Data contribution degree analysis method and device, storage medium and electronic device

Abstract

The invention discloses a data contribution degree analysis method and device, a storage medium and an electronic device, and relates to the technical field of power grids. The method comprises the steps that to-be-processed first electric power data are acquired, the first electric power data are used for representing sensing data collected in the operation process of a transformer or a nonlinear system under a preset condition, and the preset condition is used for representing the condition of data missing in the data collection, transmission or storage process; performing data preprocessing on the first power data to obtain second power data; and performing data analysis on the second power data based on the target nonlinear model to obtain an analysis result. The technical problem that a linear analysis method in the prior art cannot effectively capture the nonlinear relation of a transformer and other complex systems, and consequently the deviation of key parameter analysis results is large is solved.

Description

Technical Field

[0001] The present invention relates to the field of power grid technology, and in particular to a data contribution analysis method, device, storage medium and electronic device. Background Art

[0002] In modern industrial and power systems, the key parameter analysis of complex nonlinear systems such as transformers is crucial to ensure the stability and safety of the system. However, in practical applications, sensor failures, the complexity of the data acquisition environment, and communication errors often lead to missing or abnormal data. Processing these missing data and accurately analyzing the contribution of key parameters to system performance has become an important issue in improving system operation efficiency.

[0003] Existing key parameter analysis methods are mainly based on linear models. However, complex systems such as transformers often exhibit highly nonlinear characteristics. General linear analysis methods cannot effectively capture these complex nonlinear relationships, resulting in large deviations in the results of key parameter analysis. In addition, the problem of missing data exacerbates this challenge. In response to these problems, some methods for processing missing data and analyzing nonlinear systems have emerged in recent years, but the existing technical methods have their own limitations.

[0004] Methods for dealing with missing data in industrial systems mainly include data interpolation and outlier detection. For example, commonly used missing data filling methods include mean interpolation, nearest neighbor interpolation, and interpolation techniques based on statistical models. These methods can fill data gaps to a certain extent, but they cannot effectively deal with complex nonlinear relationships. In addition, for nonlinear systems such as transformers, general data interpolation and outlier removal methods may not be able to accurately reconstruct the dynamic characteristics of the system.

[0005] For the modeling of nonlinear systems, current technologies mostly use methods such as deep learning. These methods can better capture the nonlinear characteristics of the system, but the above methods require a large amount of training data, and the interpretability of their models is poor, making it difficult to quantitatively analyze the specific contribution of each parameter to the system performance. For example, although the deep neural network model can learn complex nonlinear relationships, its black box characteristics make it difficult to explain the specific impact of each input parameter on the system output. This is not ideal for the analysis of key parameter contribution. Summary of the invention

[0006] The embodiments of the present invention provide a data contribution analysis method, device, storage medium and electronic device to at least solve the technical problem that the linear analysis method in the related art cannot effectively capture the nonlinear relationship of complex systems such as transformers, resulting in large deviations in the results of key parameter analysis.

[0007] According to one embodiment of the present invention, a data contribution analysis method is provided, the method comprising: obtaining first power data to be processed, wherein the first power data is used to represent the perception data collected during the operation of a transformer or a nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during data collection, transmission or storage; performing data preprocessing on the first power data to obtain second power data; performing data analysis on the second power data based on a target nonlinear model to obtain an analysis result, wherein the target nonlinear model is obtained based on polynomial basis function training, the target nonlinear model is used to determine the relationship between a target performance indicator and input data, and the analysis result is used to represent the contribution of any data in the second power data to the target performance indicator.

[0008] Optionally, performing data preprocessing on the first power data to obtain the second power data includes: using a preset filtering method to perform data filtering on the first power data to obtain third power data; and using a support vector machine interpolation method to complete the third power data to obtain the second power data.

[0009] Optionally, the data contribution analysis method further includes: constructing an initial nonlinear model based on a preset polynomial basis function; and adjusting model parameters of the initial nonlinear model based on an orthogonal least squares method to obtain a target nonlinear model.

[0010] Optionally, constructing the initial nonlinear model based on a preset polynomial basis function includes: constructing the initial nonlinear model based on a preset approximation principle and a preset polynomial basis function.

[0011] Optionally, adjusting the model parameters of the initial nonlinear model based on the orthogonal least squares method to obtain the target nonlinear model includes: transforming the basis function matrix of the initial nonlinear model to obtain an orthogonal basis matrix; determining the weight matrix of the orthogonal basis matrix based on the orthogonal least squares method; and adjusting the model parameters of the initial nonlinear model based on the weight matrix and the orthogonal basis matrix to obtain the target nonlinear model.

[0012] Optionally, adjusting the model parameters of the initial nonlinear model based on the weight matrix and the orthogonal basis matrix to obtain the target nonlinear model includes: determining the orthogonal basis function based on the orthogonal basis matrix; adjusting the model parameters of the initial nonlinear model based on the orthogonal basis matrix and the orthogonal basis function to obtain the target nonlinear model.

[0013] According to one embodiment of the present invention, there is also provided a data contribution analysis device, which includes: an acquisition module, used to acquire first power data to be processed, wherein the first power data is used to represent the perception data collected during the operation of the transformer or nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process; a processing module, used to perform data preprocessing on the first power data to obtain second power data; an analysis module, used to perform data analysis on the second power data based on a target nonlinear model to obtain an analysis result, wherein the target nonlinear model is obtained based on polynomial basis function training, and the target nonlinear model is used to determine the relationship between the target performance index and the input data, and the analysis result is used to represent the contribution of any data in the second power data to the target performance index.

