method based on the Internet of Things, characterized in that: It includes the following method steps: S1. Acquire dynamic state data and dynamic operating environment data of the transformer, as well as static state data and static operating environment data of the transformer; wherein the static operating environment data is data corresponding to each environment type collected when the transformer is in an abnormal state; S2. Establish a logistic regression model based on the static state data, and use a linear regression algorithm to calculate the weight corresponding to each environment type in the static operating environment data, and sort each environment type in descending order according to the weight to obtain a sequence table; S3, the logistic regression model predicts the state of the transformer by inputting dynamic state data, wherein: If the prediction result is normal, a notification that the current transformer status is normal is sent to the administrator; If the prediction result is abnormal, proceed to S4; Wherein, after all the environment types in the sequence table are deleted and the prediction result is still abnormal, a notification is sent to the administrator that the dynamic operating environment data does not affect the current transformer; S4, adjusting the first environment type in the sequence list in S2 in the dynamic operating environment data, and deleting the environment type from the sequence list after each adjustment, and returning to S3 to re-predict the state of the mutual inductor; The dynamic state data in S1 includes current, voltage and transformer temperature; the dynamic operating environment data includes humidity, ambient temperature and vibration intensity; the static state data includes the characteristics of the dynamic state data and the abnormal state of the corresponding transformer; the static operating environment data is the data when the transformer is in an abnormal state, including humidity, ambient temperature and vibration intensity; In S2, a logistic regression model is established based on static state data, specifically including: By converting the abnormal state of the mutual inductor in the static state data into a numerical type, 1 represents abnormality and 0 represents normality, and using it as a label; Labels are used to guide the optimization direction of the loss function during the training process of the logistic regression model. The logistic regression model determines the logistic regression model parameters by learning the relationship between the input data and the output label. Build a logistic regression model by minimizing the difference between the model predictions and the actual labels by minimizing the loss function; The S2 uses a linear regression algorithm to calculate the weight corresponding to each environment type in the static operating environment data, specifically including: According to the mathematical formula of the linear regression algorithm, the least square method is used to solve the model parameters by minimizing the gap between the actual observed values ​​and the predicted values ​​of the linear regression algorithm; Minimize the residual sum of squares between the linear regression model prediction value and the actual observation value and find the optimal parameter value; The optimal parameter value is the weight corresponding to each environment type in the static operating environment data, and the environment types are sorted in descending order according to the weight to obtain a sequence table; The logistic regression model defines a hypothesis function, a cost function and parameter optimization, specifically including: Define the model's hypothesis function by initializing the model parameters, including the weight vector and bias term; Assume that the function linearly combines the input features with the parameters, adds the bias term, and uses the Sigmoid function to map this value to a probability value between 0 and 1; Define the cost function of the model, a cross-entropy-based loss function that measures the difference between the model's predictions and the actual labels; The parameter values ​​are adjusted by minimizing the cost function. The parameter optimization is done with the help of the stochastic gradient descent algorithm, which calculates the partial derivative of the cost function with respect to each parameter and updates the parameter values.

2. The method for managing a mutual inductor based on the Internet of Things according to claim 1, characterized in that: The mathematical formula of the linear regression algorithm is Y=β0+β1X1+β2X2+β3X3...+βnXn, and the label Y in the linear regression algorithm is set to 1; the independent variable X contains the data corresponding to each environment type, β1, β2, β3...βn are the weights corresponding to the data of each environment type, and β0 is the intercept.

3. A mutual inductor management system based on the Internet of Things, used in the mutual inductor management method based on the Internet of Things according to any one of claims 1-2, characterized in that: The Internet of Things-based transformer management system comprises a data collection module (100), a transformer status analysis module (200), an environmental impact analysis module (300) and a decision and adjustment module (400), wherein: The data collection module (100) is used to obtain dynamic state data and dynamic operating environment data of the transformer, as well as static state data and static operating environment data of the transformer; wherein the static operating environment data is data collected when the transformer is in an abnormal state. The mutual inductor state analysis module (200) establishes a logistic regression model based on the static state data; at the same time, the environmental impact analysis module (300) calculates the weight corresponding to each environment type in the static operating environment data using a linear regression algorithm, and sorts each environment type in descending order of weight to obtain a sequence table; The decision and adjustment module (400) is used to send a notification to the administrator and is also used to adjust the environment type in the dynamic operating environment data; The logistic regression model is used to predict the state of the transformer by inputting dynamic state data. If the prediction result is normal, the decision and adjustment module (400) sends a notification to the administrator that the current transformer state is normal. If the prediction result is abnormal, the decision and adjustment module (400) adjusts the first environment type in the corresponding sequence list in the dynamic operating environment data, and deletes the environment type from the sequence list after each adjustment is completed, and re-predicts the state of the transformer using the logistic regression model. Wherein, after all environment types in the sequence table are deleted and the prediction result is still abnormal, the decision and adjustment module (400) sends a notification to the administrator that the dynamic operating environment data does not affect the current mutual inductor.

Images