A medium- and long-term prediction method and device for multi-loads of an integrated energy system

Abstract

The present invention discloses a method and device for medium- and long-term prediction of multi-loads in an integrated energy system, the method comprising initializing a neural network; classifying samples; inputting the input samples into the neural network, outputting training prediction data, and calculating the error between the training prediction data and the historical load data of the multi-loads to obtain an error result; based on the obtained error result, correcting the optimal network weights and wavelet parameters in the neural network based on the improved bee mating optimization algorithm; iteratively correcting the optimal network weights and wavelet parameters in the neural network until the obtained error result is less than a preset threshold; inputting the obtained real-time influencing factors of the multi-load demand into the neural network, and outputting the real-time multi-load prediction result based on the iteratively corrected optimal network weights and wavelet parameters. On the basis of improving the existing method, the gene crossover of the bee colony is performed by the method of vertical and horizontal crossover mutation to improve the performance of the algorithm.

Description

Technical Field

[0001] The present invention belongs to the technical field of integrated energy systems, and in particular relates to a method and device for mid- and long-term prediction of multiple loads in an integrated energy system. Background Art

[0002] At the present stage, the planning and operation of the integrated energy system aims to ensure the supply and demand balance of various energy sources such as electricity, gas, heat and cold. In addition, various energy sources in the integrated energy system replace each other through complex coupling mechanisms, and the factors affecting various load demands are more complex. At the same time, diversified load forecasting for integrated energy systems is still in its infancy. Some preliminary explorations have been carried out at home and abroad, such as the deep structure multi-task learning forecasting method and the IPSO-WNN forecasting method. However, the influencing factors and load forecasting methods considered are not mature, and it is necessary to conduct more in-depth exploration.

[0003] In the prior art, the gradient descent algorithm is usually used to correct the neural network node weights and the wavelet function scaling factor and translation factor, but some data features in the pre-training process of the neural network are ignored, resulting in the incomplete transmission of data features and a slow convergence speed, which will affect the accuracy and speed of neural network prediction. Summary of the invention

[0004] The present invention provides a method and device for medium- and long-term prediction of multiple loads in an integrated energy system, which are used to solve at least one of the above-mentioned technical problems.

[0005] In the first aspect, the present invention provides a method for medium- and long-term prediction of multi-element loads in an integrated energy system, comprising: step 1, initializing a neural network: determining the levels in the network structure, the number of neuron nodes in each level, initializing the optimal network weights and wavelet parameters in the neural network, setting the learning rate of the network and the upper limit of the error at the end of training; step 2, sample classification: taking the historical influencing factors of multi-element load demand as input samples, and the historical load data of the multi-element load as comparison samples; step 3, inputting the input samples into the neural network, outputting training prediction data, and calculating the error between the training prediction data and the historical load data of the multi-element load, so as to obtain the error result ; Step 4, according to the error result obtained, based on the improved bee mating optimization algorithm, correct the optimal network weights and wavelet parameters in the neural network, wherein the improved bee mating optimization algorithm includes cross-mutation of the male peak genes stored in the queen bee and the queen bee genes in the bee mating optimization algorithm by horizontal crossover and vertical crossover; Step 5, iteratively correct the optimal network weights and wavelet parameters in the neural network until the error result obtained is less than the preset threshold; Step 6, input the real-time influencing factors of the obtained multi-load demand into the neural network, and output the real-time multi-load prediction results based on the iteratively corrected optimal network weights and wavelet parameters.

[0006] In a second aspect, the present invention provides a medium- and long-term prediction device for multiple loads of an integrated energy system, comprising: an initialization module, configured to initialize a neural network: determine the levels in the network structure, the number of neuron nodes in each level, and initialize the optimal network weights and wavelet parameters in the neural network, set the learning rate of the network and the upper limit of the error at the end of training; a classification module, configured to classify samples: use the historical influencing factors of multiple load demands as input samples, and the historical load data of multiple loads as comparison samples; a first input module, configured to input the input samples into the neural network, output training prediction data, and calculate the error between the training prediction data and the historical load data of the multiple loads, so that the error result is result; a correction module, configured to correct the optimal network weights and wavelet parameters in the neural network based on the improved bee mating optimization algorithm according to the obtained error result, wherein the improved bee mating optimization algorithm includes cross-mutating the male peak genes stored in the queen bee body and the queen bee genes in the bee mating optimization algorithm by horizontal crossover and vertical crossover; an iteration module, configured to iteratively correct the optimal network weights and wavelet parameters in the neural network until the obtained error result is less than a preset threshold; a second input module, configured to input the obtained real-time influencing factors of the multi-load demand into the neural network, and output the real-time multi-load prediction result based on the iteratively corrected optimal network weights and wavelet parameters.

