Load prediction method, system and device based on CPO-VMD-MLKELM combination model and medium

Abstract

The invention belongs to the technical field of load prediction, and discloses a CPO-VMD-MLKELM combined model-based load prediction method, system and device, and a medium, so as to solve the problems of low generalization ability and poor prediction accuracy of an existing model. The method comprises the steps of obtaining historical load data, meteorological feature data and date feature data of a prediction target, and performing preprocessing to obtain a preprocessed data set; carrying out feature dimension reduction on the preprocessed data set by adopting PCA (Principal Component Analysis), and retaining principal components of which the cumulative variance contribution rate is not less than 98% to obtain dimension-reduced principal component features; dynamically optimizing VMD parameters through four defense strategies by adopting a crown porcupine optimization algorithm, and decomposing historical load data into a plurality of IMF components based on the optimized VMD parameters; and constructing an MLKELM model for each IMF component, predicting each IMF component by taking the dimensionality-reduced principal component characteristics as input, superposing predicted values of each IMF component to obtain a load predicted value, the MLKELM model comprising a stacked KELM automatic encoder and a KELM regression device.

Description

Technical Field

[0001] The present invention belongs to the technical field of load forecasting, and specifically relates to a load forecasting method, system, device and medium based on a CPO-VMD-MLKELM combined model. Background Art

[0002] Power load forecasting and future power generation planning are key links in the efficient operation of power systems. With the continuous rise in global energy demand and the accelerated transformation of energy structure from fossil fuels to fluctuating renewable energy (such as wind power and solar energy), the dynamic balance of power load faces unprecedented complexity. Accurate load forecasting has become a core requirement for modern power system planning. Even a small forecast deviation may cause grid stability problems, increase operating costs, and even lead to large-scale power outages, causing significant economic losses and social impacts. Therefore, improving the accuracy and reliability of load forecasting has always been a research focus in the field of power systems. Extreme learning machine (ELM), as a shallow artificial neural network model, has become an important tool in the field of load forecasting in the past two decades due to its extremely fast training speed (weight optimization can be completed in a single iteration) and strong generalization ability. However, its shallow structure (usually only 1-2 hidden layers) limits the model's ability to model complex nonlinear relationships, making it difficult to capture multi-scale time dependencies (such as daily cycles, seasonal fluctuations, and the impact of emergencies) and high-dimensional feature interactions (such as the coupling of multiple source data such as meteorology, economy, and user behavior) in power load. In order to break through the bottleneck of shallow ELM's expressive power, researchers have constructed a deep ELM model by stacking multiple hidden layers, trying to enhance feature extraction capabilities with the help of deep structures. However, deep ELM has obvious defects: on the one hand, gradient anomalies in back propagation can easily lead to convergence difficulties; on the other hand, although complex models perform well on training data, their generalization capabilities for new scenarios are reduced. In addition, deep structures also face the contradiction between computational efficiency and interpretability - although they can improve performance, they significantly increase computational costs, and the model's decision logic is difficult to trace, which affects their application in actual engineering. Summary of the invention

[0003] Based on the above-mentioned shortcomings and deficiencies in the prior art, one of the objects of the present invention is to at least solve one or more of the above-mentioned problems in the prior art. In other words, one of the objects of the present invention is to provide a load forecasting method, system, device and medium based on the CPO-VMD-MLKELM combined model that meets one or more of the above-mentioned needs, so as to significantly improve the accuracy of load forecasting, while ensuring that the model has excellent generalization ability, so as to achieve the purpose of adapting to complex and changeable power load scenarios.

[0004] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a load forecasting method based on a CPO-VMD-MLKELM combined model, comprising the steps of: S1, obtaining historical load data, meteorological characteristic data and date characteristic data of a forecast target and preprocessing them to obtain a preprocessed data set; S2, performing feature dimensionality reduction on the preprocessed data set using PCA principal component analysis, and retaining principal components with a cumulative variance contribution rate of not less than 98% to obtain principal component characteristics after dimensionality reduction; S3, dynamically optimizing VMD parameters using a crown porcupine optimization algorithm through four defense strategies, and decomposing the historical load data into multiple IMF components based on the optimized VMD parameters, wherein the VMD parameters include a modal number K and a penalty factor α; S4, constructing an MLKELM model for each of the IMF components, using the principal component characteristics after dimensionality reduction as input to predict each of the IMF components, and superimposing the predicted values ​​of each of the IMF components to obtain a load forecast value, wherein the MLKELM model includes a stacked KELM autoencoder and a KELM regressor.

