Load side electricity consumption behavior probability prediction method, system, terminal and medium based on unscented transformation optimization random forest algorithm

Abstract

The application discloses a load-side electricity consumption behavior probability prediction method and system based on unscented transformation optimization random forest algorithm, a terminal and a medium, and relates to the technical field of electricity consumption behavior prediction. The method comprises the following steps: obtaining relevant parameters affecting the electricity consumption behavior of a load-side user, establishing a training data set and constructing a random forest model, and training the random forest algorithm model by using the training data set; converting the given relevant parameter prediction information affecting the future electricity consumption behavior of the user into deterministic sampling points by using unscented transformation; performing deterministic calculation on the sampling points by using the trained random forest model, obtaining the prediction value output variable corresponding to each sampling point, and obtaining the probability prediction information corresponding to the electricity consumption behavior of the load-side user by using the prediction value output variable and the corresponding weight coefficient. The application uses unscented transformation optimization random forest algorithm to convert the user electricity consumption behavior description problem under future uncertain conditions into a deterministic calculation problem of a small number of sampling points, thereby improving the prediction and perception ability for the future user electricity consumption behavior.

Description

Technical Field

[0001] This invention relates to the field of digital control technology, and in particular to a method, system, terminal and medium for predicting the probability of load-side electricity consumption behavior based on an unscented transform optimized random forest algorithm. Background Technology

[0002] The doubling of installed capacity of new energy sources, coupled with their strong random fluctuations, has significantly increased the pressure on power system peak shaving and absorption, necessitating more abundant regulation resources to ensure the safe, economical, and stable operation of the power system. The continuous expansion of diverse and flexible resources on the user side, such as electric vehicles, distributed wind and solar power, distributed energy storage, user-side combined cooling, heating, and power (CCHP) units, and controllable production line loads, has brought massive potential regulation resources to the power system. Accurately predicting user electricity consumption behavior is a prerequisite for optimized resource scheduling and accurate control, and it is also an urgent problem to be solved to improve the safety and economy of power system operation.

[0003] Predicting load-side electricity consumption behavior requires acquiring information such as meteorological data, user product order information, electricity spot market prices, socio-economic environment, and the impact of low-carbon and environmental protection assessment factors. This information, combined with users' historical energy consumption data, necessitates a correlation analysis between user energy consumption behavior and these influencing factors, constructing a correlation model. This model allows for the prediction of key electricity consumption patterns, such as peak electricity demand, given the future estimated values ​​of each influencing factor. Considering the randomness of these influencing factors and the difficulty of accurate prediction, using probability distribution methods can facilitate operators' description of the predicted amounts of related information.

[0004] In existing technologies, traditional methods such as time series analysis and regression analysis are generally used, or modern intelligent algorithms such as support vector machines, neural networks, and random forests are employed. Among these methods, random forests, by constructing multiple decision trees and integrating their results, can handle high-dimensional features, are insensitive to missing values, can assess feature importance, and have strong generalization ability and algorithm interpretability, making them a relatively excellent regression model. However, traditional random forest algorithms can only provide deterministic predictions and cannot handle input quantities described in probabilistic form.

[0005] Therefore, existing technologies still have shortcomings. Summary of the Invention

[0006] The technical problem to be solved by this invention is to provide a method, system, terminal, and medium for predicting the probability of load-side electricity consumption behavior based on the unscented transform optimized random forest algorithm, addressing the aforementioned deficiencies of the prior art. The technical solution adopted by this invention is as follows: In a first aspect, the present invention provides a method for probabilistic prediction of load-side electricity consumption behavior based on an unscented transform optimized random forest algorithm, wherein the method includes: Acquire historical electricity consumption data at user terminals that reflect load-side electricity consumption behavior, as well as meteorological information, user product order information, electricity spot market prices, socio-economic environment, and low-carbon environmental protection assessment factors that affect user-side energy consumption behavior, and construct a training dataset. A random forest model is constructed based on the random forest algorithm; wherein, the random forest model uses decision trees obtained by sampling through the Bagging algorithm as the basic unit, and obtains the output value of the random forest regression model by fitting the output value of the decision tree model; The random forest model is trained based on the unscented transformation method, and during training, the future expectations of the relevant factors affecting users' energy consumption behavior are converted into deterministic sampling points and corresponding weight values. The pre-trained random forest model is used to perform deterministic calculations on the sampling points to obtain the predicted output variables corresponding to each sampling point. Based on the predicted output variables and the corresponding weight coefficients, a fitting is performed to obtain the probabilistic prediction information of the electricity consumption behavior of the corresponding load-side users.