[0007] According to a third aspect, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the steps of a method for medium- and long-term forecasting of multi-loads in an integrated energy system according to any embodiment of the present invention.

[0008] In a fourth aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program includes program instructions, and when the program instructions are executed by a computer, the computer executes the steps of a method for medium- and long-term forecasting of multiple loads in an integrated energy system according to any embodiment of the present invention.

[0009] The present application discloses a method and device for medium- and long-term prediction of multiple loads in an integrated energy system. Aiming at the problem that the factors affecting the demand of multiple loads in integrated energy are complex and the impact uncertainty is strong, based on the improvement of the WNN method, a method of vertical and horizontal crossover mutation is used to perform gene crossover of the bee colony to improve the performance of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A flow chart of a method for mid- to long-term prediction of multiple loads in an integrated energy system provided by an embodiment of the present invention;

[0012] Figure 2 A structural block diagram of a medium- and long-term prediction device for multiple loads of an integrated energy system provided by an embodiment of the present invention;

[0013] Figure 3 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0014] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0015] See also Figure 1 , which shows a flow chart of a method for medium- and long-term prediction of multiple loads in an integrated energy system of the present application.

[0016] like Figure 1 As shown, a method for medium- and long-term prediction of multi-element loads in an integrated energy system includes the following steps:

[0017] Step 1: Initialize the neural network: determine the number of layers in the network structure, the number of neuron nodes in each layer, initialize the optimal network weights and wavelet parameters in the neural network, set the network's learning rate and the upper limit of the error at the end of training.

[0018] Step 2: Sample classification: The historical influencing factors of multi-load demand are used as input samples, and the historical load data of multi-load is used as comparison samples.

[0019] In this embodiment, the historical influencing factors of the multi-load demand include an electric-thermal coupling influencing factor, an electric-gas coupling influencing factor, a gas-electric coupling influencing factor, and a gas-thermal coupling influencing factor;

[0020] Among them, the expression of the electro-thermal coupling influencing factor is:

[0021]

[0022] In the formula, K P-H is the influence factor of electric-thermal coupling, N P-H is the total number of electric heating devices in the system, For nth P-H Rated heat output of Taipower heating equipment, is the annual maximum heat load value of the system;

[0023] The expression of the influencing factor of electric-gas coupling is:

[0024]

[0025] In the formula, K P-G is the influence factor of electric-gas coupling, N P-G is the total number of electric gas generation equipment in the system, For nth P-G Rated gas production capacity of Taipower gas production equipment, is the annual maximum natural gas load value of the system;

[0026] The expression of the gas-electric coupling influencing factor is:

[0027]

[0028] In the formula, K G-P is the gas-electric coupling influencing factor, N G-P is the total number of gas-fired power generation equipment in the system, For nth G-P Rated output power of gas-fired power generation equipment, is the annual maximum electrical load value of the system;

[0029] The expression of the gas-heat coupling influencing factor is:

[0030]

[0031] In the formula, K G-H is the gas-heat coupling influencing factor, N G-H is the total number of gas heating equipment in the system, For nth G-H Rated heating power of gas heating equipment, It is the maximum annual heat load value of the system.

[0032] Step 3: Input the input sample into the neural network, output the training prediction data, and calculate the error between the training prediction data and the historical load data of the multivariate load to obtain the error result.

[0033] Step 4: According to the error results obtained, the optimal network weights and wavelet parameters in the neural network are corrected based on the improved honeybee mating optimization algorithm, wherein the improved honeybee mating optimization algorithm includes cross-mutating the male peak genes stored in the queen bee body and the queen bee genes in the honeybee mating optimization algorithm by horizontal crossover and vertical crossover.

[0034] In this embodiment, the expression of the horizontal intersection is:

[0035]

[0036]

[0037] In the formula, They are The d-th dimension offspring generated by the horizontal crossover operation and The d-th dimension offspring generated by the horizontal crossover operation, The parental gene combination The d-th dimension and parent gene combination In the d-th dimension, χ1 and χ2 are both random numbers ranging from 0 to 1, and ε1 and ε2 are both random numbers ranging from -1 to 1.