[0005] As a preferred solution, the preprocessing in step S1 includes one-hot encoding the date feature data, and the one-hot encoding is specifically: weekdays are encoded as 0, holidays are encoded as 1; seasonal features are encoded as [1,0,0,0], [0,1,0,0], [0,0,1,0], [0,0,0,1] for spring, summer, autumn, and winter, respectively.

[0006] As a preferred solution, the four defense strategies are visual defense, auditory defense, odor defense and physical attack; the visual defense updates parameters based on the current optimal solution and normally distributed random numbers; the auditory defense determines the search direction through two random individual positions; the odor defense uses a diffusion factor to control the parameter adjustment range; the physical attack corrects parameters based on the inelastic collision theorem.

[0007] As a preferred solution, the value range of the diffusion factor in the odor defense is [0.3, 2.6].

[0008] As a preferred solution, the stacked KELM autoencoder includes a three-layer structure, and the number of nodes in each layer decreases in a ratio of 60%, 40%, and 20%.

[0009] As a preferred solution, the KELM regressor adopts an RBF kernel function, and its bandwidth parameter is optimized and determined by a grid search method.

[0010] As a preferred solution, the MLKELM model is evaluated by a ten-fold cross validation method.

[0011] In a second aspect, the present invention provides a load forecasting system based on a CPO-VMD-MLKELM combined model, which is used to implement the load forecasting method as described in the first aspect.

[0012] In a third aspect, the present invention provides an electronic device, wherein the computer device includes a memory, a processor, and a computer program, and when the computer program is executed by the processor, the load forecasting method as described in the first aspect is implemented.

[0013] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the load forecasting method as described in the first aspect.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention optimizes VMD parameters (modal number K and penalty factor α) through CPO, effectively solving the modal aliasing problem caused by traditional VMD relying on manual experience to adjust parameters, making the decomposition of IMF components more accurate, thereby improving the accuracy of subsequent predictions.

[0015] 2. The MLKELM model combines the unsupervised stacked autoencoder and the supervised KELM regressor, which can deeply extract features and fit nonlinear relationships. Compared with a single ELM or shallow model, the prediction error of complex load series is reduced.

[0016] Further or more detailed beneficial effects will be described in detail in conjunction with specific examples in the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 It is a flow chart of the load forecasting method described in an embodiment of the present invention.

[0019] Figure 2 It is a schematic diagram of the structure of an automatic encoder based on kernel limit learning according to an embodiment of the present invention.

[0020] Figure 3 It is a schematic diagram of the MLKELM prediction principle described in an embodiment of the present invention.

[0021] Figure 4 is a structural diagram of the electronic device provided in an embodiment of the present invention.

[0022] Figure Number: 400. Electronic equipment; 401, processor; 402, communication bus; 403, user interface; 404, network interface; 405, memory. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0024] In the following description, multiple embodiments of the present invention are provided, and different embodiments may be replaced or combined, so the present invention may also be considered to include all possible combinations of the same and / or different embodiments described. Therefore, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then the present invention should also be considered to include embodiments containing one or more of A, B, C, and D, all other possible combinations, even though the embodiment may not be clearly described in the following text.

[0025] The following description provides examples and does not limit the scope, applicability or examples set forth in the claims. Changes may be made to the functions and arrangements of the elements described without departing from the scope of the present invention. Various processes or components may be appropriately omitted, substituted or added to each example. For example, the described method may be performed in an order different from the described order, and various steps may be added, omitted or combined. In addition, the features described in some examples may be combined in other examples.

[0026] In order to facilitate a better understanding of the embodiments of the present invention, before explaining the specific implementation modes of the present invention in detail, its application scenarios are first described.

[0027] The load forecasting method described in the embodiments of this specification is applied to power system scheduling, power generation planning, and new energy grid connection management. In these scenarios, the application of the load forecasting method aims to balance supply and demand through high-precision load forecasting, reduce the waste of spare capacity, and improve the absorption capacity of the power grid with a high proportion of renewable energy.