[0007] In one implementation, a random forest model is constructed based on the random forest algorithm, including: Each time, n training samples are taken from the training dataset with replacement to form a new training set. Each training sample has d features. Each time, only k features are selected, and k≤d, to build a decision tree model. The method of repeatedly generating decision tree models generates multiple decision trees, and each decision tree produces a decision result; By fitting the decision results of multiple decision trees, the final prediction analysis results of the decision forest are obtained, which is the output value of the random forest regression model.

[0008] In one implementation, the random forest model is constructed based on the random forest algorithm, and further includes: In the input space where the training dataset is located, each region is recursively divided into two sub-regions and the output value on each sub-region is determined to construct a binary decision tree.

[0009] In one implementation, during training, the expected future outcomes of relevant factors influencing user energy consumption behavior are converted into deterministic sampling points and corresponding weight values, including: Identify a set of statistical characteristics that reflect the factors influencing user energy consumption behavior. N σ Each sampling point, i.e., a sigma point, and the corresponding weight for each sigma point. W i The weights of the sigma sampling points satisfy the normalization condition, i.e. Furthermore, the sigma point must capture some statistical characteristics of a, and the mean of x must be taken as a constant.μ x Covariance P xx Represented by sigma points.

[0010] In one implementation, during training, the expected future values ​​of relevant factors influencing user energy consumption behavior are converted into deterministic sampling points and corresponding weight values. This also includes: The symmetric sampling method is adopted. For the known future expected values ​​x of the relevant factors affecting users' energy consumption behavior, and x is an m-dimensional variable, where m reflects the number of relevant factors affecting users' energy consumption behavior; If only the mean and covariance of the user's electricity consumption behavior output variables are needed, then select... If you need to obtain the skewness coefficient and kurtosis coefficient of the output variable, you can select... .

[0011] In one implementation, let the sigma point... ,satisfy , , Where i = 1, 2, ..., m, For matrix The element in the i-th column is an m-dimensional column vector, and the required matrix can be obtained through the Cholesky transformation. ; The weight value for the sampling point is set to... , , , where i = 1, 2, ..., m, where Values ​​are set by humans.

[0012] In one implementation, the predicted value output variable corresponding to each sampling point is obtained. y i mean μ y and variance P y They are respectively , ,in y_ express y i The arithmetic mean, i.e. User-side electricity consumption behavior prediction based on unscented transformation only uses a finite number of points to describe the second-order probability prediction problem of mean and covariance; to obtain skewness coefficient and kurtosis coefficient, only a higher-order unscented transformation needs to be selected.

[0013] Secondly, embodiments of the present invention also provide a load-side electricity consumption behavior probabilistic prediction system based on the unscented transform optimized random forest algorithm. The system is used to implement the steps of the load-side electricity consumption behavior probabilistic prediction method based on the unscented transform optimized random forest algorithm described in the above scheme. The system includes: The front-end data acquisition module is used to acquire historical electricity consumption information of users that reflects the electricity consumption behavior of the load side, as well as meteorological information, user product order information, electricity spot market price, socio-economic environment, low-carbon and environmental protection assessment factors that affect the energy consumption behavior of users, and to build a training dataset. The random forest construction module is used to construct a random forest model based on the random forest algorithm. The random forest model uses decision trees obtained by sampling through the Bagging algorithm as the basic unit, and obtains the output value of the random forest regression model by fitting the output value of the decision tree model. The Unscented Transformation module is used to train a random forest model based on the Unscented Transformation method, and during training, it converts the future expectations of relevant factors affecting users' energy consumption behavior into deterministic sampling points and corresponding weight values. The random forest prediction module is used to perform deterministic calculations on the sampling points using a pre-trained random forest model to obtain the predicted output variables corresponding to each sampling point. Based on the predicted output variables and corresponding weight coefficients, the module performs fitting to obtain probabilistic prediction information of the electricity consumption behavior of the corresponding load-side users.

