CN116796639A - Short-term power load prediction method, device and equipment - Google Patents

Abstract

The embodiment of the invention provides a short-term power load prediction method, a short-term power load prediction device and short-term power load prediction equipment. The method comprises the steps of obtaining power demand time series data, and decomposing the power demand time series data into a high-pass coefficient and a low-pass coefficient; convolving the wavelet coefficients with wavelet functions to obtain wavelet coefficients of different frequency bands, and sequentially arranging the wavelet coefficients to obtain input feature vectors; optimizing the wavelet coefficient by the input feature vector through a differential evolution algorithm to obtain a radial basis function neural network model based on the differential evolution optimization algorithm; parameter adjustment is carried out on a radial basis function neural network model based on a differential evolution optimization algorithm; and predicting the load of a future time point by using the radial basis function neural network model based on the differential evolution optimization algorithm after parameter adjustment. In this way, the problems of high-frequency volatility and nonlinearity in processing load data can be solved, the prediction precision is improved, the method is suitable for load demand prediction above the factory level, and better universality and applicability are achieved.

Description

Short-term power load forecasting method, device and equipment

Technical field

The present invention generally relates to the field of load forecasting, and more specifically, to short-term power load forecasting methods, devices and equipment.

Background technique

With the reform of the electricity market, energy costs are of practical significance for industrial enterprises, especially intelligent manufacturing enterprises such as mechanical processing and metallurgy, to strengthen their competitive advantages. The diversification of the power market allows industrial enterprises to flexibly obtain competitive advantages in the market. Short-term load forecasting is the basic condition for reducing energy costs.

Demand forecasting has been a research hotspot for many scholars in recent years. Most traditional methods use probabilistic models, such as autoregressive method (ARMA), multiple linear regression and grayscale models. However, most traditional model methods have this computational limitation. These limitations cannot handle the nonlinear and missing data characteristics of power demand data. Modern forecasting techniques mostly use hybrid machine learning methods.

The advantage of the hybrid machine learning method is that it has good prediction accuracy while achieving the prediction function. Current research includes many combination methods such as fuzzy-neural network method, support vector machine-optimization method, extreme learning machine-optimization method, and BP neural network. In terms of short-term industrial load forecasting, problems such as long calculation time, low accuracy, and poor robustness are more prominent.

Contents of the invention

According to embodiments of the present invention, a short-term power load prediction scheme is provided. This solution solves the high-frequency fluctuation and nonlinear problems in processing load data, improves the prediction accuracy, is suitable for load demand prediction at the factory level and above, and has better versatility and applicability.

In a first aspect of the invention, a short-term power load forecasting method is provided. The method includes:

Obtain power demand time series data, and decompose the power demand time series data into high-pass coefficients and low-pass coefficients;

Convolve the high-pass coefficients and low-pass coefficients with wavelet functions respectively to obtain wavelet coefficients in different frequency bands, and arrange them in order to obtain the input feature vector;

Optimize the wavelet coefficients of the input feature vector through a differential evolution algorithm, and construct a radial basis function neural network model based on the differential evolution optimization algorithm;

Calculate the Gaussian function center, Gaussian function width and the weight of the hidden unit and the output unit of the radial basis function neural network model based on the differential evolution optimization algorithm, and perform the radial basis function neural network model based on the differential evolution optimization algorithm. Parameter adjustment;

The parameter-adjusted radial basis function neural network model based on differential evolution optimization algorithm is used to predict the load at future time points.

Further, the decomposition of the power demand time series data into high-pass coefficients and low-pass coefficients includes:

The original time series of the power demand time series data is filtered N times, and an approximate coefficient and a detail coefficient are obtained after each filtering. A total of N approximate coefficients and N detail coefficients are obtained; the N approximate coefficients are used as high-pass coefficients ;The N detail coefficients are used as low-pass coefficients.

Further, the wavelet coefficients of the input feature vector are optimized through a differential evolution algorithm to obtain a radial basis function neural network model based on the differential evolution optimization algorithm, including:

Initialize the population;

Randomly select several different individuals from the population, scale the vector differences of the selected individuals, and perform vector synthesis with the individuals to be mutated to obtain a mutation vector;

Perform a crossover operation on the reference vector and the mutation vector;

Use a differential evolution algorithm to select the optimal individual in the population to obtain offspring individuals;

Conduct an adaptability assessment on the offspring individuals, and obtain elite individuals based on adaptability screening;

The elite individuals when reaching the iteration conditions are used as the radial basis function neural network model based on the differential evolution optimization algorithm.