[0028] Embodiment 1: like Figure 1-3As shown, this embodiment provides a load forecasting method based on the CPO-VMD-MLKELM combined model, including the following steps: S1, obtaining historical load data, meteorological characteristic data and date characteristic data of the forecast target and preprocessing them to obtain a preprocessed data set; S2, using PCA principal component analysis to reduce the dimension of the preprocessed data set, and retaining the principal component with a cumulative variance contribution rate of not less than 98% to obtain the principal component characteristics after dimension reduction; S3, using the crown porcupine optimization algorithm to dynamically optimize the modal number K and penalty factor of the VMD parameter through four defense strategies , based on the optimized VMD parameters, the historical load data is decomposed into multiple IMF components, and the decomposition formula is: , where x is the original signal and N is the number of IMFs; S4, constructing an MLKELM model for each of the IMF components, using the principal component features after dimensionality reduction as input to predict each of the IMF components, and superimposing the predicted values ​​of each of the IMF components to obtain a load prediction value, wherein the MLKELM model includes a stacked KELM autoencoder and a KELM regressor. More specifically, each There is a frequency and amplitude modulation signal , in the above formula, is the frequency, is the amplitude. The expression of the observed signal x0 is , where x is the original signal, is additive Gaussian white noise. Using Tikhonov regularization, the original signal is restored to , thus obtaining and solving the frequency domain Euler–Lagrange equation , where is the variance of white noise, is the Fourier transform of the observed signal. The expression for constrained optimization is , to determine the bandwidth of a mode, where is the center frequency, is the Hilbert transform. In order to convert the constrained optimization model into an unconstrained form, the penalty factor a and the Lagrange multiplier k are defined, and then the model is reorganized More specifically, in the unsupervised stage of the MLKELM model, the first autoencoder randomly maps the input raw data to an ELM hidden space. The following autoencoders use this random space as input data to train each stacked ELM autoencoder separately. When the training is completed, the output expression of each hidden layer is , where g is the activation function, is the output weight matrix of the i-th autoencoder, and is the output of the i-th layer and the i-1-th layer, is to remap the hidden layer of the i-1th layer to the ith layer. Each autoencoder is an independent feature extractor. As the number of autoencoders increases, the features obtained will become more compact. Once these autoencoders are trained separately, the parameters of the stacked autoencoder structure are fixed and no fine-tuning is required. The kernel matrix of the KELM is , the solution of KELM is , the autoencoder of the kernel extreme value learning machine maps the input data to the hidden matrix through the kernel function, and then learns the transformation matrix, which transforms the hidden representation into the output data. The transformation matrix can be learned using , we can calculate the input data of the i+1th layer KELM autoencoder The multi-layer KELM performs KELM-based regression in the second stage, and the final representation matrix constitutes the input of the KELM regression model. The training process of the final stage is as follows , where T is the target matrix, is the kernel representation matrix, is the output weight matrix of the second stage, The calculation formula is , where I is the unit matrix, C is the regularization coefficient, and the predicted values ​​of each IMF component are superimposed to obtain the load prediction value. The MLKELM model includes a stacked KELM autoencoder and a KELM regressor.

[0029] Specifically, this embodiment provides a preferred implementation method, and the preprocessing in step S1 includes one-hot encoding the date feature data, and the one-hot encoding is specifically: weekdays are encoded as 0, and holidays are encoded as 1; seasonal characteristics are encoded as [1,0,0,0], [0,1,0,0], [0,0,1,0], [0,0,0,1] for spring, summer, autumn, and winter, respectively.

