Data fusion method and device, computer equipment and storage medium

Abstract

The invention relates to a data fusion method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring to-be-fused data from at least two information sources; dividing the to-be-fused data by adopting a target support vector machine to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing a decision function in the initial support vector machine according to a relaxation parameter; the relaxation parameter comprises a relaxation variable and a penalty parameter; and fusion processing is carried out on various types of data. By adopting the method, the to-be-fused data can be accurately divided and fused.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a data fusion method, device, computer equipment and storage medium. Background Art

[0002] In recent years, information model technology has been widely used in the field of smart grids. By establishing a unified information model, data standardization and protocol unification between different terminal devices can be achieved, thereby improving the integration and operation efficiency of the system. The application of information model technology not only improves the interoperability of the system, but also enhances the accuracy and real-time performance of data processing, and optimizes resource allocation and system scheduling.

[0003] With the widespread application of renewable energy and the development of smart grid technology, the number and scale of distributed energy systems have increased rapidly. However, in the existing technology, there is a lack of a unified information model between terminal devices, which leads to obstacles in information exchange and data sharing, thus affecting the overall performance of distributed energy systems. Summary of the invention

[0004] Based on this, it is necessary to provide a data fusion method, device, computer equipment and storage medium to address the above technical problems.

[0005] In a first aspect, the present application provides a data fusion method, comprising:

[0006] Acquire data to be fused from at least two information sources;

[0007] Using a target support vector machine to divide the data to be fused to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to a relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter;

[0008] Each type of data is fused and processed separately.

[0009] In one embodiment, the target support vector machine is determined by:

[0010] Constructing an objective function based on the relaxation parameters, the normal vector variables and the bias variables;

[0011] Constructing a constraint function according to the slack variable, the bias variable, the normal vector variable and the data variable;

[0012] Using initial sample data, updating the data variables in the constraint function; wherein the initial sample data includes sample data and data labels corresponding to the sample data;

[0013] Taking the minimum function value of the objective function as the goal and the updated constraint function as the constraint condition, the objective function is solved based on the evolutionary algorithm to obtain the normal vector parameter and the bias parameter;

[0014] According to the normal vector parameters and the bias parameters, the decision function in the initial support vector machine is optimized to obtain the target support vector machine.

[0015] In one embodiment, constructing an objective function according to the relaxation parameter, the normal vector variable and the bias variable comprises:

[0016] Determine a first intermediate function according to the second norm of the normal vector variable;

[0017] The product of the slack variable and the penalty parameter is used as the second intermediate function;

[0018] The sum of the first intermediate function and the second intermediate function is used as the objective function.

[0019] In one embodiment, constructing a constraint function according to the slack variable, the bias variable, the normal vector variable and the data variable comprises:

[0020] Using a nonlinear transformation function, the data variables are converted into a multidimensional change space to obtain a multidimensional data variable;

[0021] The product of the transpose of the normal vector variable and the multidimensional data variable is used as a third intermediate function;

[0022] taking the sum of the third intermediate function and the bias variable as a fourth intermediate function;

[0023] constructing an initial function according to the fourth intermediate function and the slack variable;

[0024] The penalty function algorithm is used to optimize the initial function and obtain the constraint function.

[0025] In one embodiment, the target support vector machine is used to divide the data to be fused to obtain at least two types of data, including:

[0026] Perform data cleaning on the data to be fused to obtain processed data to be fused;

[0027] Extract the data features of the processed data to be fused to obtain the target feature vector;

[0028] Normalizing the target feature vector to obtain a normalized target feature vector;

[0029] Based on the target support vector machine, the data to be fused is divided according to the target feature vector after normalization to obtain at least two types of data.

[0030] In one embodiment, performing data cleaning on the data to be fused to obtain processed data to be fused includes:

[0031] Get the mean and standard deviation of the data to be fused;

[0032] Determine the standard score of each data in the data to be fused according to the mean, standard deviation and the data to be fused;

[0033] Determine abnormal data in the data to be fused according to each standard score and score threshold;

[0034] The abnormal data is processed to obtain processed data to be fused.

[0035] In one embodiment, extracting the data features of the processed data to be fused to obtain the target feature vector includes:

[0036] Based on the feature extraction model, extract the data features in the processed data to be fused to obtain the initial data features;

[0037] Determining a covariance matrix of the initial data features;

[0038] Performing eigendecomposition on the covariance matrix to obtain initial eigenvectors and eigenvalues ​​corresponding to each initial eigenvector;

[0039] The initial feature vector corresponding to the feature value greater than the feature threshold is used as the target feature vector.

[0040] In a second aspect, the present application also provides a data fusion device, comprising:

[0041] A data acquisition module, used to acquire data to be fused from at least two information sources;

[0042] A data partitioning module, used to use a target support vector machine to partition the data to be fused to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to a relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter;

[0043] The data fusion module is used to perform fusion processing on various types of data respectively.

[0044] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0045] Acquire data to be fused from at least two information sources;

[0046] Using a target support vector machine to divide the data to be fused to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to a relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter;

[0047] Each type of data is fused and processed separately.

[0048] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0049] Acquire data to be fused from at least two information sources;

[0050] Using a target support vector machine to divide the data to be fused to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to a relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter;

[0051] Each type of data is fused and processed separately.

[0052] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0053] Acquire data to be fused from at least two information sources;

[0054] Using a target support vector machine to divide the data to be fused to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to a relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter;

[0055] Each type of data is fused and processed separately.