[0014] Thirdly, embodiments of the present invention also provide a terminal, wherein the terminal includes a memory, a processor, and a load-side electricity consumption behavior probability prediction program based on the unscented transform optimized random forest algorithm stored in the memory and executable on the processor. When the processor executes the load-side electricity consumption behavior probability prediction program based on the unscented transform optimized random forest algorithm, it implements the steps of the load-side electricity consumption behavior probability prediction method based on the unscented transform optimized random forest algorithm in any of the above schemes.

[0015] Fourthly, embodiments of the present invention also provide a computer-readable storage medium, wherein the computer-readable storage medium stores a load-side electricity consumption behavior probability prediction program based on the unscented transform optimized random forest algorithm, and the load-side electricity consumption behavior probability prediction program based on the unscented transform optimized random forest algorithm implements the steps of the load-side electricity consumption behavior probability prediction method based on the unscented transform optimized random forest algorithm as described in any of the above schemes on the computer-readable storage medium.

[0016] Beneficial Effects: Compared with existing technologies, this invention provides a probabilistic prediction method for load-side electricity consumption behavior based on an unscented transform optimized random forest algorithm. First, this invention acquires historical electricity consumption information at user terminals reflecting load-side electricity consumption behavior, as well as meteorological information, user product order information, electricity spot market prices, socio-economic environment, and low-carbon environmental protection assessment factors affecting user energy consumption behavior, and constructs a training dataset. Then, based on the random forest algorithm, a random forest model is constructed; wherein the random forest model uses decision trees obtained through Bagging algorithm sampling as basic units, and obtains the output value of the random forest regression model by fitting the output values ​​of the decision tree model. Next, the random forest model is trained based on the unscented transform method, and during training, the future expectations of relevant influencing factors affecting user energy consumption behavior are converted into deterministic sampling points and corresponding weight values. Finally, the trained random forest model is used to perform deterministic calculations on the sampling points to obtain the predicted output variables corresponding to each sampling point. Based on the predicted output variables and corresponding weight coefficients, fitting is performed to obtain the probabilistic prediction information of the corresponding load-side user electricity consumption behavior. This invention uses unscented transformation to optimize the random forest algorithm, transforming the problem of describing user electricity consumption behavior under uncertain future conditions into a deterministic computation problem with a small number of sampling points, thereby improving the ability to predict and perceive future user electricity consumption behavior. Attached Figure Description

[0017] Figure 1 This is a flowchart of a preferred embodiment of the load-side electricity consumption behavior probability prediction method based on the unscented transform optimized random forest algorithm provided in this invention.

[0018] Figure 2 The flowchart illustrates the practical application of the load-side electricity consumption behavior probability prediction method based on the unscented transformation optimized random forest algorithm provided in this embodiment of the invention.

[0019] Figure 3 The diagram below illustrates the principle of a load-side electricity consumption behavior probability prediction system based on an unscented transform optimized random forest algorithm, as provided in an embodiment of the present invention.

[0020] Figure 4 A schematic diagram of the terminal provided in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and effects of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0022] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content, operations, or steps, nor does it require execution in the described order. For example, some operations or steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0023] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0024] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. For example, the first control information and the second control information are only used to distinguish different control information and do not limit their order.

[0025] Those skilled in the art will understand that the words "first" and "second" do not limit the quantity or the order of execution, and that the words "first" and "second" do not necessarily imply that they are different.

[0026] It should also be understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0027] To address the problems of existing technologies, this invention provides a probabilistic prediction method for load-side electricity consumption behavior based on an unscented transform optimized random forest algorithm. The method based on this embodiment can use the unscented transform optimized random forest algorithm to transform the problem of describing user electricity consumption behavior under uncertain future conditions into a deterministic calculation problem with a small number of sampling points, thereby improving the ability to predict and perceive future user electricity consumption behavior.

[0028] The load-side electricity consumption behavior probabilistic prediction method based on the unscented transform optimized random forest algorithm in this embodiment can be applied to a terminal, which can be an intelligent product terminal such as a computer. Figure 1 As shown, the load-side electricity consumption behavior probabilistic prediction method based on the unscented transform optimized random forest algorithm includes the following steps: Step S100: Obtain historical electricity consumption information of user terminals that reflects the electricity consumption behavior of the load side, as well as meteorological information, user product order information, electricity spot market price, socio-economic environment, low-carbon and environmental protection assessment factors that affect the energy consumption behavior of the user side, and construct a training dataset.