[0014] Optionally, the processing module is further used to perform data filtering on the first power data using a preset filtering method to obtain third power data; and to perform data completion on the third power data using a support vector machine interpolation method to obtain second power data.

[0015] Optionally, the processing module is further used to construct an initial nonlinear model based on a preset polynomial basis function; and adjust model parameters of the initial nonlinear model based on an orthogonal least squares method to obtain a target nonlinear model.

[0016] Optionally, the processing module is further used to construct an initial nonlinear model based on a preset approximation principle and a preset polynomial basis function.

[0017] Optionally, the processing module is also used to transform the basis function matrix of the initial nonlinear model to obtain an orthogonal basis matrix; determine the weight matrix of the orthogonal basis matrix based on the orthogonal least squares method; and adjust the model parameters of the initial nonlinear model based on the weight matrix and the orthogonal basis matrix to obtain a target nonlinear model.

[0018] Optionally, the processing module is further used to determine an orthogonal basis function based on an orthogonal basis matrix; and adjust model parameters of the initial nonlinear model based on the orthogonal basis matrix and the orthogonal basis function to obtain a target nonlinear model.

[0019] According to one embodiment of the present invention, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute any of the above-mentioned data contribution analysis methods when running on a computer or a processor.

[0020] According to one embodiment of the present invention, there is also provided an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute any of the above data contribution analysis methods.

[0021] According to one embodiment of the present invention, there is also provided a computer program product, including a computer program, which implements any of the above data contribution analysis methods when executed by a processor.

[0022] In an embodiment of the present invention, by obtaining first power data to be processed, wherein the first power data is used to represent the perception data collected during the operation of the transformer or nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process; data preprocessing is performed on the first power data to obtain second power data; data analysis is performed on the second power data based on a target nonlinear model to obtain an analysis result, wherein the target nonlinear model is obtained based on polynomial basis function training, and the target nonlinear model is used to determine the relationship between the target performance index and the input data, and the analysis result is used to represent the contribution of any data in the second power data to the target performance index, thereby achieving the purpose of accurately analyzing the contribution of key parameters of complex nonlinear systems in the case of data loss, thereby achieving the technical effect of accurately identifying the contribution of each parameter to the key performance index in the missing data under complex environments, thereby solving the technical problem that the linear analysis method in the related technology cannot effectively capture the nonlinear relationship of complex systems such as transformers, resulting in large deviations in the results of key parameter analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0024] Figure 1 is a flow chart of a data contribution analysis method according to one embodiment of the present invention;

[0025] Figure 2 is a structural block diagram of a data contribution analysis device according to one embodiment of the present invention. DETAILED DESCRIPTION

[0026] To facilitate understanding, some descriptions of concepts related to the embodiments of the present invention are exemplarily provided for reference.

[0027] As shown below:

[0028] Volterra Polynomial Basis Functions: Volterra polynomials are a special class of polynomials used to approximate the dynamic characteristics of nonlinear systems. In nonlinear system analysis, the Volterra series provides a method to decompose the response of a nonlinear system into basis functions of different orders, which represent the effects of different combinations of input variables on the system output. Volterra polynomial basis functions are used in this invention to construct nonlinear models to describe the relationship between key performance indicators and input parameters.

[0029] Hampel Filtering: Hampel filtering is a statistical method for detecting and removing outliers in a data sequence. Hampel filtering is based on the median and median absolute deviation (MAD) to identify points that deviate from the median by more than a certain threshold and mark them as outliers. In the data preprocessing stage of the present invention, Hampel filtering is used to remove abnormal data points caused by sensor failure or communication errors, thereby improving the accuracy of subsequent analysis.

[0030] Support Vector Machine Imputation (SVM Imputation): Support Vector Machine Imputation is a technique that uses a support vector machine (SVM) to fill in data. The support vector machine interpolation method predicts missing values ​​by building an SVM model, training the model based on existing input and output data, and then using the model to estimate the missing input values. In the present invention, support vector machine interpolation is used to handle missing data caused by sensor failures and other reasons, and complete the data set by predicting missing values ​​to ensure the integrity of complex nonlinear system analysis.

[0031] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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 should fall within the scope of protection of the present invention.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In the description of this embodiment, unless otherwise specified, the meaning of "multiple" is two or more. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] According to one embodiment of the present invention, an embodiment of a data contribution analysis method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0034] The method embodiment can be executed in an electronic device including a memory and a processor, a similar control device or a system. Taking an electronic device as an example, the electronic device may include one or more processors and a memory for storing data. Optionally, the electronic device may also include a communication device and a display device for communication functions. It can be understood by a person of ordinary skill in the art that the above structural description is only illustrative and does not limit the structure of the above electronic device. For example, the electronic device may also include more or fewer components than the above structural description, or have a configuration different from the above structural description.

[0035] The processor may include one or more processing units. For example, the processor may include a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processing (DSP) chip, a microcontroller unit (MCU), a field-programmable gate array (FPGA), a neural network processor (NPU), a tensor processing unit (TPU), an artificial intelligence (AI) type processor, and the like. Among them, different processing units may be independent components or integrated into one or more processors. In some instances, the electronic device may also include one or more processors.