[0056] The above data fusion method, device, computer equipment and storage medium obtain the data to be fused from at least two information sources; use the target support vector machine to divide the data to be fused to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to the relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter; and each type of data is fused separately. The above scheme improves the calculation accuracy of the target support vector machine by introducing the target support vector machine, that is, by optimizing the decision function in the initial support vector machine through the relaxation parameter, and at the same time improves the generalization ability of the target support vector machine, further ensures the accuracy of the division of the data to be fused by the target support vector machine, and thus improves the accuracy and efficiency of the data fusion process. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 is an application environment diagram of a data fusion method in an embodiment;

[0059] Figure 2A is a flow chart of a data fusion method in one embodiment;

[0060] Figure 2B A schematic diagram of a hyperplane in a case where linear separability of two-dimensional classification is possible in one embodiment;

[0061] Figure 3 A schematic diagram of a process for constructing a target support vector machine in one embodiment;

[0062] Figure 4 A schematic diagram of a process for constructing a constraint function in one embodiment;

[0063] Figure 5 A schematic diagram of a process of dividing the data to be fused in one embodiment;

[0064] Figure 6 A schematic diagram of a process for performing data cleaning on fused data in one embodiment;

[0065] Figure 7 A schematic diagram of a process for determining a target feature vector in one embodiment;

[0066] Figure 8 A unified information model architecture diagram of a control terminal in one embodiment;

[0067] Fig. 9 A schematic diagram of a flow chart of a data fusion method in another embodiment;

[0068] Fig.10 is a structural block diagram of a data fusion device in one embodiment;

[0069] Fig.11 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0071] The data fusion method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the information source 101 communicates with the server 102 through the network. The data storage system can store the data that the server 102 needs to process. The data storage system can be integrated on the server 102, or it can be placed on the cloud or other network servers. Optionally, the server 102 obtains the data to be fused from at least two information sources 101, and uses a target support vector machine to divide the data to be fused to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to the relaxation parameters; the relaxation parameters of the server 102 include relaxation variables and penalty parameters; and each type of data is fused separately. The server 102 can be implemented with an independent server or a server cluster consisting of multiple servers.

[0072] In one embodiment, Figure 2A As shown, a data fusion method is provided, which is applied to Figure 1 The server 102 in the example is used as an example to illustrate, and specifically includes the following steps:

[0073] S201, obtaining data to be fused from at least two information sources.

[0074] The information source is the source of data, for example, the information source may be a sensor, a database, a network, etc.

[0075] Optionally, information can be obtained from at least two sources:

[0076] S202, using a target support vector machine to divide the data to be fused into at least two categories of data.

[0077] The target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to the relaxation parameters; the relaxation parameters include relaxation variables and penalty parameters.

[0078] Optionally, the data to be fused may be input into a trained target support vector machine, so that the target support vector machine processes and analyzes the data to be fused, divides the data to be fused, and obtains at least two types of data output by the target support vector machine.

[0079] It should be noted that, in order to optimize the initial support vector machine, in the embodiment of the present application, slack variables and penalty parameters are introduced to relax the objective function and constraints in the initial support vector machine to obtain the target support vector machine.

[0080] S203, performing fusion processing on each type of data.

[0081] Optionally, a data fusion algorithm is used to fuse various types of data separately.

[0082] In the above data fusion method, the data to be fused are obtained from at least two information sources; the target support vector machine is used to divide the data to be fused to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to the relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter; and each type of data is fused separately. The above scheme improves the calculation accuracy of the target support vector machine by introducing the target support vector machine, that is, by optimizing the decision function in the initial support vector machine through the relaxation parameter, and at the same time improves the generalization ability of the target support vector machine, further ensures the accuracy of the division of the data to be fused by the target support vector machine, and thus improves the accuracy and efficiency of the data fusion process.

[0083] Optionally, as a machine learning algorithm for small samples, support vector machines are based on the principle of structural risk minimization in statistical theory. Support vector machines can effectively handle high-dimensional data and nonlinear problems by finding the best decision boundary in high-dimensional space, that is, the hyperplane that maximizes the interval between classes. The core idea is to find a hyperplane in the feature space to maximize the interval between two classes, thereby improving the accuracy and generalization ability of classification. The construction of the optimal hyperplane can be divided into two cases: linearly separable and linearly inseparable. The original support vector machine method was proposed based on the linearly separable case.

[0084] Based on the initial support vector machine, assuming there are two types of sample sets , ;in, is the total number of samples, is the spatial dimension of the sample, is the category mark of the sample.

[0085] like Figure 2B As shown in the figure, it is a schematic diagram of the hyperplane in the case of linear separability in two-dimensional classification. The hollow circle on the top represents the first category, and the solid circle on the bottom represents the second category. For the sake of uniform explanation, a boundary that divides these two categories in any dimension, whether it is a point on a one-dimensional line, a line on a two-dimensional plane, a plane in three-dimensional space, or even a hyperplane in a higher dimension, is collectively called a hyperplane. Let H be the hyperplane, represented by the normal vector, , Parallel to the hyperplane, where , The vertical length between is called the classification interval. The optimal hyperplane not only meets the requirement of correctly separating the two categories, but also must ensure that the classification interval is maximized. The expression of the hyperplane H is:

[0086] (1)

[0087] in, is the normal vector variable in the initial support vector machine; is the bias variable in the initial support vector machine. The set of linearly separable training samples ,in, Here we can list and solve for variables and The optimal solution formula is:

[0088] (2)

[0089] (3)

[0090] Among them, formula (2) is the initial objective function of the initial support vector machine; formula (3) is the initial constraint function of the initial support vector machine.

[0091] Optionally, the optimal solution is obtained through the above formula and , and construct a hyperplane , we can obtain the decision function:

[0092] (4)

[0093] However, when the training set is no longer linearly separable, the above process cannot accurately obtain the optimal hyperplane. Therefore, in the embodiment of the present application, the above formula (2) and formula (3) are relaxed and optimized by introducing relaxation parameters, that is, relaxation variables and penalty parameters, to obtain the target support vector machine. Based on this, in one embodiment, Figure 3As shown, a process for constructing a target support vector machine is provided, which specifically includes the following steps:

[0094] S301, constructing an objective function according to the relaxation parameters, the normal vector variables and the bias variables.