Further, the calculation of the Gaussian function center, Gaussian function width, and weights of hidden units and output units of the radial basis function neural network model based on the differential evolution optimization algorithm includes:

Using the K-means clustering algorithm, the wavelet coefficients of the different frequency bands are clustered, and the cluster center is used as the center of the Gaussian function;

Using the mean square error as the loss function, the weights of the hidden units and output units are calculated through the backpropagation algorithm;

The optimal width is selected as the width of the Gaussian function through cross-validation.

Further, the use of the parameter-adjusted radial basis function neural network model based on the differential evolution optimization algorithm to predict the load at future time points includes:

The wavelet coefficients of different frequency bands are input into the radial basis function neural network model based on the differential evolution optimization algorithm after parameter adjustment, and the load value of the future continuous time series is output.

In a second aspect of the present invention, a short-term power load prediction device is provided. The device includes:

A decomposition module, used to obtain power demand time series data and decompose the power demand time series data into high-pass coefficients and low-pass coefficients;

The convolution module is used to convolve the high-pass coefficient and the low-pass coefficient with the wavelet function respectively to obtain the wavelet coefficients of different frequency bands, and arrange them in order to obtain the input feature vector;

An optimization module, used to optimize the wavelet coefficients of the input feature vector through a differential evolution algorithm to obtain a radial basis function neural network model based on the differential evolution optimization algorithm;

A calculation module used to calculate the Gaussian function center, Gaussian function width, and the weight of the hidden unit and the output unit of the radial basis function neural network model based on the differential evolution optimization algorithm;

A prediction module is used to predict the load at future time points using the radial basis function neural network model based on the differential evolution optimization algorithm.

Further, the decomposition of the power demand time series data into high-pass coefficients and low-pass coefficients includes:

The original time series of the power demand time series data is filtered N times, and an approximate coefficient and a detail coefficient are obtained after each filtering. A total of N approximate coefficients and N detail coefficients are obtained; the N approximate coefficients are used as high-pass coefficients ;The N detail coefficients are used as low-pass coefficients.

Further, the wavelet coefficients of the input feature vector are optimized through a differential evolution algorithm to obtain a radial basis function neural network model based on the differential evolution optimization algorithm, including:

Initialize the population;

Randomly select several different individuals from the population, scale the vector differences of the selected individuals, and perform vector synthesis with the individuals to be mutated to obtain a mutation vector;

Perform a crossover operation on the reference vector and the mutation vector;

Use a differential evolution algorithm to select the optimal individual in the population to obtain offspring individuals;

Conduct an adaptability assessment on the offspring individuals, and obtain elite individuals based on adaptability screening;

The elite individuals when reaching the iteration conditions are used as the radial basis function neural network model based on the differential evolution optimization algorithm.

Further, the calculation of the Gaussian function center, Gaussian function width, and weights of hidden units and output units of the radial basis function neural network model based on the differential evolution optimization algorithm includes:

Using the K-means clustering algorithm, the wavelet coefficients of the different frequency bands are clustered, and the cluster center is used as the center of the Gaussian function;

Using the mean square error as the loss function, the weights of the hidden units and output units are calculated through the backpropagation algorithm;

The optimal width is selected as the width of the Gaussian function through cross-validation.

Further, the use of the parameter-adjusted radial basis function neural network model based on the differential evolution optimization algorithm to predict the load at future time points includes:

The wavelet coefficients of different frequency bands are input into the radial basis function neural network model based on the differential evolution optimization algorithm after parameter adjustment, and the load value of the future continuous time series is output.

In a third aspect of the invention, an electronic device is provided. The electronic device has at least one processor; and a memory communicatively connected to the at least one processor; the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor, To enable the at least one processor to execute the method of the first aspect of the invention.

In a fourth aspect of the present invention, there is provided a non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the method of the first aspect of the present invention.

It should be understood that the content described in this summary is not intended to identify key or important features of the embodiments of the invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description.