[0036] The memory can be used to store computer programs, such as storing the computer program corresponding to the data contribution analysis method in the embodiment of the present invention, and the processor implements the above-mentioned data contribution analysis method by running the computer program stored in the memory. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0037] The communication device is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the communication device includes a network adapter (network interface controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the communication device can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0038] The display device may be, for example, a touch screen type liquid crystal display (LCD) and a touch display (also referred to as a "touch screen" or "touch display screen"). The liquid crystal display may enable a user to interact with a user interface of a mobile terminal. In some embodiments, the mobile terminal has a graphical user interface (GUI), and a user may interact with the GUI by finger contact and / or gestures on a touch-sensitive surface, wherein the human-computer interaction functions here may optionally include the following interactions: creating web pages, drawing, word processing, making electronic documents, games, video conferencing, instant messaging, sending and receiving emails, call interfaces, playing digital videos, playing digital music and / or web browsing, etc. The executable instructions for executing the above human-computer interaction functions are configured / stored in a computer program product or a readable storage medium executable by one or more processors.

[0039] In this embodiment, a data contribution analysis method running on an electronic device is provided. Figure 1 is a flow chart of a data contribution analysis method according to one embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0040] Step S10, obtaining first power data to be processed, wherein the first power data is used to represent the sensed data collected during the operation of the transformer or the nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process;

[0041] In an embodiment of the present invention, the first power data can be understood as an original data set containing some missing or abnormal values, reflecting the actual state of the system under the actual operating environment, but the data is incomplete due to various factors such as sensor failure, data transmission interruption, etc., which is not limited here.

[0042] Preset conditions can be understood as special situations that may be encountered during data collection, transmission or storage, mainly including data missing or data quality issues. For example, the existence of preset conditions means that some points or some parameter values ​​in the data set may not be recorded or recorded inaccurately, which may be caused by sensor failure, signal loss during data transmission, storage medium damage, etc., which are not limited here.

[0043] A nonlinear system can be understood as a physical or engineering system whose behavior cannot be simply described by a linear relationship. For example, in a power system, a transformer is a typical nonlinear system, and its performance indicators such as voltage regulation, efficiency, stability, etc., and the relationship between multiple input parameters such as input voltage, load current, temperature, etc. are often nonlinear, which is not limited here.

[0044] The sensed data can be understood as a component of the first power data collected under preset conditions, which is used for subsequent defect data processing and key parameter contribution analysis of nonlinear systems. Exemplarily, the sensed data is a direct reflection of the system operating status and can be used to monitor, analyze and control the performance of the system, which is not limited here.

[0045] Acquiring the first power data to be processed can be understood as acquiring a collected original data set containing some missing or abnormal values ​​during the operation of the transformer or the nonlinear system under preset conditions.

[0046] In an embodiment of the present invention, first power data to be processed is obtained, wherein the first power data is used to represent the sensed data collected during the operation of the transformer or nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process. Therefore, by obtaining and properly processing the first power data, the accuracy of the nonlinear system key parameter contribution analysis and the effectiveness of the model can be significantly improved.

[0047] Step S11, performing data preprocessing on the first power data to obtain second power data;

[0048] In the embodiment of the present invention, data preprocessing can be understood as a series of operations performed on the original data, that is, the first power data, before data analysis or modeling to ensure the cleanliness, integrity and consistency of the data, thereby improving the accuracy and efficiency of subsequent analysis. Exemplarily, data preprocessing includes removing outliers and filling missing values ​​to ensure the integrity and accuracy of the data, which is not limited here.

[0049] The second power data can be understood as a cleaned and prepared data set obtained after the data preprocessing step is performed on the first power data. Exemplarily, through data preprocessing, the outliers and missing values ​​in the second power data have been processed, the data format has been unified, and the features have been selected or converted, thereby providing a high-quality data foundation for subsequent in-depth analysis, which can more accurately reflect the true state of the system and improve the reliability and effectiveness of the analysis results, which is not limited here.

[0050] Performing data preprocessing on the first power data to obtain the second power data can be understood as performing data preprocessing on the first power data including operations such as removing outliers and filling missing values, so as to obtain high-quality data, namely, the second power data.

[0051] In an embodiment of the present invention, the second power data is obtained by performing data preprocessing on the first power data, including steps such as removing outliers and filling missing values, thereby avoiding the adverse effects of outliers and missing values ​​on statistical results and model training, and ensuring the accuracy and effectiveness of subsequent analysis.

[0052] Step S12, performing data analysis on the second power data based on the target nonlinear model to obtain analysis results, wherein the target nonlinear model is obtained based on polynomial basis function training, the target nonlinear model is used to determine the relationship between the target performance index and the input data, and the analysis results are used to represent the contribution of any data in the second power data to the target performance index.

[0053] In an embodiment of the present invention, the target nonlinear model can be understood as a model used to describe and predict the complex nonlinear relationship between target performance indicators in the second power data, such as transformer efficiency, failure rate, etc., and multiple input data, such as temperature, current, voltage, and other power system operating parameters.

[0054] The analysis result can be understood as the specific information obtained after in-depth analysis of the second power data based on the target nonlinear model, that is, the contribution of each data point in the second power data to the target performance indicator. For example, the analysis result reveals the specific influence of various operating parameters on system performance in a complex nonlinear system, which is not limited here.

[0055] The polynomial basis function can be understood as a mathematical function used to construct a target nonlinear model. For example, in power system analysis, the polynomial basis function is used to fit complex nonlinear relationships into a linear combination of basis functions, where each basis function is a polynomial expression of input data, which is not limited here.

[0056] The target performance index can be understood as a key parameter or index for understanding the change law or predicting the state of the power system in the analysis. For example, the target performance index may include the efficiency, failure rate, temperature fluctuation range, power output stability, etc. of the transformer, which are not limited here.

[0057] The input data may be understood to refer to all monitored and measured data points in the second power data that have an impact on the target performance indicator. For example, the input data may include current and voltage, temperature, humidity, etc., which are not limited here.