[0095] The objective function is the function after optimizing the above formula (2), that is, the objective function of the target support vector machine.

[0096] Optionally, the initial objective function consisting of the normal vector variable and the bias variable may be optimized by relaxing parameters according to preset rules to construct the objective function.

[0097] Optionally, in order to ensure the accuracy of the constructed objective function, in the embodiment of the present application, the first intermediate function can also be determined according to the second norm of the normal vector variable; the product of the slack variable and the penalty parameter is used as the second intermediate function; and the sum of the first intermediate function and the second intermediate function is used as the objective function. Specifically, the objective function can be expressed by the following formula (5):

[0098] (5)

[0099] in, is the penalty parameter, is the slack variable; is the number of data in the data to be fused.

[0100] S302, constructing a constraint function according to the slack variables, the bias variables, the normal vector variables and the data variables.

[0101] The constraint function is the function after optimizing the above formula (3), that is, the constraint function of the target support vector machine.

[0102] Optionally, the initial constraint function consisting of the bias variable, the normal vector variable and the data variable may be optimized through slack variables according to preset rules to construct a constraint function.

[0103] S303, using the initial sample data, updating the data variables in the constraint function.

[0104] The initial sample data includes sample data and data labels corresponding to the sample data.

[0105] S304, taking the minimum function value of the objective function as the goal and the updated constraint function as the constraint condition, solving the objective function based on the evolutionary algorithm to obtain the normal vector parameter and the bias parameter.

[0106] Among them, the evolutionary algorithm is a search algorithm inspired by the biological evolution mechanism in nature. In the embodiment of the present application, the evolutionary algorithm is used to solve the objective function to find the optimal solution of the bias variable and the normal vector variable.

[0107] Optionally, the objective function is minimized as the goal, the updated constraint function is used as the constraint condition, and the objective function is solved by an evolutionary algorithm to obtain normal vector parameters and bias parameters.

[0108] Exemplarily, the process of the evolutionary algorithm is as follows:

[0109] Optional, randomly generate a population of initial solutions , each individual represents a solution vector ; Perform mutation operations on individuals in the current population to generate new individual vectors. The basic differential evolution algorithm will adopt one of the following mutation strategies.

[0110] The first mutation strategy is to randomly select three different individuals from the population , , , generate new individuals through differential operation :

[0111] (6)

[0112] Among them, F is the scaling factor, which is used to control the degree of variation.

[0113] Another mutation strategy is to select the current best individual from the population and two random individuals , , generate new individuals :

[0114] (7)

[0115] Furthermore, the mutant individuals With the original individual Perform crossover operations to generate trial solutions :

[0116] (8)

[0117] in, is the crossover probability, which controls the probability of the mutant individual being selected. It is the problem dimension.

[0118] By comparing the solution With the original individual The fitness of the next generation of individuals is selected based on the fitness of the individuals. When the preset termination condition (such as the maximum number of iterations or fitness convergence) is reached, the optimal solution is output.

[0119] S305, optimizing the decision function in the initial support vector machine according to the normal vector parameters and the bias parameters to obtain a target support vector machine.

[0120] Optionally, the decision function in the initial support vector machine is optimized according to the normal vector parameter and the bias parameter to obtain an optimized decision function. Specifically, the optimized decision function can be expressed by the following formula (9):

[0121] (9)

[0122] In this embodiment, the initial support vector machine is optimized by relaxing parameters to obtain the objective function and constraint function; at the same time, the objective function is solved by an evolutionary algorithm to ensure the accuracy of the obtained normal vector parameters and bias parameters, further ensuring the accuracy of the constructed target support vector machine.

[0123] Optionally, in one embodiment, Figure 4 As shown, a method for constructing a constraint function is provided, which specifically includes the following steps:

[0124] S401, using a nonlinear transformation function to transform the data variable into a multi-dimensional change space to obtain a multi-dimensional data variable.

[0125] Among them, the nonlinear transformation function is a function that converts data variables into a multidimensional variation space.

[0126] Optionally, a nonlinear transformation function may be used to calculate the data variables to transform the data variables into a multidimensional variation space to obtain a multidimensional data variable.

[0127] S402, taking the transposition of the normal vector variable and the product of the multidimensional data variable as the third intermediate function.

[0128] Optionally, the process of using the transpose of the normal vector variable and the product of the multidimensional data variable as the third intermediate function can be expressed by the following formula (11):

[0129] (11)

[0130]

[0131] S403: Taking the sum of the third intermediate function and the bias variable as the fourth intermediate function.

[0132] Optionally, the process of using the sum of the third intermediate function and the bias variable as the fourth intermediate function can be expressed by the following formula (12):

[0133] (12)

[0134] S404: construct an initial function according to the fourth intermediate function and the slack variables.

[0135] Optionally, the process of constructing the initial function according to the fourth intermediate function and the slack variable can be expressed by the following formula (13):

[0136] (13)

[0137] S405, using a penalty function algorithm to optimize the initial function to obtain a constraint function.

[0138] Optionally, in order to further optimize the initial function, a penalty function algorithm can be used to convert the constraints of the initial function into a penalty function so that solutions that do not meet the constraints will be penalized. Specifically, the constraint function can be expressed by the following formula (14):

[0139] (14)

[0140] (15)

[0141] In the embodiment of the present application, the accuracy of the constructed initial function is ensured by introducing the third intermediate function and the fourth intermediate function; further, the penalty function algorithm is used to optimize the initial function, thereby improving the solution accuracy of the constraint function.

[0142] Optionally, in one embodiment, Figure 5 As shown, a method for dividing the fused data is provided, which specifically includes the following steps:

[0143] S501, performing data cleaning processing on the data to be fused to obtain processed data to be fused.