Description of the drawings

The above and other features, advantages and aspects of various embodiments of the invention will become more apparent with reference to the following detailed description taken in conjunction with the accompanying drawings. In the drawings, the same or similar reference numbers represent the same or similar elements, where:

Figure 1 shows a flow chart of a short-term power load forecasting method according to an embodiment of the present invention;

Figure 2 shows a schematic diagram of the optimization process of optimizing wavelet coefficients through a differential evolution algorithm according to an embodiment of the present invention;

Figure 3 shows a block diagram of a short-term power load prediction device according to an embodiment of the present invention;

4 illustrates a block diagram of an exemplary electronic device capable of implementing embodiments of the invention;

Among them, 400 is electronic equipment, 401 is computing unit, 402 is ROM, 403 is RAM, 404 is bus, 405 is I/O interface, 406 is input unit, 407 is output unit, 408 is storage unit, and 409 is communication unit. .

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments These are some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

In addition, the term "and/or" in this article is only an association relationship that describes related objects, indicating that there can be three relationships. For example, A and/or B can mean: A alone exists, and A and B exist simultaneously. There are three cases of B alone. In addition, the character "/" in this article generally indicates that the related objects are an "or" relationship.

Figure 1 shows a flow chart of a short-term power load prediction method according to an embodiment of the present invention. The DE-RBFNN model represents the radial basis function neural network model based on the differential evolution optimization algorithm.

The method includes:

S101. Obtain power demand time series data, and decompose the power demand time series data into high-pass coefficients and low-pass coefficients.

In this embodiment, the user historical data obtained by the power system is user load data, which can be a 15-minute time interval, the unit is KW; or the time interval is a 1-hour time interval, the unit is MW.

In this embodiment, the original time series of the power demand time series data is filtered N times, and an approximate coefficient and a detail coefficient are obtained after each filtering, and a total of N approximate coefficients and N detail coefficients are obtained; the N The approximation coefficients are used as high-pass coefficients; the N detail coefficients are used as low-pass coefficients.

Specifically, DWT (Discrete Wavelet Transformation) discrete wavelet transform is used as a preprocessing technology to decompose the power demand time series data into the frequency domain and time domain. Reduce the noise present in the data and make the power data smoother. Using the discrete wavelet transform, the signal is decomposed into a set of approximate high-pass and low-pass coefficients. Decompose power data into three or more levels of information. For example, the decomposition process is four levels, N=4, the number of 4 approximate coefficients is (A ₁ , A ₂ , A ₃ , A ₄ ) and 4 detail coefficients (D ₁ , D ₂ , D ₃ , D ₄ ) respectively. for four.

In the above embodiment, the original time series is filtered once through the filter to obtain the first-level approximation coefficient A ₁ and the first-level detail coefficient D ₁ .

Then, the first-level approximation coefficient A ₁ and the first-level detail coefficient D ₁ are filtered twice through the filter to obtain the second-level approximation coefficient A ₂ and the second-level detail coefficient D ₂ .

Then, the second-level approximation coefficient A ₂ and the second-level detail coefficient D ₂ are filtered three times through the filter to obtain the third-level approximation coefficient A ₃ and the third-level detail coefficient D ₃ .

Finally, the third-level approximation coefficient A ₃ and the third-level detail coefficient D ₃ are filtered four times through the filter to obtain the fourth-level approximation coefficient A ₄ and the third-level detail coefficient D ₄ .

As an embodiment of the present invention, from the above four approximate coefficients (A ₁ , A ₂ , A ₃ , A ₄ ) and the four detail coefficients (D ₁ , D ₂ , D ₃ , D ₄ ), select The last level of approximation coefficients is the fourth level approximation coefficient A ₄ and all detail coefficients. Reconstruct the time series x(t):

In the above embodiment, DWT is used as a preprocessing technique to decompose the power demand time series data into frequency domain and time domain. Reduce the noise present in the data and make the power data smoother.

S102. Convolve the high-pass coefficients and low-pass coefficients with wavelet functions respectively to obtain wavelet coefficients in different frequency bands, and arrange them in order to obtain input feature vectors.