[0058] The second power data is analyzed based on the target nonlinear model, and the analysis result can be understood as performing data analysis on the second power data using the target nonlinear model trained with the polynomial basis function, exploring the relationship between the target performance index and the input data, thereby obtaining the contribution of any data in the second power data to the target performance index.

[0059] The target nonlinear model is obtained by training based on polynomial basis functions, which can be understood as the target nonlinear model is obtained by training based on polynomial basis functions, that is, a set of polynomial functions are used as basis functions to approximate or construct the relationship between power system performance and operating parameters. Exemplarily, the target nonlinear model can not only process the nonlinear characteristics of the data, but also can be trained and optimized using the complete second power data after data preprocessing to provide more accurate prediction and analysis capabilities, which is not limited here.

[0060] The target nonlinear model is used to determine the relationship between the target performance index and the input data. It can be understood that the target nonlinear model describes the nonlinear relationship between the input data and the target performance index through a polynomial basis function.

[0061] The analysis result is used to indicate the contribution of any data in the second power data to the target performance index. The analysis result can be understood as revealing the direct impact size and nature of each specific operating parameter, that is, each data point in the second power data, on the target performance index in a complex system.

[0062] In an embodiment of the present invention, data analysis is performed on the second power data based on a target nonlinear model obtained through training with a polynomial basis function, thereby determining the relationship between the target performance indicator and the input data, and obtaining the contribution of any data in the second power data to the target performance indicator, thereby significantly improving the accuracy and depth of power system analysis, and providing strong data support for fault prediction, system optimization, maintenance decisions, and resource management.

[0063] The first power data to be processed is obtained, wherein the first power data is used to represent the sensed data collected during the operation of the transformer or nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process. Therefore, by obtaining and properly processing the first power data, the accuracy of the contribution analysis of the key parameters of the nonlinear system and the effectiveness of the model can be significantly improved. The second power data is obtained by performing data preprocessing on the first power data, including steps such as removing outliers and filling missing values, thereby avoiding the adverse effects of outliers and missing values ​​on statistical results and model training, and ensuring the accuracy and effectiveness of subsequent analysis. The second power data is analyzed based on the target nonlinear model obtained by polynomial basis function training, so as to determine the relationship between the target performance index and the input data, and obtain the contribution of any data in the second power data to the target performance index, which significantly improves the accuracy and depth of power system analysis, and provides strong data support for fault prediction, system optimization, maintenance decision-making and resource management.

[0064] Through the above steps, by obtaining the first power data to be processed, wherein the first power data is used to represent the perception data collected during the operation of the transformer or nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process; performing data preprocessing on the first power data to obtain the second power data; performing data analysis on the second power data based on the target nonlinear model to obtain the analysis result, wherein the target nonlinear model is obtained based on polynomial basis function training, the target nonlinear model is used to determine the relationship between the target performance index and the input data, and the analysis result is used to represent the contribution of any data in the second power data to the target performance index, thereby achieving the purpose of accurately analyzing the contribution of key parameters of complex nonlinear systems in the case of data loss, thereby achieving the technical effect of accurately identifying the contribution of each parameter to the key performance index in the missing data under complex environment, thereby solving the technical problem that the linear analysis method in the related technology cannot effectively capture the nonlinear relationship of complex systems such as transformers, resulting in large deviations in the results of key parameter analysis.

[0065] Optionally, in step S11, performing data preprocessing on the first power data to obtain the second power data may include the following execution steps:

[0066] Step S111, filtering the first power data using a preset filtering method to obtain third power data;

[0067] Step S112: Use a support vector machine interpolation method to complete the third power data to obtain second power data.

[0068] In the embodiment of the present invention, the preset filtering method can be understood as a signal processing technology for processing and cleaning raw data, the main purpose of which is to remove noise, outliers or unnecessary information in the data to improve the accuracy and reliability of subsequent data analysis. Exemplarily, the preset filtering method can be a Hampel filtering method, a sliding average filter, a wavelet transform filter, and a low-pass filter, which are not limited here.

[0069] The third power data can be understood as the first power data processed by a preset filtering method, that is, a data set after noise, outliers or other interference factors are removed on the basis of the original data.

[0070] The support vector machine interpolation method can be understood as a machine learning-based technique for dealing with missing data problems, especially in prediction and regression problems. Exemplarily, the support vector machine interpolation method uses the powerful learning and prediction capabilities of the support vector machine (SVM) model to estimate the possible values ​​of missing data by analyzing the inherent structure and patterns in the data, which is not limited here.

[0071] The first power data is filtered by a preset filtering method to obtain the third power data. It can be understood that the third power data is obtained after removing noise, outliers or other interference factors on the basis of the first power data. Compared with the first power data, the third power data has higher reliability and less interference, and is the basis for building an accurate model and conducting in-depth analysis. Exemplarily, in this process, the Hampel filtering method is used to detect outliers and remove abnormal data. Support vector machine interpolation is used to complete the missing data to ensure the effectiveness of subsequent analysis. As an efficient and robust outlier identification method, Hampel filtering uses the median to estimate the data position and the median absolute deviation to approximate the standard deviation of the data. The process is described as follows: for the data sequence x1, x2, x3,…, x n , local median m i It can be calculated by formula (1).

[0072] m i =median(x i-k ,x i-k+1 ,...,x i ,...,x i+k-1 ,x i+k ) Formula (1)

[0073] Where median is the median, k is the number of neighbors on each side, and the width of the sliding window is 2k+1. The standard deviation is σ i =κ·m i ,in is an unbiased estimate of the Gaussian distribution. Outliers can be detected using formula (2).