[0144] Optionally, the data to be fused obtained from multiple information sources will inevitably contain abnormal data. In order to ensure the availability of the data to be fused, it is necessary to perform data cleaning on the data to be fused to obtain the processed data to be fused.

[0145] S502, extracting data features of the processed data to be fused to obtain a target feature vector.

[0146] The target feature vector is a vector containing data features of the data to be fused.

[0147] Optionally, the data to be fused can be input into the feature extraction network through the feature extraction network to obtain the target feature vector.

[0148] S503, normalizing the target feature vector to obtain a normalized target feature vector.

[0149] Optionally, the target feature vector is normalized to convert the target feature vector into a form with a mean of 0 and a variance of 1, thereby obtaining a normalized target feature vector.

[0150] S504, based on the target support vector machine, according to the normalized target feature vector, the data to be fused is divided to obtain at least two types of data.

[0151] Optionally, the normalized target feature vector is input into a target support vector machine, so that the target support vector machine divides the data to be fused according to the target feature vector to obtain at least two types of data.

[0152] In the embodiment of the present application, by performing data cleaning processing on the data to be fused, introducing the target feature vector, and further normalizing the target feature vector, the accuracy of the subsequent target support vector machine's division process of the data to be fused is guaranteed.

[0153] Optionally, in one embodiment, Figure 6 As shown, a method for data cleaning of fused data is provided, which specifically includes the following steps:

[0154] S601, obtaining the mean and standard deviation of the data to be fused.

[0155] Optionally, statistical analysis is performed on the data to be fused to calculate the mean and standard deviation of the data to be fused.

[0156] S602: Determine the standard score of each data in the data to be fused according to the mean, the standard deviation and the data to be fused.

[0157] Among them, the standard score is a quantity with equal units, which is the quotient obtained by dividing the difference between the original score and the group mean by the standard deviation; in the embodiment of the present application, the standard score of each data is the value obtained by dividing the difference between the mean of each data and the data to be fused by the standard deviation.

[0158] Optionally, the standard score of each data in the data to be fused can be expressed by the following formula (16):

[0159] (16)

[0160] in, represents the standard score; is the data to be fused; is the mean; is the standard deviation.

[0161] S603: Determine abnormal data in the data to be fused according to each standard score and the score threshold.

[0162] The score threshold is a preset standard score threshold; for example, the score threshold may be 3.

[0163] Optionally, for any data, if the standard score of the data is greater than the score threshold, the data is regarded as abnormal data.

[0164] S604, processing the abnormal data to obtain processed data to be fused.

[0165] Optionally, abnormal data may be deleted from the data to be fused; meanwhile, for missing data in the data to be fused, interpolation may be used to fill in the missing values. For example, the mean may be used to fill in the missing values.

[0166] In the embodiment of the present application, the integrity and availability of the data to be fused are ensured by detecting abnormal data in the data to be fused according to the standard score of each data in the data to be fused and processing the abnormal data.

[0167] Optionally, in one embodiment, Figure 7 As shown, a method for determining a target feature vector is provided, which specifically includes the following steps:

[0168] S701, based on the feature extraction model, extract the data features in the processed data to be fused to obtain the initial data features.

[0169] Among them, the feature extraction model is a model for extracting features in the data to be fused; the initial data features are data features in the fused data.

[0170] Optionally, the processed data to be fused may be input into a feature extraction model, so that the feature extraction model extracts data features from the processed data to be fused to obtain initial data features.

[0171] S702, determining the covariance matrix of the initial data features.

[0172] Among them, the covariance matrix represents the correlation between the various dimensions of the initial data features.

[0173] Optionally, statistical analysis is performed on the initial data features to calculate a covariance matrix of the initial data features.

[0174] S703, performing eigendecomposition on the covariance matrix to obtain initial eigenvectors and eigenvalues ​​corresponding to each initial eigenvector.

[0175] Optionally, an eigendecomposition method is used to perform eigendecomposition on the covariance matrix to obtain initial eigenvectors and eigenvalues ​​corresponding to each initial eigenvector.

[0176] S704: Use the initial feature vector corresponding to the feature value greater than the feature threshold as the target feature vector.

[0177] The characteristic threshold is a preset characteristic value threshold.

[0178] Optionally, an initial feature vector corresponding to a feature value greater than a feature threshold may be used as a target feature vector.

[0179] In the embodiment of the present application, the accuracy of the determined target feature vector is ensured by introducing the initial data features and the covariance matrix of the initial data features.

[0180] Optionally, based on the data fusion method provided in the embodiment of the present application, taking the smart grid as the application scenario and each control terminal in the smart grid as the information source, an application example of applying the data fusion method to the unified information model architecture of the distributed interactive resource control terminal is provided. Figure 8 The unified information model architecture of the control terminal is shown in FIG. The unified information model architecture of the distributed interactive resource control terminal is logically divided into six five-level layers.

[0181] At the physical layer, detailed physical properties of various types of equipment in the distributed energy system are defined, including information such as the manufacturer, installation date, and operating status of the equipment. These properties provide a comprehensive description of the performance and operating status of the equipment. The geographical location of each device, accurate to the longitude and latitude, ensures the accurate positioning and management of the equipment in space. In addition, the definition of power levels and electrical characteristics allows the system to perform precise regulation and energy optimization based on the technical specifications of the equipment. All of this information is identified by a unique object code, ensuring that each physical device can be uniquely identified in the entire system, thereby achieving accurate identification and management of equipment within the system, providing a solid foundation for the stable operation and efficient scheduling of distributed energy systems.