Wavelet decomposition is a technique that decomposes a signal into frequency bands of different scales. DWT obtains wavelet coefficients in different frequency bands by convolving the signal with the wavelet function. The wavelet coefficient features of different scales are arranged in a certain order to form an input feature vector.

S103. Use the differential evolution algorithm to optimize the wavelet coefficients of the input feature vector to obtain the DE-RBFNN model.

DE (Differential Evolution Algorithm) is a differential evolution algorithm. RBFNN (Radial basis function neural network) is a radial basis function neural network.

In this embodiment, as shown in Figure 2, the optimization process of optimizing wavelet coefficients through differential evolution algorithm includes:

S201. Initialize the population p ^k .

Specifically, a population consists of a group of individuals Each individual/> Represents a parameter set of an RBF neural network. According to the formula/> Randomly select the initial value, where:/> Represents the j-th value of the i-th individual in the 0th generation; U(0,1) represents a uniformly distributed random number in the interval (0,1).

S202. Randomly select several different individuals from the population, scale the vector differences of the selected individuals, and perform vector synthesis with the individuals to be mutated to obtain a mutation vector.

Specifically, the mutation is completed by randomly selecting two different individuals in the population, scaling their vector differences, and then performing vector synthesis with the individual to be mutated. In the formula: x _j1 , x _j2 , x _j3 are three different individuals randomly selected from the current population; F is the scaling factor, which determines the step length and speed of the search, and the value range is [0,1].

S203. Perform a crossover operation on the reference vector and the mutation vector.

Specifically, the baseline vector and the mutation vector are crossed, and the binomial crossover operator is calculated as In the formula: rand _i is a uniformly distributed random number in the interval [0,1]; i _n is a uniformly distributed random number in the interval [1,n]; CR is the crossover probability, which determines the weight of genetic information before and after mutation, taking The value range is [0,1].

S204. Use a differential evolution algorithm to select the optimal individual in the population to obtain offspring individuals.

The differential evolution algorithm, or DE algorithm, uses a greedy selection mechanism to ensure that the population always evolves towards the global optimum. The DE algorithm has the advantages of fewer control parameters, fast convergence speed, and high reliability in solving nonlinear problems, and has been widely used to solve various problems. However, the DE algorithm has great pressure to select control parameters, and the algorithm performance is highly dependent on the control parameters. Population individuals are prone to problems such as premature convergence, local optimality, and search stagnation, which have certain limitations in the application solution process.

S205. Conduct an adaptability assessment on the offspring individuals, and obtain elite individuals based on adaptability screening.

Specifically, for each individual, its fitness function value is calculated. The fitness function can be a measure of training error, such as root mean square error (RMSE) or cross-entropy loss. Fitness is calculated by evaluating the individual's predictive performance on the training set. For newly generated individuals Calculate its fitness function value. Will the newly generated individual/> With the parent individual/> Compare and retain individuals with better fitness as elite individuals.

S206. Use the elite individuals when the iteration conditions are met as the DE-RBFNN model.

Specifically, check whether the termination condition is met. The termination condition can be that the maximum number of iterations is reached or the fitness function value reaches a predefined threshold. Finally, the individual with the best fitness is selected as the final DE-RBFNN model.

S104. Calculate the Gaussian function center, Gaussian function width, and weights of hidden units and output units of the DE-RBFNN model, and adjust parameters of the DE-RBFNN model.

In this embodiment, the K-means clustering algorithm is used to cluster the wavelet coefficients in the training set, and the cluster center is used as the center of the Gaussian function. The mean square error (MSE) is used as the loss function, and through the reverse The propagation algorithm updates the weights and selects the optimal width through cross-validation or trial and error. You can try different width values, train and validate the model for each value, and then choose the width value that performs best on the validation set. The relative deviation values of all compared models are calculated, which can be confirmed more directly by calculating the MAD, where the proposed model returns the best value for the training set and the test set.

Gaussian function center: Gaussian function center represents the position of each hidden unit (radial basis function) in the input space. Each hidden unit has a corresponding Gaussian function center. These center points determine the degree to which the network responds to different features and samples in the input space, thereby affecting the representation ability of the network and the fitting ability of the model.