[0038] The expression of the longitudinal intersection is:

[0039]

[0040] In the formula, The parent gene combination The offspring generated by the vertical crossover operation between the d1th dimension and the d2th dimension, The parental gene combination The d1th dimension and parent gene combination The d2th dimension of , χ is a random number between 0 and 1.

[0041] Step 5: Iteratively modify the optimal network weights and wavelet parameters in the neural network until the error result obtained is less than a preset threshold.

[0042] Step 6: Input the acquired real-time influencing factors of the multivariate load demand into the neural network, and output the real-time multivariate load forecasting results based on the iteratively corrected optimal network weights and wavelet parameters.

[0043] In summary, the method of this embodiment uses the bee mating algorithm to iteratively correct the network weights and wavelet parameters in the neural network to improve the prediction accuracy, and uses the horizontal and vertical crossover method to improve the gene crossover part of the bee mating algorithm. The horizontal crossover is to execute the crossover operator to randomly combine all the drone genes in the queen bee with the queen bee genes, and then perform horizontal crossover (crossover between parent gene combinations) on all dimensions of each combined gene to update the gene combination. The vertical crossover is the crossover between different gene combinations of the parent generation, the offspring generation, etc., to solve the local optimal problem that the bee mating algorithm may fall into.

[0044] In a specific embodiment, the improved BMO-WNN method is used to predict the medium- and long-term multi-load demand of the integrated energy system, and the traditional WNN prediction method and the BMO-WNN prediction method are used for comparison. This technology verifies the effectiveness of the proposed prediction method through examples and compares the prediction efficiency of different methods. The electricity, gas, heat, and cooling load data and related influencing factor data of a selected area in Beijing from 2003 to 2018 for a total of 15 years are used as training samples to perform multi-load prediction of the integrated energy system.

[0045] In order to verify the effectiveness of the improved BMO-WNN prediction method proposed in this method, the training time and prediction relative error of each algorithm are selected as the main comparison indicators to compare the effects of each prediction algorithm, as shown in the following table.

[0046] Comparison of the performance of various prediction algorithms

[0047]

[0048] It can be seen from the table that the prediction accuracy of the proposed improved BM0-WNN method is better than that of the BMO-WNN and traditional WNN prediction methods. The optimization improvement of network weights and wavelet function parameters based on the WNN method effectively overcomes the problem that the WNN method uses the gradient descent algorithm to ignore some data features in the pre-training process of the neural network, which leads to a slow convergence speed. In terms of training time, the BM0-WNN method and the improved BMO-WNN method are better than the traditional WNN method.

[0049] See also Figure 2 , which shows a structural block diagram of a medium- and long-term prediction device for multiple loads of an integrated energy system of the present application.

[0050] like Figure 2 As shown, the medium- and long-term prediction device 200 for multiple loads in an integrated energy system includes an initialization module 210 , a classification module 220 , a first input module 230 , a correction module 240 , an iteration module 250 and a second input module 260 .

[0051] Among them, the initialization module 210 is configured to initialize the neural network: determine the layers in the network structure, the number of neuron nodes in each layer, and initialize the optimal network weights and wavelet parameters in the neural network, set the learning rate of the network and the upper limit of the error at the end of training; the classification module 220 is configured to classify samples: take the historical influencing factors of multivariate load demand as input samples, and the historical load data of multivariate loads as comparison samples; the first input module 230 is configured to input the input samples into the neural network, output training prediction data, and calculate the error between the training prediction data and the historical load data of multivariate loads, so as to obtain the error result; the correction module 240 is configured to The error result obtained is based on the use of an improved bee mating optimization algorithm to correct the optimal network weights and wavelet parameters in the neural network, wherein the improved bee mating optimization algorithm includes cross-mutating the male peak genes stored in the queen bee body and the queen bee genes in the bee mating optimization algorithm by horizontal crossover and vertical crossover; the iteration module 250 is configured to iteratively correct the optimal network weights and wavelet parameters in the neural network until the error result obtained is less than a preset threshold; the second input module 260 is configured to input the obtained real-time influencing factors of the multi-load demand into the neural network, and output the real-time multi-load prediction result based on the iteratively corrected optimal network weights and wavelet parameters.