[0029] Combination Figure 2 As shown in the diagram, this embodiment can acquire historical electricity consumption information reflecting user electricity consumption behavior through the system's front-end acquisition unit, as well as meteorological information, user product order information, electricity spot market prices, socio-economic environment, and parameters related to low-carbon and environmental protection assessment factors that affect user-side energy consumption behavior. The collected historical parameters can be obtained through pre-entry or real-time acquisition methods.

[0030] Before predicting user-side energy consumption behavior, the system obtains information through a pre-collection module. This includes user historical energy consumption information, response speed, response duration, and parameters related to resource regulation characteristics, which are declared by the resource provider during registration and confirmed through system testing. User participation costs in the spot market and user product order information are assigned default values ​​by the resource provider during registration and updated by the user after each invitation. Meteorological information is provided by the enterprise-level meteorological service center of the power grid management system and can be imported into the system via E-file format. Socioeconomic environment and low-carbon environmental protection assessment information for the user's location can be obtained by the user providing their electricity account number, and the user's specific location can be obtained through the account number and resource access relationship in the power grid marketing system. Based on this location relationship, the user's socioeconomic environment and low-carbon environmental protection assessment information can be obtained.

[0031] Step S200: Construct a random forest model based on the random forest algorithm; wherein the random forest model uses decision trees obtained by sampling through the Bagging algorithm as basic units, and obtains the output value of the random forest regression model by fitting the output value of the decision tree model.

[0032] Random Forest is a combined classification and prediction algorithm based on Bagging (Bootstrap Aggregating) and ensemble learning. It uses a bootstrap substitution method to locally sample the original samples, ensuring that each different subsample has the same size and class distribution. Multiple base classifiers based on different principles are combined using Bagging, effectively preventing homogeneous base classifiers from consistently misclassifying specific samples. Compared to traditional single decision trees, Random Forest constructs multiple different decision trees by training on different sample subsets. Based on an optimal combination of sensitive parameters, it selects one sensitive parameter as the root node and the remaining sensitive parameters as child nodes, comprehensively judging reservoir types according to their priority. Guided by feature engineering, the Random Forest model can be more stable, focusing only on more effective sensitive parameters, improving both accuracy and efficiency in reservoir identification.

[0033] The construction process of a random forest model consists of two parts: algorithm parameter selection and model training. Specifically, the parameters involved in the random forest algorithm include the number of decision trees in the decision forest, the maximum depth of the trees, the minimum number of samples required for internal node re-splits, the minimum number of samples required for a leaf node, the maximum number of features considered when finding the optimal split, and the selection of a split quality metric. For the split quality metric, either the Gini coefficient or the information entropy gain can be chosen. For other random forest hyperparameters, a random search method can be used to obtain them.

[0034] This embodiment uses a regression tree generation algorithm to train a random forest model. In the input space of the training dataset, each region is recursively divided into two sub-regions, and the output value for each sub-region is determined, constructing a binary decision tree. Specifically, n training samples are taken from the training dataset with replacement each time to form a new training set. Each training sample has d features, and only k features are selected each time, where k ≤ d, to construct a decision tree model. This method of generating decision tree models is repeated multiple times to generate multiple decision trees, each producing a decision result. The decision results of multiple decision trees are fitted to obtain the final prediction analysis result of the decision forest, which is the output value of the random forest regression model.

[0035] Step S300: Train a random forest model based on the unscented transformation method, and during training, convert the future expectations of relevant influencing factors affecting user energy consumption behavior into deterministic sampling points and corresponding weight values.

[0036] This embodiment identifies a set of statistical characteristics of factors a related to influencing users' energy consumption behavior. N σ Each sampling point, i.e., a sigma point, and the corresponding weight for each sigma point. W i The weights of the sigma sampling points satisfy the normalization condition, i.e. Furthermore, the sigma point must capture some statistical characteristics of a, and the mean of x must be taken as a constant. μ x Covariance P xx Represented by sigma points.

[0037] Specifically, this embodiment can use the symmetrical sampling method. For the known future expected value x of the relevant influencing factors affecting user energy consumption behavior, and x is an m-dimensional variable, where m reflects the number of relevant influencing factors affecting user energy consumption behavior; If only the mean and covariance of the user's electricity consumption behavior output variables are needed, then select... If you need to obtain the skewness coefficient and kurtosis coefficient of the output variable, you can select... .