Gaussian function width: Gaussian function width determines the coverage of the radial basis function in the input space. It controls the activation degree and amplitude decay rate of the radial basis function. A smaller width will lead to a sharper activation function curve, which is sensitive to local changes in the input space; a larger width will lead to a flatter activation function curve, which is more sensitive to the overall characteristics of the input space. The choice of Gaussian function width usually needs to be tuned during the training process so that the network can better fit the data and have appropriate generalization capabilities.

Weights of Hidden Units and Output Units: The weights of hidden units and output units are used to pass and transform the input signal to the next layer. The weights of the hidden units determine the weighted response of the radial basis function to the input data, and the weights of the output units determine the calculation of the output value. These weights are adjusted during training by optimization algorithms such as differential evolution to minimize the network's prediction error or loss function. By adjusting the weights, the network can learn the characteristics and patterns of the data in order to make accurate predictions and classifications.

In this embodiment, the Gaussian function center and width determine the position and activation range of the radial basis function in RBFNN, while the weights of the hidden units and output units determine the network connection and signal transmission. The selection and adjustment of these parameters have an important impact on the modeling ability and performance of RBFNN.

S105. Use the parameter-adjusted DE-RBFNN model to predict the load at future time points. That is, after the above-mentioned parameters such as the center of the Gaussian function, the width of the Gaussian function, and the weights of the hidden units and output units are adjusted, the wavelet coefficients of different frequency bands are input into the parameter-adjusted DE-RBFNN model, and the load values of the future continuous time series are output to realize the control. Load forecast at future time points.

According to embodiments of the present invention, a short-term load prediction method is proposed, which effectively handles high-frequency fluctuations and nonlinear problems in load data and improves prediction accuracy. This method is suitable for load demand forecasting at the factory level and above, and has better versatility and applicability.

It should be noted that for the sake of simple description, the foregoing method embodiments are expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequence. Because in accordance with the present invention, certain steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily necessary for the present invention.

The above is an introduction to the method embodiments. The solution of the present invention will be further described below through device embodiments.

As shown in Figure 3, device 300 includes:

Decomposition module 310 is used to obtain power demand time series data and decompose the power demand time series data into high-pass coefficients and low-pass coefficients;

The convolution module 320 is used to convolve the high-pass coefficient and the low-pass coefficient with the wavelet function respectively to obtain the wavelet coefficients of different frequency bands, and arrange them in order to obtain the input feature vector;

The optimization module 330 is used to optimize the wavelet coefficients of the input feature vector through a differential evolution algorithm to obtain the DE-RBFNN model;

The calculation module 340 is used to calculate the Gaussian function center, Gaussian function width, and the weight of the hidden unit and the output unit of the DE-RBFNN model, and adjust parameters of the DE-RBFNN model;

The prediction module 350 is used to predict the load at future time points using the adjusted DE-RBFNN model.

In this embodiment, the decomposition module 310 filters the original time series of the power demand time series data N times, and obtains an approximate coefficient and a detail coefficient after each filtering. A total of N approximate coefficients and N Detail coefficients; the N approximate coefficients serve as high-pass coefficients; the N detail coefficients serve as low-pass coefficients.

In this embodiment, the optimization module 330 is specifically used to:

Initialize the population;

Randomly select several different individuals from the population, scale the vector differences of the selected individuals, and perform vector synthesis with the individuals to be mutated to obtain a mutation vector;

Perform a crossover operation on the reference vector and the mutation vector;

Use a differential evolution algorithm to select the optimal individual in the population to obtain offspring individuals;

Conduct an adaptability assessment on the offspring individuals, and obtain elite individuals based on adaptability screening;

The elite individuals when reaching the iteration conditions are used as the DE-RBFNN model.

In this embodiment, the calculation module 340 is specifically used for:

The K-means clustering algorithm is used to cluster the wavelet coefficients of different frequency bands, and the cluster center is used as the center of the Gaussian function; the mean square error is used as the loss function, and the hidden unit and output are calculated through the back propagation algorithm. The weight of the unit; the optimal width is selected as the width of the Gaussian function through cross-validation.

Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the described module can be referred to the corresponding process in the foregoing method embodiment, and will not be described again here.

In the technical solution of the present invention, the acquisition, storage and application of the user's personal information are all in compliance with relevant laws and regulations and do not violate public order and good customs.

According to embodiments of the present invention, the present invention also provides an electronic device and a readable storage medium.