[0074]

[0075] Among them, n σ is the set threshold.

[0076] The support vector machine interpolation method is used to complete the third power data to obtain the second power data. It can be understood that the support vector machine interpolation method is used to complete the missing data of the third power data from which noise and abnormal values ​​have been removed, so as to obtain a complete data set that reflects the system status as accurately as possible, that is, the second power data. For example, support vector machine interpolation is a method based on support vector machines that can be used to fill in missing data. For data pairs Support vector machine interpolation can be described as f(x) = (w·x) + b, where The minimization of the regularized risk function can be expressed as: Where C is the regularization constant, l ε() is the l-insensitive loss function. By introducing slack variables and Lagrange coefficients, the regression estimation can be expressed as formula (3).

[0077]

[0078] Here, k is a nonlinear kernel function.

[0079] In an embodiment of the present invention, by adopting a preset filtering method and a support vector machine interpolation method to perform data filtering and data completion on the first power data, higher quality second power data is finally obtained. That is, the use of the preset filtering method and the support vector machine interpolation method not only improves the quality of the original power data, but also provides a solid data foundation for the accurate analysis of the contribution of key parameters and the optimal management of the power system.

[0080] Optionally, the data contribution analysis method further includes the following execution steps:

[0081] Step S113, constructing an initial nonlinear model based on a preset polynomial basis function;

[0082] Step S114: adjusting the model parameters of the initial nonlinear model based on the orthogonal least squares method to obtain a target nonlinear model.

[0083] In the embodiment of the present invention, the preset polynomial basis function can be understood as a set of polynomial functions pre-selected in the process of building a nonlinear model, which functions are used as basic building blocks of the model to fit the complex relationship between input parameters and output responses. Exemplarily, the preset polynomial basis function can be a Volterra polynomial basis function, which is not limited here.

[0084] The initial nonlinear model can be understood as a mathematical model constructed using preset polynomial basis functions to describe the relationship between input data and target performance indicators.

[0085] The orthogonal least squares method can be understood as a statistical method for estimating model parameters, which is particularly suitable for solving the overfitting problem in nonlinear models.

[0086] The target nonlinear model can be understood as a model that is based on the initial nonlinear model and is obtained after orthogonal least squares parameter adjustment and optimization to effectively describe the nonlinear relationship between the input parameters and the target performance indicators.

[0087] Constructing an initial nonlinear model based on a preset polynomial basis function can be understood as constructing an initial nonlinear model using a preset polynomial basis function. Exemplarily, there is a complex nonlinear relationship between the key performance indicator and each input parameter. In this step, based on the universal approximation principle, the nonlinear function is approximated by a model composed of Volterra polynomial basis functions. Consider a nonlinear discrete system as shown in formula (4).

[0088] X t+1 =G(X t ,u t ), y t =h(X t ,u t ) Formula (4)

[0089] Where G(·) is a nonlinear function vector, h(·) is a nonlinear function, and X t is the state matrix, u t is the input, y t is the output. The nonlinear discrete system can be described by formula (5):

[0090] y t =f(y t-1 ,y t-2 ,…,y t-m ,u t-1 ,u t-2 ,…,u t-n ) Formula (5)

[0091] Where m and n are the maximum delays, and f(·) is a nonlinear function. According to the universal approximation principle, a single-layer neural network can accurately approximate any nonlinear function with a finite number of basis functions. The Volterra polynomial is used to approximate the nonlinear function f(x t ), where x t =[y t-1 ,y t-2 ,…,y t-m ,u t-1 ,u t-2 ,…,u t-n ]. Nonlinear function f(x t ) can be expressed as formula (6).

[0092]

[0093] The set of Volterra polynomials is formula (7).

[0094]

[0095] The number of basis functions is

[0096] Using the Volterra polynomial basis function, the nonlinear function f(x t ) can be expressed as formula (8).

[0097]

[0098] where o(x t l ) is the approximate error.

[0099] Adjusting the model parameters of the initial nonlinear model based on the orthogonal least squares method to obtain the target nonlinear model can be understood as optimizing the model parameters of the initial nonlinear model using the orthogonal least squares method, selecting and adjusting the weights and model parameters that best reflect the actual system performance, thereby obtaining a target nonlinear model that can accurately fit the data and avoid overfitting and has higher performance. Exemplarily, in this step, the weights and parameters of the initial nonlinear model constructed in step S113 are determined by the orthogonal least squares method. Assuming that a series of input and output data of the system have been given, based on formula (6) and formula 8, the input-output relationship can be written as formula (9).

[0100] Y=Φ(x)W+O(x l ) Formula (9)

[0101] The output matrix Weight Matrix Basis function matrix Estimation error matrix They are expressed as formula (10) to formula (13) respectively.

[0102] Y=[y1 y2 ... y M ] T Formula (10)

[0103] W=[w1 w2 ... w N ] T Formula (11)

[0104]

[0105]

[0106] Using the least squares method, the weight matrix W can be determined by formula (14).

[0107]

[0108] matrix Construct a series of basis function matrices. Orthogonal least squares method converts Φ iis converted into a series of orthogonal basis vectors. Then, the individual contribution of each basis vector to the output can be calculated. The orthogonal decomposition of the basis function matrix is ​​given by Φ(x)=PQ, where P is an M×N matrix with orthogonal columns and Q is an N×N unit upper triangular matrix with 1s on the diagonal and 0s below the diagonal. Using the P and Q matrices, Equation (9) can be written in the form of Equation (15).

[0109]

[0110] Where V = [v1v2...v N ] T To minimize ||Y-PV||2, the best estimate It can be calculated by formula (16).