[0038] Let sigma point ,satisfy , , Where i = 1, 2, ..., m, For matrix The element in the i-th column is an m-dimensional column vector, and the required matrix can be obtained through the Cholesky transformation. ; The weight value for the sampling point is set to... , , , where i = 1, 2, ..., m, where Values ​​are set by humans and are generally taken as follows: =1 / 3.

[0039] Step S400: Use the pre-trained random forest model to perform deterministic calculations on the sampling points to obtain the predicted output variables corresponding to each sampling point. Based on the predicted output variables and the corresponding weight coefficients, perform fitting to obtain the probability prediction information of the electricity consumption behavior of the corresponding load-side users.

[0040] After training and obtaining the random forest model, this embodiment treats the random forest model as a nonlinear transformation function and the load electricity consumption behavior prediction problem as a nonlinear transformation problem. The known input variable x is the influencing factor related to user electricity consumption behavior, and the statistical characteristics of x have been obtained. The output variable y is used to describe electricity consumption behavior, which can be described by the amount of electricity disconnected from the grid. The output variable y and x satisfy the nonlinear transformation y=f(x) constructed by the random forest model, and it is necessary to solve for the statistical characteristics of y. Based on the intuitive inference that "the probability distribution of an approximate nonlinear function is easier to approximate than the nonlinear function itself", a set of sampling points that can reflect the statistical characteristics of x, namely sigma points, and the corresponding weights W of the sigma points are first determined. Then, each sigma point is input into the random forest model to perform a deterministic nonlinear transformation.

[0041] Next, this embodiment uses decision trees to perform regression calculations for each sampling point; then, the results generated by each decision tree are fitted, which can be done using the mean fitting method. Finally, based on the unscented transform optimized random forest algorithm, probabilistic prediction information of load-side electricity consumption behavior is generated, so as to realize the regulation of user-side resources according to the probabilistic prediction information. Specifically, this embodiment obtains the predicted value output variable corresponding to each sampling point. y i mean μ y and variance P y They are respectively , ,iny_ express y i The arithmetic mean, i.e. User-side electricity consumption behavior prediction based on unscented transformation only uses a finite number of points to describe the second-order probability prediction problem of mean and covariance; to obtain skewness coefficient and kurtosis coefficient, only a higher-order unscented transformation needs to be selected.

[0042] This invention proposes a probabilistic prediction method for load-side electricity consumption behavior based on an optimized random forest algorithm using unscented transform. This method enables the predictive analysis of user-side energy consumption behavior under the influence of various uncertainties, supporting the safe, economical, and low-carbon operation of the power system. The unscented transform method uses only a finite number of sampling points to describe complex probabilistic prediction problems, extending the ability of the random forest method to handle probabilistic prediction issues. While maintaining the interpretability of the random forest algorithm, it optimizes the algorithm's performance, allowing direct application of mature random forest algorithms. Furthermore, it facilitates the extension of higher-order statistical characteristic analysis models of user-side energy consumption behavior, demonstrating strong engineering application value.

[0043] Based on the above implementations, the present invention also provides a load-side electricity consumption behavior probabilistic prediction system based on an unscented transform optimized random forest algorithm. This system is used to implement the method steps in the above method embodiments. Figure 3 As shown, the system includes: a pre-acquisition input module 10, a random forest construction module 20, an unscented transformation module 30, and a random forest prediction module 40. Specifically, the pre-acquisition input module is used to acquire historical electricity consumption information reflecting load-side electricity consumption behavior, as well as meteorological information, user product order information, electricity spot market prices, socio-economic environment, and low-carbon environmental protection assessment factors affecting user-side energy consumption behavior, and to construct a training dataset. The random forest construction module 20 is used to construct a random forest model based on the random forest algorithm; wherein the random forest model uses decision trees obtained through Bagging algorithm sampling as basic units, and obtains the output value of the random forest regression model through a fitting method of the decision tree model output value. The unscented transformation module 30 is used to train the random forest model based on the unscented transformation method, and during training, converts the future expectations of relevant influencing factors affecting user energy consumption behavior into deterministic sampling points and corresponding weight values. The random forest prediction module 30 is used to perform deterministic calculations on the sampling points using a pre-trained random forest model to obtain the predicted output variables corresponding to each sampling point. Based on the predicted output variables and the corresponding weight coefficients, it performs fitting to obtain the probability prediction information of the electricity consumption behavior of the corresponding load-side users.