Figure 4 shows a schematic block diagram of an electronic device 400 that may be used to implement embodiments of the invention. Electronic devices are intended to refer to various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are examples only and are not intended to limit the implementation of the invention described and/or claimed herein.

The device 400 includes a computing unit 401 that can perform various appropriate actions according to a computer program stored in a read-only memory (ROM) 402 or loaded from a storage unit 408 into a random access memory (RAM) 403 and deal with. In the RAM 403, various programs and data required for the operation of the device 400 can also be stored. Computing unit 401, ROM 402 and RAM 403 are connected to each other via bus 404. An input/output (I/O) interface 405 is also connected to bus 404.

Multiple components in the device 400 are connected to the I/O interface 405, including: input unit 406, such as a keyboard, mouse, etc.; output unit 407, such as various types of displays, speakers, etc.; storage unit 408, such as a magnetic disk, optical disk, etc. ; and communication unit 409, such as a network card, modem, wireless communication transceiver, etc. The communication unit 409 allows the device 400 to exchange information/data with other devices through computer networks such as the Internet and/or various telecommunications networks.

Computing unit 401 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processing processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 401 executes various methods and processes described above, such as methods S101 to S105. For example, in some embodiments, methods S101 to S105 may be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as the storage unit 408. In some embodiments, part or all of the computer program may be loaded and/or installed onto device 400 via ROM 402 and/or communication unit 409. When the computer program is loaded into the RAM 403 and executed by the computing unit 401, one or more steps of the methods S101 to S105 described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to perform methods S101 to S105 in any other suitable manner (eg, by means of firmware).

Various implementations of the systems and techniques described above may be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip implemented in a system (SOC), load programmable logic device (CPLD), computer hardware, firmware, software, and/or a combination thereof. These various embodiments may include implementation in one or more computer programs executable and/or interpreted on a programmable system including at least one programmable processor, the programmable processor The processor, which may be a special purpose or general purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device. An output device.

Program code for implementing the methods of the invention may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that the program codes, when executed by the processor or controller, cause the functions specified in the flowcharts and/or block diagrams/ The operation is implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of this disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media would include one or more wire-based electrical connections, laptop disks, hard drives, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.

To provide interaction with a user, the systems and techniques described herein may be implemented on a computer having a display device (eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user ); and a keyboard and pointing device (eg, a mouse or a trackball) through which a user can provide input to the computer. Other kinds of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and may be provided in any form, including Acoustic input, voice input or tactile input) to receive input from the user.

The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., A user's computer having a graphical user interface or web browser through which the user can interact with implementations of the systems and technologies described herein), or including such backend components, middleware components, or any combination of front-end components in a computing system. The components of the system may be interconnected by any form or medium of digital data communication (eg, a communications network). Examples of communication networks include: local area network (LAN), wide area network (WAN), and the Internet.

Computer systems may include clients and servers. Clients and servers are generally remote from each other and typically interact over a communications network. The relationship of client and server is created by computer programs running on corresponding computers and having a client-server relationship with each other. The server can be a cloud server, a distributed system server, or a server combined with a blockchain.

It should be understood that various forms of the process shown above may be used, with steps reordered, added or deleted. For example, each step described in the present invention can be executed in parallel, sequentially, or in different orders. As long as the desired results of the technical solution of the present invention can be achieved, there is no limitation here.

The above-mentioned specific embodiments do not constitute a limitation on the scope of the present invention. It will be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions are possible depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims (10)