[0111]

[0112] The variance of the output can be expressed as formula (17).

[0113]

[0114] in is the increment of the expected output variance. The error reduction rate can be defined as formula (18).

[0115]

[0116] Formula (18) expresses the individual contribution of each basis function to the output. i The search for a subset of basis functions can be implemented using the classic GramSchmidt method.

[0117] In the embodiment of the present invention, by constructing an initial nonlinear model based on a preset polynomial basis function and then adjusting the parameters using the orthogonal least squares method, an accurate, simple and highly interpretable target nonlinear model can be obtained. This model provides a powerful mathematical tool for key parameter analysis, performance prediction and optimization of the power system, helps to improve the operating efficiency and safety of the system, and promotes scientific decision-making based on data.

[0118] Optionally, in step S113, constructing an initial nonlinear model based on a preset polynomial basis function may include the following execution steps:

[0119] Step S1131, constructing an initial nonlinear model based on a preset approximation principle and a preset polynomial basis function.

[0120] In the embodiment of the present invention, the preset approximation principle can be understood as the Universal Approximation Theorem. Exemplarily, the core idea of ​​the Universal Approximation Theorem is that for a continuous real-valued function, as long as it is within a finite interval, it can be approximated with arbitrary precision by a single-layer feedforward neural network with enough hidden units (in this application, it can be understood as a model composed of polynomial basis functions), which is not limited here.

[0121] Constructing the initial nonlinear model based on the preset approximation principle and the preset polynomial basis function can be understood as constructing the initial nonlinear model using the universal approximation principle and the Volterra polynomial basis function to approximately describe the nonlinear relationship between the key performance indicators of the system and each input parameter.

[0122] In an embodiment of the present invention, an initial nonlinear model is constructed based on a preset approximation principle and a preset polynomial basis function, which provides a powerful mathematical framework for the analysis of complex systems such as power systems, thereby not only helping to more accurately describe system behavior, but also laying the foundation for further model optimization and data analysis.

[0123] Optionally, in step S114, adjusting the model parameters of the initial nonlinear model based on the orthogonal least squares method to obtain the target nonlinear model may include the following execution steps:

[0124] Step S1141, transforming the basis function matrix of the initial nonlinear model to obtain an orthogonal basis matrix;

[0125] Step S1142, determining a weight matrix of an orthogonal basis matrix based on an orthogonal least squares method;

[0126] Step S1143, adjusting the model parameters of the initial nonlinear model based on the weight matrix and the orthogonal basis matrix to obtain a target nonlinear model.

[0127] In the embodiment of the present invention, the basis function matrix can be understood as a matrix composed of polynomial basis functions, which contains the calculation results of all preset basis functions on the input parameters. Each column represents a basis function, and each row corresponds to a data sample. Through the basis function matrix, we can convert the nonlinear combination of input parameters into the form of matrix operations, which is convenient for subsequent linear regression analysis. Exemplarily, the basis function matrix can be expressed as

[0128] The orthogonal basis matrix can be understood as a matrix obtained by converting the basis function matrix during the orthogonalization process. Exemplarily, the basis functions in the basis function matrix are converted into a set of orthogonal bases by the Gram-Schmidt orthogonalization method or other orthogonalization techniques. Each column of the orthogonal basis matrix constitutes an orthogonal basis vector, and all orthogonal basis vectors are independent of each other in the model and are not limited here.

[0129] The weight matrix can be understood as a matrix determined by the orthogonal least squares method, which contains the contribution of each orthogonal basis vector to the output. Each weight value corresponds to a column in the orthogonal basis matrix (i.e., an orthogonal basis vector), indicating the importance of the orthogonal basis vector in the model output.

[0130] Transforming the basis function matrix of the initial nonlinear model to obtain an orthogonal basis matrix can be understood as generating a set of new, orthogonal basis functions based on the original basis function matrix through orthogonalization technology (such as the Gram-Schmidt orthogonalization process).

[0131] Determining the weight matrix of the orthogonal basis matrix based on the orthogonal least squares method can be understood as finding a set of weight parameters in the constructed orthogonal basis function model to minimize the error between the model prediction output and the actual observed data.

[0132] Adjusting the model parameters of the initial nonlinear model based on the weight matrix and the orthogonal basis matrix to obtain the target nonlinear model can be understood as using the calculated weight matrix and orthogonal basis matrix to optimize the model parameters so that the model is more in line with the actual data, thereby obtaining the target nonlinear model.

[0133] In an embodiment of the present invention, an orthogonal transformation is performed on the basis function matrix of the initial nonlinear model, the weight matrix is ​​determined by applying the orthogonal least squares method, and finally the model parameters are adjusted based on the weight matrix and the orthogonal basis matrix to obtain a target nonlinear model. This can significantly improve the performance of the model in processing complex nonlinear systems and defective data, ensuring that the model is not only accurate in prediction, but also has strong interpretability, efficient calculation, simple structure and strong generalization ability.

[0134] Optionally, adjusting the model parameters of the initial nonlinear model based on the weight matrix and the orthogonal basis matrix to obtain the target nonlinear model may include the following execution steps:

[0135] Determining an orthogonal basis function based on an orthogonal basis matrix;

[0136] The model parameters of the initial nonlinear model are adjusted based on the orthogonal basis matrix and the orthogonal basis function to obtain the target nonlinear model.

[0137] In the embodiment of the present invention, the orthogonal basis function can be understood as a group of functions in the function space, and the orthogonal basis functions satisfy the orthogonality condition.