[0044] The load-side electricity consumption behavior probability prediction system based on the unscented transform optimized random forest algorithm in this embodiment is based on the same principle as the steps in the above method embodiments, and will not be repeated here.

[0045] Based on the above embodiments, the present invention also provides a terminal, the principle block diagram of which can be as follows: Figure 4 As shown. The terminal may include one or more processors 100 ( Figure 4 (Only one is shown in the image), memory 101, and computer program 102 stored in memory 101 and executable on one or more processors 100. For example, a load-side electricity consumption behavior probability prediction program based on an unscented transform optimized random forest algorithm. When one or more processors 100 execute computer program 102, they can implement the various steps in the embodiment of the load-side electricity consumption behavior probability prediction method based on an unscented transform optimized random forest algorithm. Alternatively, when one or more processors 100 execute computer program 102, they can implement the functions of each module / unit in the embodiment of the load-side electricity consumption behavior probability prediction system based on an unscented transform optimized random forest algorithm, which is not limited here.

[0046] In one embodiment, the processor 100 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0047] In one embodiment, memory 101 may be an internal storage unit of an electronic device, such as a hard drive or RAM. Memory 101 may also be an external storage device of the electronic device, such as a plug-in hard drive, smart media card (SMC), secure digital card (SD), flash card, etc. Furthermore, memory 101 may include both internal and external storage units. Memory 101 is used to store computer programs and other programs and data required by the terminal. Memory 101 can also be used to temporarily store data that has been output or will be output.

[0048] Those skilled in the art will understand that Figure 4The block diagram shown is merely a partial structural diagram related to the present invention and does not constitute a limitation on the terminal to which the present invention is applied. A specific terminal may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0049] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, operational databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual operating data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0050] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for probabilistic prediction of load-side electricity consumption behavior based on unscented transform optimized random forest algorithm, characterized in that, The method includes: Acquire historical electricity consumption data at user terminals that reflect load-side electricity consumption behavior, as well as meteorological information, user product order information, electricity spot market prices, socio-economic environment, and low-carbon environmental protection assessment factors that affect user-side energy consumption behavior, and construct a training dataset. A random forest model is constructed based on the random forest algorithm; wherein, the random forest model uses decision trees obtained by sampling through the Bagging algorithm as the basic unit, and obtains the output value of the random forest regression model by fitting the output value of the decision tree model; Identify a set of statistical characteristics of factors a that influence user energy consumption behavior. N σ Each sampling point, i.e., a sigma point, and the corresponding weight for each sigma point. W i The weights of the sigma sampling points satisfy the normalization condition, i.e. Furthermore, the sigma point must acquire partial statistical characteristics of the relevant influencing factor 'a' affecting user energy consumption behavior, and the mean of the known future expected values ​​'x' of the relevant influencing factors affecting user energy consumption behavior. μ x Covariance P xx Represented by sigma points; among them, some statistical characteristics of the factors a related to user energy consumption behavior include: the mean, covariance, skewness coefficient, and kurtosis coefficient of the output variable of user electricity consumption behavior; The random forest model is treated as a nonlinear transformation function, and the problem of predicting load electricity consumption behavior is also treated as a nonlinear transformation problem. The pre-trained random forest model is used to perform deterministic calculations on the sampling points to obtain the predicted output variables corresponding to each sampling point. Based on the predicted output variables and the corresponding weight coefficients, fitting is performed to obtain the probabilistic prediction information of the corresponding load-side user electricity consumption behavior.

2. The load-side electricity consumption behavior probabilistic prediction method based on unscented transform optimized random forest algorithm according to claim 1, characterized in that, Based on the random forest algorithm, a random forest model is constructed, including: Each time, n training samples are taken from the training dataset with replacement to form a new training set. Each training sample has d features. Each time, only k features are selected, and k≤d, to build a decision tree model. The method of repeatedly generating decision tree models generates multiple decision trees, and each decision tree produces a decision result; By fitting the decision results of multiple decision trees, the final prediction analysis results of the decision forest are obtained, which is the output value of the random forest regression model.

3. The load-side electricity consumption behavior probabilistic prediction method based on unscented transform optimized random forest algorithm according to claim 2, characterized in that, Building a random forest model based on the random forest algorithm also includes: In the input space where the training dataset is located, each region is recursively divided into two sub-regions and the output value on each sub-region is determined to construct a binary decision tree.