[0182] In the communication layer, the data exchange protocol between various devices is first established, and international standardized protocols such as IEC 61850 and DL / T 860 are adopted to ensure the standardization and interoperability of data transmission and information exchange between devices inside and outside the system. This standardized protocol enables devices produced by different manufacturers to communicate under a unified data format and communication specification, effectively reducing compatibility issues that may arise during system integration and operation. In terms of communication parameter settings, the communication layer defines the specific communication parameters of each device, including IP (Internet Protocol) address, port number, transmission rate, etc. Through unified parameter settings, the system can ensure stable connection and efficient data transmission between devices. The allocation of IP addresses and port numbers of devices is managed according to the network topology and security policies to ensure the security and reliability of communication. In addition, the communication layer also covers the communication protocol adaptation and upgrade mechanism of the device. With the development of technology and changes in system requirements, communication protocols may need to be regularly upgraded or adapted to new communication technologies. The communication layer is responsible for planning and implementing updates to these protocols to ensure the long-term reliability and sustainable development of the system.

[0183] In the service layer, the interfaces and functions of multiple core services are defined. These services include data acquisition, status monitoring, fault detection and diagnosis, and energy management. By adopting a unified interface standard, the service layer ensures efficient interoperability between various services and the overall integration of the system. This not only helps real-time data processing and decision support, but also improves the system's ability to optimize energy consumption and the accuracy of fault prediction, providing reliable technical support for the stable operation and intelligent management of distributed energy systems.

[0184] At the application layer, a variety of core application scenarios and business logics are defined, including energy scheduling and optimization, demand response, energy trading, and forecasting analysis. The application layer implements optimized control and efficient management of the system by calling various services at the service layer through a unified information model interface. These applications not only improve energy utilization efficiency and market responsiveness through real-time data analysis and intelligent algorithms, but also promote the market-oriented operation and economic benefits of the energy market. Intelligent management and continuous optimization of the application layer provide solid support for the stable operation and sustainable development of distributed energy systems in a complex and changing energy environment.

[0185] In the security layer, first, the encryption mechanism of data is designed to ensure that all sensitive data can be effectively protected during transmission and storage. Advanced encryption algorithms and technologies are used to encrypt data end-to-end to prevent data leakage and tampering. Secondly, the security layer manages and limits the access rights of different users and devices in the system through strict access control mechanisms. This includes role-based access control to ensure that only authorized users and devices can access specific resources and functions. Identity authentication is also an important part of the security layer. Through advanced authentication methods such as multi-factor authentication, the legitimacy of the identity of users and devices is verified to prevent unauthorized access and identity forgery. In addition, the audit tracking function helps the security layer monitor the operation and access activities of the system, record and analyze the logs of key operations, and promptly discover and respond to potential security threats and abnormal behaviors. The security layer adopts standardized security protocols and mechanisms, such as OAuth open authorization, to ensure the security of data transmission and the legitimacy of user access. Through these security measures, the security layer not only protects the system from malicious attacks and data leakage, but also enhances the system's anti-attack capability and stability, providing a solid security foundation for the sustainable development of distributed energy systems.

[0186] Fig. 9 FIG. 2 is a flow chart of a data fusion method in another embodiment. Based on the above embodiment, this embodiment provides an optional example of a data fusion method. Fig. 9 The specific implementation process is as follows:

[0187] S901, constructing an objective function according to the relaxation parameters, the normal vector variables and the bias variables.

[0188] Optionally, a first intermediate function is determined based on the second norm of the normal vector variable; the product of the slack variable and the penalty parameter is used as the second intermediate function; and the sum of the first intermediate function and the second intermediate function is used as the objective function.

[0189] S902, constructing a constraint function according to the slack variables, the bias variables, the normal vector variables and the data variables.

[0190] Optionally, a nonlinear transformation function is used to convert data variables into a multidimensional change space to obtain multidimensional data variables; the transpose of the normal vector variable and the product of the multidimensional data variable are used as the third intermediate function; the sum of the third intermediate function and the bias variable is used as the fourth intermediate function; an initial function is constructed based on the fourth intermediate function and the slack variable; and a penalty function algorithm is used to optimize the initial function to obtain a constraint function.

[0191] S903, using the initial sample data, updating the data variables in the constraint function.

[0192] The initial sample data includes sample data and data labels corresponding to the sample data.

[0193] S904, taking the minimum function value of the objective function as the goal and the updated constraint function as the constraint condition, solving the objective function based on the evolutionary algorithm to obtain the normal vector parameter and the bias parameter.

[0194] S905, optimizing the decision function in the initial support vector machine according to the normal vector parameter and the bias parameter to obtain a target support vector machine.

[0195] S906, performing data cleaning processing on the data to be fused to obtain processed data to be fused.

[0196] Optionally, obtain the mean and standard deviation of the data to be fused; determine the standard score of each data in the data to be fused based on the mean, standard deviation and the data to be fused; determine the abnormal data in the data to be fused based on each standard score and the score threshold; process the abnormal data to obtain the processed data to be fused.

[0197] S907, extracting the data features of the processed data to be fused to obtain a target feature vector.

[0198] Optionally, based on a feature extraction model, data features in the processed data to be fused are extracted to obtain initial data features; the covariance matrix of the initial data features is determined; the covariance matrix is ​​eigen-decomposed to obtain initial eigenvectors and eigenvalues ​​corresponding to each initial eigenvector; and the initial eigenvector corresponding to the eigenvalue greater than the feature threshold is used as the target eigenvector.

[0199] S908, normalizing the target feature vector to obtain a normalized target feature vector.

[0200] S909, based on the target support vector machine, according to the normalized target feature vector, the data to be fused is divided to obtain at least two types of data.

[0201] S910, performing fusion processing on various types of data respectively.

[0202] The specific process of S901-S910 above can refer to the description of the above method embodiment, and its implementation principle and technical effect are similar, which will not be repeated here.