[0052] It should be understood that Figure 2 Modules and references documented in Figure 1 Therefore, the operations and features described above for the method and the corresponding technical effects are also applicable to Figure 2 The modules in it will not be described in detail here.

[0053] In other embodiments, the present invention also provides a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions can execute the medium- and long-term prediction method for multi-loads of the integrated energy system in any of the above method embodiments;

[0054] As an implementation mode, the computer-readable storage medium of the present invention stores computer-executable instructions, and the computer-executable instructions are configured as follows:

[0055] Initialize the neural network: determine the layers in the network structure, the number of neuron nodes in each layer, initialize the optimal network weights and wavelet parameters in the neural network, set the network's learning rate and the upper limit of the error at the end of training;

[0056] Sample classification: The historical influencing factors of multi-load demand are used as input samples, and the historical load data of multi-load is used as comparison samples;

[0057] Input the input sample to the neural network, output training prediction data, and calculate the error between the training prediction data and the historical load data of the multivariate load to obtain an error result;

[0058] According to the error results obtained, the optimal network weights and wavelet parameters in the neural network are corrected based on the improved honey bee mating optimization algorithm, wherein the improved honey bee mating optimization algorithm includes crossover mutation of the male peak gene stored in the queen bee body and the queen bee gene in the honey bee mating optimization algorithm by horizontal crossover and vertical crossover;

[0059] Iteratively modify the optimal network weights and wavelet parameters in the neural network until the error result obtained is less than a preset threshold;

[0060] The acquired real-time influencing factors of multivariate load demand are input into the neural network, and the real-time multivariate load forecasting results are output based on the iteratively corrected optimal network weights and wavelet parameters.

[0061] The computer-readable storage medium may include a program storage area and a data storage area, wherein the program storage area may store an operating system and application programs required for at least one function; the data storage area may store data created according to the use of the medium- and long-term prediction device for multiple loads of an integrated energy system, etc. In addition, the computer-readable storage medium may include a high-speed random access memory, and may also include a memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the medium- and long-term prediction device for multiple loads of an integrated energy system via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0062] Figure 3 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 3 As shown, the device includes: a processor 310 and a memory 320. The electronic device may also include: an input device 330 and an output device 340. The processor 310, the memory 320, the input device 330 and the output device 340 may be connected via a bus or other means. Figure 3The example of the bus connection is taken. The memory 320 is the computer-readable storage medium mentioned above. The processor 310 executes various functional applications and data processing of the server by running the non-volatile software programs, instructions and modules stored in the memory 320, that is, the medium- and long-term prediction method of multiple loads in the integrated energy system of the above method embodiment is realized. The input device 330 can receive input digital or character information, and generate key signal input related to user settings and function control of the medium- and long-term prediction device of multiple loads in the integrated energy system. The output device 340 may include display devices such as display screens.

[0063] The electronic device can execute the method provided by the embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in this embodiment, please refer to the method provided by the embodiment of the present invention.

[0064] As an implementation mode, the electronic device is applied to a medium- and long-term prediction device for multiple loads of an integrated energy system, and is used for a client, and includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can:

[0065] Initialize the neural network: determine the layers in the network structure, the number of neuron nodes in each layer, initialize the optimal network weights and wavelet parameters in the neural network, set the network's learning rate and the upper limit of the error at the end of training;

[0066] Sample classification: The historical influencing factors of multi-load demand are used as input samples, and the historical load data of multi-load is used as comparison samples;

[0067] Input the input sample to the neural network, output training prediction data, and calculate the error between the training prediction data and the historical load data of the multivariate load to obtain an error result;

[0068] According to the error results obtained, the optimal network weights and wavelet parameters in the neural network are corrected based on the improved honey bee mating optimization algorithm, wherein the improved honey bee mating optimization algorithm includes crossover mutation of the male peak gene stored in the queen bee body and the queen bee gene in the honey bee mating optimization algorithm by horizontal crossover and vertical crossover;

[0069] Iteratively modify the optimal network weights and wavelet parameters in the neural network until the error result obtained is less than a preset threshold;

[0070] The acquired real-time influencing factors of multivariate load demand are input into the neural network, and the real-time multivariate load forecasting results are output based on the iteratively corrected optimal network weights and wavelet parameters.