4. The load-side electricity consumption behavior probabilistic prediction method based on unscented transform optimized random forest algorithm according to claim 1, characterized in that, Also includes: The symmetric sampling method is adopted. For the known future expected values ​​x of the relevant factors affecting users' energy consumption behavior, and x is an m-dimensional variable, where m reflects the number of relevant factors affecting users' energy consumption behavior; If only the mean and covariance of the user's electricity consumption behavior output variables are needed, then select... If you need to obtain the skewness coefficient and kurtosis coefficient of the output variable, then select... .

5. The load-side electricity consumption behavior probabilistic prediction method based on unscented transform optimized random forest algorithm according to claim 4, characterized in that, Let sigma point ,satisfy , , ,in =1,2,…,m, For matrix The The column elements are m-dimensional column vectors, and the required matrix is ​​obtained through the Cholesky transformation. ; The weight value for the sampling point is set to... , , , where i = 1, 2, ..., m, where Values ​​are set by humans.

6. The load-side electricity consumption behavior probabilistic prediction method based on unscented transform optimized random forest algorithm according to claim 5, characterized in that, Obtain the predicted value output variable corresponding to each sampling point. y i mean μ y and variance P y They are respectively , ,in express y i The arithmetic mean, i.e. User-side electricity consumption behavior prediction based on unscented transformation only uses a finite number of points to describe the second-order probability prediction problem of mean and covariance; to obtain skewness coefficient and kurtosis coefficient, only a higher-order unscented transformation needs to be selected.

7. A load-side electricity consumption behavior probabilistic prediction system based on unscented transform optimized random forest algorithm, characterized in that, The system is used to implement the steps of the load-side electricity consumption behavior probabilistic prediction method based on the unscented transform optimized random forest algorithm as described in any one of claims 1-6, and the system includes: The front-end data acquisition module is used to acquire historical electricity consumption information of users that reflects the electricity consumption behavior of the load side, as well as meteorological information, user product order information, electricity spot market price, socio-economic environment, low-carbon and environmental protection assessment factors that affect the energy consumption behavior of users, and to build a training dataset. The random forest construction module is used to construct a random forest model based on the random forest algorithm. The random forest model uses decision trees obtained by sampling through the Bagging algorithm as the basic unit, and obtains the output value of the random forest regression model by fitting the output value of the decision tree model. The unscented transformation module is used to determine a set of statistical characteristics that reflect the factors 'a' that influence user energy consumption behavior. N σ Each sampling point, i.e., a sigma point, and the corresponding weight for each sigma point. W i The weights of the sigma sampling points satisfy the normalization condition, i.e. Furthermore, the sigma point must acquire partial statistical characteristics of the relevant influencing factor 'a' affecting user energy consumption behavior, and the mean of the known future expected values ​​'x' of the relevant influencing factors affecting user energy consumption behavior. μ x Covariance P xx Represented by sigma points; among them, some statistical characteristics of the factors a related to user energy consumption behavior include: the mean, covariance, skewness coefficient, and kurtosis coefficient of the output variable of user electricity consumption behavior; The random forest prediction module treats the random forest model as a nonlinear transformation function and the load electricity consumption behavior prediction problem as a nonlinear transformation problem. It uses the pre-trained random forest model to perform deterministic calculations on the sampling points to obtain the predicted output variables corresponding to each sampling point. Based on the predicted output variables and corresponding weight coefficients, it performs fitting to obtain the probabilistic prediction information of the corresponding load-side user electricity consumption behavior.

8. A terminal, characterized in that, The terminal includes a memory, a processor, and a load-side electricity consumption behavior probability prediction program based on the unscented transform optimized random forest algorithm stored in the memory and executable on the processor. When the processor executes the load-side electricity consumption behavior probability prediction program based on the unscented transform optimized random forest algorithm, it implements the steps of the load-side electricity consumption behavior probability prediction method based on the unscented transform optimized random forest algorithm as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a load-side electricity consumption behavior probability prediction program based on the unscented transform optimized random forest algorithm. The load-side electricity consumption behavior probability prediction program based on the unscented transform optimized random forest algorithm implements the steps of the load-side electricity consumption behavior probability prediction method based on the unscented transform optimized random forest algorithm as described in any one of claims 1-6 on the computer-readable storage medium.

Images