[0203] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0204] Based on the same inventive concept, the embodiment of the present application also provides a data fusion device for implementing the data fusion method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more data fusion device embodiments provided below can refer to the limitations on the data fusion method above, and will not be repeated here.

[0205] In one embodiment, Fig.10 As shown, a data fusion device 1000 is provided, comprising: a data acquisition module 1010, a data division module 1020 and a data fusion module 1030, wherein:

[0206] The data acquisition module 1010 is used to acquire data to be fused from at least two information sources.

[0207] The data partitioning module 1020 is used to use a target support vector machine to partition the fused data to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to the relaxation parameters; the relaxation parameters include relaxation variables and penalty parameters.

[0208] The data fusion module 1030 is used to perform fusion processing on various types of data respectively.

[0209] The above-mentioned data fusion device obtains the data to be fused from at least two information sources; uses the target support vector machine to divide the data to be fused to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to the relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter; and each type of data is fused separately. The above-mentioned scheme improves the calculation accuracy of the target support vector machine by introducing the target support vector machine, that is, by optimizing the decision function in the initial support vector machine through the relaxation parameter, and at the same time improves the generalization ability of the target support vector machine, further ensures the accuracy of the division of the data to be fused by the target support vector machine, and thus improves the accuracy and efficiency of the data fusion process.

[0210] In one embodiment, the data fusion device 1000 further includes:

[0211] Function building module, used to build the objective function based on the relaxation parameters, normal vector variables and bias variables.

[0212] A constraint building module, used to build constraint functions based on slack variables, bias variables, normal vector variables and data variables;

[0213] The data updating module is used to update the data variables in the constraint function using the initial sample data; wherein the initial sample data includes the sample data and the data labels corresponding to the sample data.

[0214] The variable solving module is used to solve the objective function based on the evolutionary algorithm with the minimum function value of the objective function as the goal and the updated constraint function as the constraint condition to obtain the normal vector parameters and bias parameters.

[0215] The vector machine determination module is used to optimize the decision function in the initial support vector machine according to the normal vector parameters and the bias parameters to obtain the target support vector machine.

[0216] In one embodiment, the function building module is specifically used to:

[0217] According to the second norm of the normal vector variable, a first intermediate function is determined; the product of the slack variable and the penalty parameter is used as the second intermediate function; and the sum of the first intermediate function and the second intermediate function is used as the objective function.

[0218] In one embodiment, the constraint building module is specifically used to:

[0219] A nonlinear transformation function is used to transform the data variables into a multidimensional change space to obtain a multidimensional data variable; the transpose of the normal vector variable and the product of the multidimensional data variable are used as the third intermediate function; the sum of the third intermediate function and the bias variable is used as the fourth intermediate function; an initial function is constructed based on the fourth intermediate function and the slack variable; a penalty function algorithm is used to optimize the initial function to obtain a constraint function.

[0220] In one embodiment, the data partitioning module 1020 includes:

[0221] The data processing unit is used to perform data cleaning on the data to be fused to obtain processed data to be fused.

[0222] The feature extraction unit is used to extract the data features of the processed data to be fused to obtain the target feature vector.

[0223] The feature processing unit is used to normalize the target feature vector to obtain the normalized target feature vector.

[0224] The data partitioning unit is used to partition the data to be fused based on the target support vector machine and the target feature vector after normalization to obtain at least two types of data.

[0225] In one embodiment, the data processing unit is specifically used for:

[0226] Obtain the mean and standard deviation of the data to be fused; determine the standard score of each data in the data to be fused according to the mean, standard deviation and the data to be fused; determine the abnormal data in the data to be fused according to each standard score and score threshold; process the abnormal data to obtain the processed data to be fused.

[0227] In one embodiment, the feature extraction unit is specifically used for:

[0228] Based on the feature extraction model, the data features in the processed data to be fused are extracted to obtain the initial data features; the covariance matrix of the initial data features is determined; the covariance matrix is ​​eigen-decomposed to obtain the initial eigenvectors and the eigenvalues ​​corresponding to each initial eigenvector; the initial eigenvector corresponding to the eigenvalue greater than the feature threshold is used as the target eigenvector.

[0229] Each module in the above data fusion device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0230] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Fig.11 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data to be fused obtained from at least two information sources. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a data fusion method is implemented.

[0231] Those skilled in the art will understand that Fig.11 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0232] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0233] Acquire data to be fused from at least two information sources;

[0234] The target support vector machine is used to divide the fused data to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to the relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter;

[0235] Each type of data is fused and processed separately.

[0236] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0237] According to the slack parameters, normal vector variables and bias variables, an objective function is constructed; according to the slack variables, bias variables, normal vector variables and data variables, a constraint function is constructed; the data variables in the constraint function are updated using the initial sample data; wherein the initial sample data includes the sample data and the data labels corresponding to the sample data; with the minimum function value of the objective function as the goal and the updated constraint function as the constraint condition, the objective function is solved based on the evolutionary algorithm to obtain the normal vector parameters and bias parameters; according to the normal vector parameters and bias parameters, the decision function in the initial support vector machine is optimized to obtain the target support vector machine.

[0238] In one embodiment, when the processor executes the computer program to construct the objective function according to the relaxation parameters, the normal vector variables and the bias variables, the following steps are also implemented:

[0239] According to the second norm of the normal vector variable, a first intermediate function is determined; the product of the slack variable and the penalty parameter is used as the second intermediate function; and the sum of the first intermediate function and the second intermediate function is used as the objective function.

[0240] In one embodiment, when the processor executes the computer program to construct the constraint function according to the slack variables, the bias variables, the normal vector variables and the data variables, the following steps are also implemented:

[0241] A nonlinear transformation function is used to transform the data variables into a multidimensional change space to obtain a multidimensional data variable; the transpose of the normal vector variable and the product of the multidimensional data variable are used as the third intermediate function; the sum of the third intermediate function and the bias variable is used as the fourth intermediate function; an initial function is constructed based on the fourth intermediate function and the slack variable; a penalty function algorithm is used to optimize the initial function to obtain a constraint function.