[0138] Determining the orthogonal basis functions based on the orthogonal basis matrix can be understood as utilizing the characteristics of the orthogonal basis matrix to ensure that the final determined basis function set meets the orthogonality requirement, that is, the inner product of any two basis functions is equal to zero (except that the inner product between itself and itself is equal to a non-zero constant, usually 1).

[0139] Adjusting the model parameters of the initial nonlinear model based on the orthogonal basis matrix and the orthogonal basis function to obtain the target nonlinear model can be understood as optimizing the structure and parameters of the initial nonlinear model by the orthogonal basis matrix and the orthogonal basis function, so that it can more accurately and stably reflect the nonlinear relationship between the input variables and the output results in the complex system, thereby obtaining the target nonlinear model.

[0140] In an embodiment of the present invention, by determining the orthogonal basis functions based on the orthogonal basis matrix, and adjusting the parameters of the nonlinear model based on the orthogonal basis matrix and the orthogonal basis functions, a target nonlinear model with numerical stability, strong interpretability, avoidance of overfitting, high computational efficiency, simplified structure and high prediction accuracy can be constructed. This not only provides an effective way to optimize the nonlinear model in theory, but also demonstrates a powerful ability to process complex system data in practical applications.

[0141] The specific implementation steps of this application are:

[0142] The first step is to initialize the acceptable error e * , use Hampel filtering and support vector machine interpolation to process the sampled data.

[0143] The second step, when k = 1, execute the for loop of i: = i + 1, where i∈[1,N]. In each loop, set P1 (i) =Φ i , calculate V1 using formula (16) (i) , calculate r1 using formula (18) (i) . End the for loop.

[0144] Step 3: Let s1 = argmax{r1 (i) ,i=1,2,...,N},

[0145] Step 4: Execute the while loop, k:=k+1. In each k loop, execute the for loop i:=i+1, where i∈[1,N] and i≠s1,...,i≠s k-1 ,calculate:

[0146]

[0147]

[0148]

[0149] After the for loop ends, set

[0150] End the while loop.

[0151] Step 5: Return the orthogonal subsets P and V.

[0152] This application proposes a key parameter contribution analysis method based on polynomial basis functions, which can accurately identify the contribution of each parameter to the key performance indicators in the defective data under complex environments. Through this method, quantitative evaluation and priority sorting of key parameters can be achieved, thereby providing scientific basis and data support for subsequent transformer optimization design, operation and maintenance, and fault prediction. First, the method performs data preprocessing on the defective data, including removing outliers and filling missing values ​​to ensure the integrity and accuracy of the data. Next, the polynomial basis function method is used to construct a nonlinear model between the key performance indicators and each input parameter. This method uses polynomial basis functions to convert complex nonlinear relationships into a combination model of multiple basis functions, and establishes the association between performance indicators and their influencing parameters. Finally, the orthogonal least squares method is used to determine the weights and parameters of the model. At this point, the key performance indicators are decomposed into a weighted combination of multiple input parameter basis functions, so that the contribution of each parameter to the performance indicators can be quantitatively described. The method mainly includes three steps: data preprocessing, polynomial basis function model construction, and polynomial basis function model parameter determination. By executing the above steps, a nonlinear model between the key performance indicators and each input parameter is constructed using polynomial basis functions, and the complex nonlinear relationship is converted into a basis function combination model. The orthogonal least squares method is used to accurately determine the weights and parameters of the model, realize the association between the key performance indicators and each input parameter, and finally quantitatively describe the specific contribution of each parameter to the system performance.

[0153] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0154] In this embodiment, a data contribution analysis device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and will not be repeated here. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0155] Figure 2 is a structural block diagram of a data contribution analysis device according to one embodiment of the present invention. Figure 2 As shown, a data contribution analysis device 200 is used as an example, and the device includes: an acquisition module 201, used to acquire first power data to be processed, wherein the first power data is used to represent the perception data collected during the operation of the transformer or nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process; a processing module 202, used to perform data preprocessing on the first power data to obtain second power data; an analysis module 203, used to perform data analysis on the second power data based on a target nonlinear model to obtain an analysis result, wherein the target nonlinear model is obtained based on polynomial basis function training, and the target nonlinear model is used to determine the relationship between the target performance index and the input data, and the analysis result is used to represent the contribution of any data in the second power data to the target performance index.

[0156] Optionally, the processing module 202 is further used to perform data filtering on the first power data using a preset filtering method to obtain third power data; and to perform data completion on the third power data using a support vector machine interpolation method to obtain second power data.

[0157] Optionally, the processing module 202 is further configured to construct an initial nonlinear model based on a preset polynomial basis function; and adjust model parameters of the initial nonlinear model based on an orthogonal least squares method to obtain a target nonlinear model.

[0158] Optionally, the processing module 202 is further configured to construct an initial nonlinear model based on a preset approximation principle and a preset polynomial basis function.

[0159] Optionally, the processing module 202 is also used to transform the basis function matrix of the initial nonlinear model to obtain an orthogonal basis matrix; determine the weight matrix of the orthogonal basis matrix based on the orthogonal least squares method; and adjust the model parameters of the initial nonlinear model based on the weight matrix and the orthogonal basis matrix to obtain a target nonlinear model.

[0160] Optionally, the processing module 202 is further configured to determine an orthogonal basis function based on an orthogonal basis matrix; and adjust model parameters of the initial nonlinear model based on the orthogonal basis matrix and the orthogonal basis function to obtain a target nonlinear model.

[0161] It should be noted that the above modules can be implemented by software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.

[0162] An embodiment of the present invention further provides a computer-readable storage medium, in which a computer program is stored. The computer program is configured to execute the steps of any of the above method embodiments when running on a computer or a processor.