1. A short-term power load forecasting method, characterized by including: Obtain power demand time series data, and decompose the power demand time series data into high-pass coefficients and low-pass coefficients; Convolve the high-pass coefficients and low-pass coefficients with wavelet functions respectively to obtain wavelet coefficients in different frequency bands, and arrange them in order to obtain the input feature vector; Optimize the wavelet coefficients of the input feature vector through a differential evolution algorithm, and construct a radial basis function neural network model based on the differential evolution optimization algorithm; Calculate the Gaussian function center, Gaussian function width and the weight of the hidden unit and the output unit of the radial basis function neural network model based on the differential evolution optimization algorithm, and perform the radial basis function neural network model based on the differential evolution optimization algorithm. Parameter adjustment; The parameter-adjusted radial basis function neural network model based on differential evolution optimization algorithm is used to predict the load at future time points. 2. The method of claim 1, wherein decomposing the power demand time series data into high-pass coefficients and low-pass coefficients includes: The original time series of the power demand time series data is filtered N times, and an approximate coefficient and a detail coefficient are obtained after each filtering. A total of N approximate coefficients and N detail coefficients are obtained; the N approximate coefficients are used as high-pass coefficients ;The N detail coefficients are used as low-pass coefficients. 3. The method according to claim 1, characterized in that the wavelet coefficients of the input feature vector are optimized through a differential evolution algorithm to obtain a radial basis function neural network model based on the differential evolution optimization algorithm, including: Initialize the population; Randomly select several different individuals from the population, scale the vector differences of the selected individuals, and perform vector synthesis with the individuals to be mutated to obtain a mutation vector; Perform a crossover operation on the reference vector and the mutation vector; Use a differential evolution algorithm to select the optimal individual in the population to obtain offspring individuals; Conduct an adaptability assessment on the offspring individuals, and obtain elite individuals based on adaptability screening; The elite individuals when reaching the iteration conditions are used as the radial basis function neural network model based on the differential evolution optimization algorithm. 4. The method according to claim 1, characterized in that the calculation of the Gaussian function center, Gaussian function width and the weight of the hidden unit and the output unit of the radial basis function neural network model based on the differential evolution optimization algorithm, include: Using the K-means clustering algorithm, the wavelet coefficients of the different frequency bands are clustered, and the cluster center is used as the center of the Gaussian function; Using the mean square error as the loss function, the weights of the hidden units and output units are calculated through the backpropagation algorithm; The optimal width is selected as the width of the Gaussian function through cross-validation. 5. The method according to claim 1, characterized in that the use of a parameter-adjusted radial basis function neural network model based on a differential evolution optimization algorithm to predict the load at a future time point includes: The wavelet coefficients of different frequency bands are input into the radial basis function neural network model based on the differential evolution optimization algorithm after parameter adjustment, and the load value of the future continuous time series is output. 6. A short-term power load forecasting device, characterized by including: A decomposition module, used to obtain power demand time series data and decompose the power demand time series data into high-pass coefficients and low-pass coefficients; The convolution module is used to convolve the high-pass coefficient and the low-pass coefficient with the wavelet function respectively to obtain the wavelet coefficients of different frequency bands, and arrange them in order to obtain the input feature vector; An optimization module, used to optimize the wavelet coefficients of the input feature vector through a differential evolution algorithm to obtain a radial basis function neural network model based on the differential evolution optimization algorithm; A calculation module used to calculate the Gaussian function center, Gaussian function width, and the weight of the hidden unit and the output unit of the radial basis function neural network model based on the differential evolution optimization algorithm; A prediction module is used to predict the load at future time points using the radial basis function neural network model based on the differential evolution optimization algorithm. 7. The device according to claim 6, wherein decomposing the power demand time series data into high-pass coefficients and low-pass coefficients includes: The original time series of the power demand time series data is filtered N times, and an approximate coefficient and a detail coefficient are obtained after each filtering. A total of N approximate coefficients and N detail coefficients are obtained; the N approximate coefficients are used as high-pass coefficients ;The N detail coefficients are used as low-pass coefficients. 8. The device according to claim 6, wherein the input feature vector is optimized by using a differential evolution algorithm to optimize the wavelet coefficients to obtain a radial basis function neural network model based on the differential evolution optimization algorithm, including: Initialize the population; Randomly select several different individuals from the population, scale the vector differences of the selected individuals, and perform vector synthesis with the individuals to be mutated to obtain a mutation vector; Perform a crossover operation on the reference vector and the mutation vector; Use a differential evolution algorithm to select the optimal individual in the population to obtain offspring individuals; Conduct an adaptability assessment on the offspring individuals, and obtain elite individuals based on adaptability screening; The elite individuals when reaching the iteration conditions are used as the radial basis function neural network model based on the differential evolution optimization algorithm. 9. An electronic device including at least one processor; and A memory communicatively connected with the at least one processor; characterized in that, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform any one of claims 1-5. Methods. 10. A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the method according to any one of claims 1-5.