[0242] In one embodiment, when the processor executes the computer program to use the target support vector machine to divide the fused data to obtain at least two types of data, the processor further implements the following steps:

[0243] The data to be fused is cleaned to obtain processed data to be fused; data features of the processed data to be fused are extracted to obtain a target feature vector; the target feature vector is normalized to obtain a normalized target feature vector; based on a target support vector machine, the data to be fused is divided according to the normalized target feature vector to obtain at least two types of data.

[0244] In one embodiment, when the processor executes the computer program to perform data cleaning on the data to be fused and obtains the processed data to be fused, the processor further implements the following steps:

[0245] Obtain the mean and standard deviation of the data to be fused; determine the standard score of each data in the data to be fused according to the mean, standard deviation and the data to be fused; determine the abnormal data in the data to be fused according to each standard score and score threshold; process the abnormal data to obtain the processed data to be fused.

[0246] In one embodiment, when the processor executes the computer program to extract the data features of the processed data to be fused and obtains the target feature vector, the processor also implements the following steps:

[0247] Based on the feature extraction model, the data features in the processed data to be fused are extracted to obtain the initial data features; the covariance matrix of the initial data features is determined; the covariance matrix is ​​eigen-decomposed to obtain the initial eigenvectors and the eigenvalues ​​corresponding to each initial eigenvector; the initial eigenvector corresponding to the eigenvalue greater than the feature threshold is used as the target eigenvector.

[0248] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0249] Acquire data to be fused from at least two information sources;

[0250] The target support vector machine is used to divide the fused data to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to the relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter;

[0251] Each type of data is fused and processed separately.

[0252] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0253] According to the slack parameters, normal vector variables and bias variables, an objective function is constructed; according to the slack variables, bias variables, normal vector variables and data variables, a constraint function is constructed; the data variables in the constraint function are updated using the initial sample data; wherein the initial sample data includes the sample data and the data labels corresponding to the sample data; with the minimum function value of the objective function as the goal and the updated constraint function as the constraint condition, the objective function is solved based on the evolutionary algorithm to obtain the normal vector parameters and bias parameters; according to the normal vector parameters and bias parameters, the decision function in the initial support vector machine is optimized to obtain the target support vector machine.

[0254] In one embodiment, when the processor executes the computer program to construct the objective function according to the relaxation parameters, the normal vector variables and the bias variables, the following steps are also implemented:

[0255] According to the second norm of the normal vector variable, a first intermediate function is determined; the product of the slack variable and the penalty parameter is used as the second intermediate function; and the sum of the first intermediate function and the second intermediate function is used as the objective function.

[0256] In one embodiment, when the processor executes the computer program to construct the constraint function according to the slack variables, the bias variables, the normal vector variables and the data variables, the following steps are also implemented:

[0257] A nonlinear transformation function is used to transform the data variables into a multidimensional change space to obtain a multidimensional data variable; the transpose of the normal vector variable and the product of the multidimensional data variable are used as the third intermediate function; the sum of the third intermediate function and the bias variable is used as the fourth intermediate function; an initial function is constructed based on the fourth intermediate function and the slack variable; a penalty function algorithm is used to optimize the initial function to obtain a constraint function.

[0258] In one embodiment, when the processor executes the computer program to use the target support vector machine to divide the fused data to obtain at least two types of data, the processor further implements the following steps:

[0259] The data to be fused is cleaned to obtain processed data to be fused; data features of the processed data to be fused are extracted to obtain a target feature vector; the target feature vector is normalized to obtain a normalized target feature vector; based on a target support vector machine, the data to be fused is divided according to the normalized target feature vector to obtain at least two types of data.

[0260] In one embodiment, when the processor executes the computer program to perform data cleaning on the data to be fused and obtains the processed data to be fused, the processor further implements the following steps:

[0261] Obtain the mean and standard deviation of the data to be fused; determine the standard score of each data in the data to be fused according to the mean, standard deviation and the data to be fused; determine the abnormal data in the data to be fused according to each standard score and score threshold; process the abnormal data to obtain the processed data to be fused.

[0262] In one embodiment, when the processor executes the computer program to extract the data features of the processed data to be fused and obtains the target feature vector, the processor also implements the following steps:

[0263] Based on the feature extraction model, the data features in the processed data to be fused are extracted to obtain the initial data features; the covariance matrix of the initial data features is determined; the covariance matrix is ​​eigen-decomposed to obtain the initial eigenvectors and the eigenvalues ​​corresponding to each initial eigenvector; the initial eigenvector corresponding to the eigenvalue greater than the feature threshold is used as the target eigenvector.

[0264] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0265] Acquire data to be fused from at least two information sources;

[0266] The target support vector machine is used to divide the fused data to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to the relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter;

[0267] Each type of data is fused and processed separately.

[0268] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0269] According to the slack parameters, normal vector variables and bias variables, an objective function is constructed; according to the slack variables, bias variables, normal vector variables and data variables, a constraint function is constructed; the data variables in the constraint function are updated using the initial sample data; wherein the initial sample data includes the sample data and the data labels corresponding to the sample data; with the minimum function value of the objective function as the goal and the updated constraint function as the constraint condition, the objective function is solved based on the evolutionary algorithm to obtain the normal vector parameters and bias parameters; according to the normal vector parameters and bias parameters, the decision function in the initial support vector machine is optimized to obtain the target support vector machine.

[0270] In one embodiment, when the processor executes the computer program to construct the objective function according to the relaxation parameters, the normal vector variables and the bias variables, the following steps are also implemented:

[0271] According to the second norm of the normal vector variable, a first intermediate function is determined; the product of the slack variable and the penalty parameter is used as the second intermediate function; and the sum of the first intermediate function and the second intermediate function is used as the objective function.