[0030] Specifically, this embodiment provides a preferred implementation method, wherein the four defense strategies are visual defense, auditory defense, odor defense, and physical attack; the visual defense updates parameters based on the current optimal solution and normally distributed random numbers; the auditory defense determines the search direction through two random individual positions; the odor defense uses a diffusion factor to control the parameter adjustment range; and the physical attack corrects parameters based on the inelastic collision theorem. More specifically, the use of the crowned porcupine optimization algorithm to dynamically optimize VMD parameters through four defense strategies also includes population initialization, that is, each crowned porcupine in the crowned porcupine group represents a candidate solution, and the initialization mathematical expression of the crowned porcupine group is: , where i=1,2,…,N, N is the population size, x i is the position of the ith individual, rand is a random number between [0,1], UB and LB are the upper and lower limits of the search interval respectively. CPO introduces a cyclic population reduction strategy, which allows some crested porcupines to leave the group during the optimization process and then rejoin the group to increase population diversity, thereby accelerating its convergence speed. Its mathematical expression is: , where C is the number of cycles, and in this embodiment, C = 2 is preferred. is the current iteration number, T max is the maximum number of iterations, mod represents the modulus operation, N min It is the minimum number of individuals in the newly generated population, and is generally 0.8 times of N. The first defense strategy among the four defense strategies is that the crested porcupine raises and flaps its feathers to warn predators. The mathematical model is , where For iteration Time The location of an individual, is the current optimal solution, is a random number based on normal distribution, is a random number on [0, 1], For predators The position at the iteration is expressed as , where is a random integer on [1, N]. The second defense strategy among the four defense strategies is that the crested porcupine makes noise and further threatens the predator. The mathematical model is , where and are the locations of two randomly selected crested porcupines, is a random number on [0, 1], Randomly takes 0 or 1, and y is the position of the predator, which is between the current crested porcupine and a crested porcupine randomly selected from the population. =0, , which means that the second defense strategy has successfully frightened the predator, and the predator no longer moves towards the crested porcupine. =1, which means that the predator is moving towards the crested porcupine, but the predator can be close to the crested porcupine or away from it. The second defense strategy uses two random individuals and To determine whether a predator is approaching or moving away from a crested porcupine, When the predator approaches the crested porcupine, When the predator moves away from the crested porcupine, the third defense strategy among the four defense strategies is that the crested porcupine will secrete a foul odor that spreads in the surrounding area to prevent the predator from approaching further. The mathematical model is , where r3 is a random integer on [1, N], is a random number on [0, 1], As the parameter used to control the search direction, use the formula definition; For the defense factor, use the formula definition, is the odor diffusion factor. In this embodiment, the preferred odor diffusion factor is in the range of [0.3, 2.6]. When U1=0, the crested porcupine will stop odor diffusion, and the predator will stop moving because of fear of the crested porcupine. At this time, the distance between the predator and the crested porcupine remains unchanged. When U1=1, the crested porcupine will emit a clear odor. The fourth defense strategy among the four defense strategies is that when all previous strategies fail, the crested porcupine will launch a physical attack on the predator. The mathematical model is , where is the convergence speed factor. = 0.2, is a random number on [0, 1], is the average force of the crested porcupine attacking the i-th predator, calculated according to the inelastic collision theorem.

[0031] Specifically, this embodiment provides a preferred implementation, and the value range of the diffusion factor in the odor defense is [0.3, 2.6].

[0032] Specifically, this embodiment provides a preferred implementation, the stacked KELM autoencoder includes a three-layer structure, and the number of nodes in each layer decreases in a ratio of 60%, 40%, and 20%.

[0033] Specifically, this embodiment provides a preferred implementation, wherein the KELM regressor adopts an RBF kernel function, and its bandwidth parameter is optimized and determined by a grid search method.

[0034] Specifically, this embodiment provides a preferred implementation, which also includes using a ten-fold cross validation method to evaluate the MLKELM model.

[0035] Embodiment 2: This embodiment provides a load forecasting system based on the CPO-VMD-MLKELM combined model, which is used to implement the load forecasting method described in the first embodiment.

[0036] Embodiment three: like Figure 4 As shown, this embodiment provides an electronic device, which may include: at least one processor, at least one network interface, a user interface, a memory, and at least one communication bus.

[0037] The communication bus can be used to realize the connection and communication among the above-mentioned components.

[0038] The user interface may include buttons, and the optional user interface may also include a standard wired interface or a wireless interface.

[0039] The network interface may include but is not limited to a Bluetooth module, an NFC module, a Wi-Fi module, etc.

[0040] Among them, the processor may include one or more processing cores. The processor uses various interfaces and lines to connect the various parts of the entire electronic device, and executes various functions of the electronic device and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory, and calling data stored in the memory. Optionally, the processor can be implemented in at least one hardware form of DSP, FPGA, and PLA. The processor can integrate one or a combination of CPU, GPU, modem, etc. Among them, the CPU mainly processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display; the modem is used to handle wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor, but may be implemented separately through a chip.

[0041] Among them, the memory may include RAM or ROM. Optionally, the memory includes a non-transitory computer-readable medium. The memory can be used to store instructions, programs, codes, code sets or instruction sets. The memory may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory may also be at least one storage device located away from the aforementioned processor. The memory as a computer storage medium may include an operating system, a network communication module, a user interface module and a prediction application. The processor may be used to call the prediction application stored in the memory and execute the steps of the load prediction method mentioned in the above-mentioned embodiment.