[0071] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiment.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A medium- and long-term forecasting method for multi-loads in an integrated energy system. It is characterized in that include: Step 1: Initialize the neural network: determine the levels in the network structure, the number of neuron nodes in each level, initialize the optimal network weights and wavelet parameters in the neural network, set the learning rate of the network and the upper limit of the error at the end of training; Step 2, sample classification: taking the historical influencing factors of the multi-load demand as input samples and the historical load data of the multi-load as comparison samples, the historical influencing factors of the multi-load demand include electric-thermal coupling influencing factors, electric-gas coupling influencing factors, gas-electric coupling influencing factors and gas-thermal coupling influencing factors; The expression of the electro-thermal coupling influencing factor is: ， In the formula, is the influencing factor of electrical-thermal coupling, is the total number of electric heating devices in the system, For the Rated heat output of Taipower heating equipment, is the annual maximum heat load value of the system; The expression of the influencing factor of electric-gas coupling is: ， In the formula, is the influencing factor of electric-gas coupling, is the total number of electric gas generation equipment in the system, For the Rated gas production capacity of Taipower gas production equipment, is the annual maximum natural gas load value of the system; The expression of the gas-electric coupling influencing factor is: ， In the formula, is the gas-electric coupling influencing factor, is the total number of gas-fired power generation equipment in the system, For the Rated output power of gas-fired power generation equipment, is the annual maximum electrical load value of the system; The expression of the gas-heat coupling influencing factor is: ， In the formula, is the gas-heat coupling influencing factor, is the total number of gas heating equipment in the system, For the Rated heating power of gas heating equipment, is the annual maximum heat load value of the system; Step 3: Input the input sample into the neural network, output the training prediction data, and calculate the error between the training prediction data and the historical load data of the multivariate load to obtain the error result; Step 4: According to the error results obtained, the optimal network weights and wavelet parameters in the neural network are corrected based on the improved honey bee mating optimization algorithm, wherein the improved honey bee mating optimization algorithm includes crossover and vertical crossover to perform crossover mutation on the male peak genes stored in the queen bee and the queen bee genes in the honey bee mating optimization algorithm, and the expression of the horizontal crossover is: ， In the formula, , They are The first Dimensional The first Wikipedia, , The parental gene combination No. Genetic combination of the parent No. dimension, , are all random numbers between 0 and 1. , All are random numbers between -1 and 1; The expression of the longitudinal intersection is: ， In the formula, The parent gene combination No. Peace The offspring produced by the vertical crossover operation, , The parental gene combination No. Genetic combination of the parent No. dimension, A random number between 0 and 1; Step 5: Iteratively modify the optimal network weights and wavelet parameters in the neural network until the error result obtained is less than a preset threshold; Step 6: Input the acquired real-time influencing factors of the multivariate load demand into the neural network, and output the real-time multivariate load forecasting results based on the iteratively corrected optimal network weights and wavelet parameters.

2. A medium- and long-term forecasting device for multiple loads of an integrated energy system according to the method of claim 1, It is characterized in that include: The initialization module is configured to initialize the neural network: determine the layers in the network structure, the number of neuron nodes in each layer, initialize the optimal network weights and wavelet parameters in the neural network, set the learning rate of the network and the upper limit of the error at the end of training; The classification module is configured as sample classification: the historical influencing factors of multivariate load demand are used as input samples, and the historical load data of multivariate loads are used as comparison samples; A first input module is configured to input the input sample into the neural network, output training prediction data, and calculate the error between the training prediction data and the historical load data of the multivariate load to obtain an error result; A correction module is configured to correct the optimal network weights and wavelet parameters in the neural network based on the obtained error results and based on the use of an improved honeybee mating optimization algorithm, wherein the improved honeybee mating optimization algorithm includes crossover mutation of the male peak gene stored in the queen bee body and the queen bee gene in the honeybee mating optimization algorithm by using horizontal crossover and vertical crossover; An iteration module, configured to iteratively correct the optimal network weights and wavelet parameters in the neural network until the error result obtained is less than a preset threshold; The second input module is configured to input the acquired real-time influencing factors of the multivariate load demand into the neural network, and output the real-time multivariate load forecasting result based on the iteratively corrected optimal network weights and wavelet parameters.

3. An electronic device, It is characterized in that include: At least one processor, and a memory communicatively connected to the at least one processor, wherein 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 the method of claim 1.

4. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the program is executed by a processor, the method according to claim 1 is implemented.

Images