[0272] In one embodiment, when the processor executes the computer program to construct the constraint function according to the slack variables, the bias variables, the normal vector variables and the data variables, the following steps are also implemented:

[0273] A nonlinear transformation function is used to transform the data variables into a multidimensional change space to obtain a multidimensional data variable; the transpose of the normal vector variable and the product of the multidimensional data variable are used as the third intermediate function; the sum of the third intermediate function and the bias variable is used as the fourth intermediate function; an initial function is constructed based on the fourth intermediate function and the slack variable; a penalty function algorithm is used to optimize the initial function to obtain a constraint function.

[0274] In one embodiment, when the processor executes the computer program to use the target support vector machine to divide the fused data to obtain at least two types of data, the processor further implements the following steps:

[0275] The data to be fused is cleaned to obtain processed data to be fused; data features of the processed data to be fused are extracted to obtain a target feature vector; the target feature vector is normalized to obtain a normalized target feature vector; based on a target support vector machine, the data to be fused is divided according to the normalized target feature vector to obtain at least two types of data.

[0276] In one embodiment, when the processor executes the computer program to perform data cleaning on the data to be fused and obtains the processed data to be fused, the processor further implements the following steps:

[0277] Obtain the mean and standard deviation of the data to be fused; determine the standard score of each data in the data to be fused according to the mean, standard deviation and the data to be fused; determine the abnormal data in the data to be fused according to each standard score and score threshold; process the abnormal data to obtain the processed data to be fused.

[0278] In one embodiment, when the processor executes the computer program to extract the data features of the processed data to be fused and obtains the target feature vector, the processor also implements the following steps:

[0279] Based on the feature extraction model, the data features in the processed data to be fused are extracted to obtain the initial data features; the covariance matrix of the initial data features is determined; the covariance matrix is ​​eigen-decomposed to obtain the initial eigenvectors and the eigenvalues ​​corresponding to each initial eigenvector; the initial eigenvector corresponding to the eigenvalue greater than the feature threshold is used as the target eigenvector.

[0280] It should be noted that the data involved in this application (including but not limited to data used for analysis, storage, display, etc.) are all information and data fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0281] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0282] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0283] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A data fusion method, characterized in that: The method comprises: Acquire data to be fused from at least two information sources; Using a target support vector machine to divide the data to be fused to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to a relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter; Each type of data is fused and processed separately.

2. The method according to claim 1, characterized in that The target support vector machine is determined by: Constructing an objective function based on the relaxation parameters, the normal vector variables and the bias variables; Constructing a constraint function according to the slack variable, the bias variable, the normal vector variable and the data variable; Using initial sample data, updating the data variables in the constraint function; wherein the initial sample data includes sample data and data labels corresponding to the sample data; Taking the minimum function value of the objective function as the goal and the updated constraint function as the constraint condition, the objective function is solved based on the evolutionary algorithm to obtain the normal vector parameter and the bias parameter; According to the normal vector parameters and the bias parameters, the decision function in the initial support vector machine is optimized to obtain the target support vector machine.

3. The method according to claim 2, characterized in that The constructing of the objective function according to the relaxation parameter, the normal vector variable and the bias variable comprises: Determine a first intermediate function according to the second norm of the normal vector variable; The product of the slack variable and the penalty parameter is used as the second intermediate function; The sum of the first intermediate function and the second intermediate function is used as the objective function.

4. The method according to claim 2, characterized in that: The step of constructing a constraint function according to the slack variable, the bias variable, the normal vector variable and the data variable comprises: Using a nonlinear transformation function, the data variables are converted into a multidimensional change space to obtain a multidimensional data variable; The product of the transpose of the normal vector variable and the multidimensional data variable is used as a third intermediate function; taking the sum of the third intermediate function and the bias variable as a fourth intermediate function; constructing an initial function according to the fourth intermediate function and the slack variable; The penalty function algorithm is used to optimize the initial function and obtain the constraint function.

5. The method according to claim 1, characterized in that The target support vector machine is used to divide the data to be fused to obtain at least two types of data, including: Perform data cleaning on the data to be fused to obtain processed data to be fused; Extract the data features of the processed data to be fused to obtain the target feature vector; Normalizing the target feature vector to obtain a normalized target feature vector; Based on the target support vector machine, the data to be fused is divided according to the target feature vector after normalization to obtain at least two types of data.

6. The method according to claim 5, characterized in that The data to be fused is cleaned to obtain processed data to be fused, including: Get the mean and standard deviation of the data to be fused; Determine a standard score for each data in the data to be fused according to the mean, the standard deviation and each data in the data to be fused; Determine abnormal data in the data to be fused according to each standard score and score threshold; The abnormal data is processed to obtain processed data to be fused.

7. The method according to claim 5, characterized in that The data features of the data to be fused after extraction and processing are obtained to obtain a target feature vector, including: Based on the feature extraction model, the data features of the processed data to be fused are extracted to obtain the initial data features; Determining a covariance matrix of the initial data features; Performing eigendecomposition on the covariance matrix to obtain at least two initial eigenvectors and eigenvalues ​​corresponding to the initial eigenvectors; The initial feature vector corresponding to the feature value greater than the feature threshold is used as the target feature vector.

8. A data fusion device, characterized in that: The device comprises: A data acquisition module, used to acquire data to be fused from at least two information sources; A data partitioning module, used to use a target support vector machine to partition the data to be fused to obtain at least two types of data; wherein the target support vector machine is obtained by optimizing the decision function in the initial support vector machine according to a relaxation parameter; the relaxation parameter includes a relaxation variable and a penalty parameter; The data fusion module is used to perform fusion processing on various types of data respectively.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images