[0042] Embodiment 4: This embodiment provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium is executed on a computer or a processor, the computer or the processor executes the above-mentioned Figure 1 One or more steps in the illustrated embodiment. If the components of the electronic device described above are implemented in the form of software functional units and sold or used as independent products, they can be stored in the computer-readable storage medium.

[0043] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of this specification is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (Digital Subscriber Line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital versatile disc (DVD)), or a semiconductor medium (eg, a solid state drive (SSD)).

[0044] A person of ordinary skill in the art can understand that all or part of the processes in the method of the first embodiment can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. The aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes. In the absence of conflict, the technical features in this embodiment and the implementation scheme can be combined arbitrarily.

[0045] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described order of actions, because according to the present invention, certain steps can 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 preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0046] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0047] The above is only an exemplary embodiment of the present invention and cannot be used to limit the scope of the present invention. That is, any equivalent changes and modifications made according to the teachings of the present invention are still within the scope of the present invention. After considering the specification and practicing the disclosure here, it will be easy for those skilled in the art to think of the implementation scheme of the present invention. The present invention is intended to cover any modification, use or adaptation of the present invention, which follows the general principles of the present invention and includes common knowledge or customary technical means in the art that are not recorded in the present invention. The description and examples are only regarded as exemplary, and the scope and spirit of the present invention are defined by the claims.

Claims

1. A load forecasting method based on the CPO-VMD-MLKELM combined model, characterized in that: Includes steps: S1, obtaining historical load data, meteorological characteristic data and date characteristic data of the forecast target and preprocessing them to obtain a preprocessed data set; S2, using PCA principal component analysis to perform feature dimensionality reduction on the preprocessed data set, and retaining the principal components with cumulative variance contribution rate not less than 98% to obtain the principal component features after dimensionality reduction; S3, using the crown porcupine optimization algorithm to dynamically optimize VMD parameters through four defense strategies, and decomposing the historical load data into multiple IMF components based on the optimized VMD parameters, wherein the VMD parameters include a modal number K and a penalty factor α; S4. Construct an MLKELM model for each of the IMF components, use the principal component features after dimensionality reduction as input to predict each of the IMF components, and superimpose the predicted values ​​of each of the IMF components to obtain a load prediction value. The MLKELM model includes a stacked KELM autoencoder and a KELM regressor.

2. The load forecasting method based on the CPO-VMD-MLKELM combined model according to claim 1, characterized in that: The preprocessing in step S1 includes one-hot encoding the date feature data, and the one-hot encoding is specifically: Weekdays are coded as 0 and holidays are coded as 1; Seasonal features are coded as [1,0,0,0], [0,1,0,0], [0,0,1,0], [0,0,0,1] for spring, summer, autumn, and winter, respectively.

3. The load forecasting method based on the CPO-VMD-MLKELM combined model according to claim 2, characterized in that: The four defense strategies are visual defense, auditory defense, odor defense and physical attack; The visual defense updates parameters based on the current optimal solution and a normally distributed random number; The auditory defense determines the search direction through two random individual positions; The odor defense adopts a diffusion factor to control the parameter adjustment range; The physical attack is based on the inelastic collision theorem to modify the parameters.

4. The load forecasting method based on the CPO-VMD-MLKELM combined model according to claim 3 is characterized by: The value range of the diffusion factor in the odor defense is [0.3, 2.6].

5. The load forecasting method based on the CPO-VMD-MLKELM combined model according to claim 4, characterized in that: The stacked KELM autoencoder includes a three-layer structure, and the number of nodes in each layer decreases in the ratio of 60%, 40%, and 20%.

6. The load forecasting method based on the CPO-VMD-MLKELM combined model according to claim 5, characterized in that: The KELM regressor adopts the RBF kernel function, and its bandwidth parameter is optimized and determined by the grid search method.

7. The load forecasting method based on the CPO-VMD-MLKELM combined model according to claim 6, characterized in that: It also includes evaluating the MLKELM model using a ten-fold cross validation method.

8. A load forecasting system based on the CPO-VMD-MLKELM combined model, characterized in that: Used to implement the load forecasting method as described in any one of claims 1 to 7.

9. A computer device, comprising a memory, a processor and a computer program, characterized in that: When the computer program is executed by a processor, the load forecasting method according to any one of claims 1 to 7 is implemented.

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

Images