[0163] Optionally, in this embodiment, the computer-readable storage medium may be configured to store a computer program for performing the following steps:

[0164] Step S10, obtaining first power data to be processed, wherein the first power data is used to represent the sensed data collected during the operation of the transformer or the nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process;

[0165] Step S11, performing data preprocessing on the first power data to obtain second power data;

[0166] Step S12, performing data analysis on the second power data based on the target nonlinear model to obtain analysis results, wherein the target nonlinear model is obtained based on polynomial basis function training, the target nonlinear model is used to determine the relationship between the target performance index and the input data, and the analysis results are used to represent the contribution of any data in the second power data to the target performance index.

[0167] Optionally, in this embodiment, the above-mentioned computer-readable storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store computer programs.

[0168] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0169] Optionally, in this embodiment, the processor in the electronic device may be configured to run a computer program to perform the following steps:

[0170] Step S10, obtaining first power data to be processed, wherein the first power data is used to represent the sensed data collected during the operation of the transformer or the nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process;

[0171] Step S11, performing data preprocessing on the first power data to obtain second power data;

[0172] Step S12, performing data analysis on the second power data based on the target nonlinear model to obtain analysis results, wherein the target nonlinear model is obtained based on polynomial basis function training, the target nonlinear model is used to determine the relationship between the target performance index and the input data, and the analysis results are used to represent the contribution of any data in the second power data to the target performance index.

[0173] An embodiment of the present invention further provides a computer program product, including a computer program, which implements the steps in any of the above method embodiments when executed by a processor.

[0174] Optionally, in this embodiment, the computer program in the above computer program product may be configured to perform the following steps when executed by a processor:

[0175] Step S10, obtaining first power data to be processed, wherein the first power data is used to represent the sensed data collected during the operation of the transformer or the nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process;

[0176] Step S11, performing data preprocessing on the first power data to obtain second power data;

[0177] Step S12, performing data analysis on the second power data based on the target nonlinear model to obtain analysis results, wherein the target nonlinear model is obtained based on polynomial basis function training, the target nonlinear model is used to determine the relationship between the target performance index and the input data, and the analysis results are used to represent the contribution of any data in the second power data to the target performance index.

[0178] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0179] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0180] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0181] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0182] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0183] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0184] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, magnetic disk or optical disk, etc., various media that can store program codes.

[0185] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A data contribution analysis method, characterized in that: The method comprises: Acquire first power data to be processed, wherein the first power data is used to represent the sensed data collected during the operation of the transformer or the nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process; performing data preprocessing on the first power data to obtain second power data; The second power data is subjected to data analysis based on a target nonlinear model to obtain an analysis result, wherein the target nonlinear model is obtained based on polynomial basis function training, the target nonlinear model is used to determine the relationship between a target performance indicator and input data, and the analysis result is used to represent the contribution of any data in the second power data to the target performance indicator.

2. The method according to claim 1, characterized in that The performing data preprocessing on the first power data to obtain second power data comprises: Using a preset filtering method to perform data filtering on the first power data to obtain third power data; The third power data is completed using a support vector machine interpolation method to obtain the second power data.

3. The method according to claim 2, characterized in that The method further comprises: Constructing an initial nonlinear model based on a preset polynomial basis function; The model parameters of the initial nonlinear model are adjusted based on the orthogonal least squares method to obtain the target nonlinear model.

4. The method according to claim 3, characterized in that The constructing of the initial nonlinear model based on the preset polynomial basis function comprises: The initial nonlinear model is constructed based on a preset approximation principle and the preset polynomial basis function.

5. The method according to claim 3, characterized in that: The adjusting the model parameters of the initial nonlinear model based on the orthogonal least squares method to obtain the target nonlinear model comprises: Transforming the basis function matrix of the initial nonlinear model to obtain an orthogonal basis matrix; Determine a weight matrix of the orthogonal basis matrix based on the orthogonal least squares method; The model parameters of the initial nonlinear model are adjusted based on the weight matrix and the orthogonal basis matrix to obtain the target nonlinear model.

6. The method according to claim 5, characterized in that The adjusting the model parameters of the initial nonlinear model based on the weight matrix and the orthogonal basis matrix to obtain the target nonlinear model comprises: Determining an orthogonal basis function based on the orthogonal basis matrix; The model parameters of the initial nonlinear model are adjusted based on the orthogonal basis matrix and the orthogonal basis function to obtain the target nonlinear model.

7. A data contribution analysis device, characterized in that: The device comprises: An acquisition module, used for acquiring first power data to be processed, wherein the first power data is used to represent the sensed data collected during the operation of the transformer or the nonlinear system under preset conditions, and the preset conditions are used to represent the situation where data is missing during the data collection, transmission or storage process; a processing module, configured to perform data preprocessing on the first power data to obtain second power data; An analysis module is used to perform data analysis on the second power data based on a target nonlinear model to obtain an analysis result, wherein the target nonlinear model is obtained based on polynomial basis function training, the target nonlinear model is used to determine the relationship between the target performance index and the input data, and the analysis result is used to represent the contribution of any data in the second power data to the target performance index.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program is configured to execute the data contribution analysis method described in any one of claims 1 to 6 when running on a computer or a processor.

9. A computer program product, characterized in that It comprises a computer program which, when executed by a processor, implements the data contribution analysis method described in any one of claims 1 to 6.

10. An electronic device comprising a memory and a processor, characterized in that: The memory stores a computer program, and the processor is configured to run the computer program to execute the data contribution analysis method described in any one of claims 1 to 6.

Images