Multi-model training method for tracking load prediction

Abstract

The invention relates to a multi-model training method for tracking load prediction, and the method comprises the steps: obtaining a tracking load prediction data set, taking the historical meteorological data, historical electricity price data and historical power station operation data of a region as a sample main body of a sample, and taking the historical tracking load data of the region as a sample label; dividing the tracking load prediction data set into a training set, a verification set and a test set; training a plurality of different preset models according to the training set and the verification set to obtain a plurality of tracking load prediction models; performing tracking load prediction on each sample in the test set by using the plurality of tracking load prediction models; according to the tracking load prediction results corresponding to the plurality of tracking load prediction models and the sample label of each sample in the test set, calculating tracking load prediction weights corresponding to the plurality of tracking load prediction models, and carrying out weighted integration on the plurality of tracking load prediction models to obtain a comprehensive tracking load prediction model, and thus, the tracking load is accurately predicted.

Description

Technical Field

[0001] The present invention relates to the technical field of centralized load forecasting, and in particular to a multi-model training method for centralized load forecasting. Background Art

[0002] Since the efficient dispatch of the power system depends on the balance between the power generation side and the power consumption side, and the power load is volatile and uncertain, accurate load forecasting is helpful to provide an effective reference for the balance between power generation and power consumption.

[0003] The common method of centralized load forecasting mainly inputs meteorological data and centralized load data into a machine learning model, trains the model parameters to obtain a forecasting model, and then inputs the currently acquired meteorological data into the forecasting model to obtain the centralized load forecasting result. However, this method of centralized load forecasting generally relies on a single model for forecasting, which is easily affected by the shortcomings of the model itself and has poor forecasting accuracy. In view of this, how to accurately predict the centralized load has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] An embodiment of the present invention provides a multi-model training method for centralized load forecasting.

[0005] In a first aspect, an embodiment of the present invention provides a multi-model training method for centralized load forecasting, the method comprising:

[0006] Obtain a unified dispatching load forecasting data set, wherein the samples in the unified dispatching load forecasting data set are based on the historical meteorological data, historical electricity price data, and historical power station operation data of the region, and the historical unified dispatching load data of the region is used as the sample label;

[0007] The centralized load forecasting data set is divided into a training set, a validation set, and a test set;

[0008] A plurality of different preset models are trained according to the training set and the validation set to obtain a plurality of unified dispatching load forecasting models;

[0009] Using multiple unified load forecasting models to forecast unified loads for each sample in the test set, and obtaining unified load forecasting results corresponding to the multiple unified load forecasting models;

[0010] Calculate the unified load prediction weights corresponding to the multiple unified load prediction models according to the unified load prediction results corresponding to the multiple unified load prediction models and the sample labels of each sample in the test set;

[0011] According to the centralized load prediction weights corresponding to the multiple centralized load prediction models, the multiple centralized load prediction models are weighted integrated to obtain a comprehensive centralized load prediction model.

[0012] In some embodiments, obtaining a centralized load forecasting data set includes:

[0013] Obtain regional meteorological data, electricity price data, power station operation data, and coordinated load data within a historical period;

[0014] Samples are generated based on the acquired meteorological data, electricity price data, power station operation data and unified load data, and the unified load forecast data set is generated based on this.

[0015] In some embodiments, the centralized load forecasting data set is divided into a training set, a validation set, and a test set, including:

[0016] Preprocessing the unified dispatching load forecasting data set, where the preprocessing includes: outlier removal, missing value filling, and normalization processing;

[0017] Dividing the historical time period into a first historical time period, a second historical time period, and a third historical time period according to a preset division ratio;

[0018] Dividing the samples corresponding to the first historical time period in the preprocessed centralized load forecasting data set into a training set;

[0019] Dividing the samples corresponding to the second historical time period in the preprocessed centralized load forecasting data set into a validation set;

[0020] The samples corresponding to the third historical time period in the preprocessed coordinated load forecasting data set are divided into a test set.

[0021] In some embodiments, according to the centralized load prediction results corresponding to the multiple centralized load prediction models and the sample labels of each sample in the test set, the centralized load prediction weights corresponding to the multiple centralized load prediction models are calculated, including:

[0022] According to the unified dispatching load forecasting results corresponding to the multiple unified dispatching load forecasting models and the sample labels of each sample in the test set, the unified dispatching load forecasting variance corresponding to the multiple unified dispatching load forecasting models is calculated;

[0023] The unified dispatching load prediction weights corresponding to the multiple unified dispatching load prediction models are calculated according to the unified dispatching load prediction variances corresponding to the multiple unified dispatching load prediction models.

[0024] In some embodiments, the unified load prediction variance corresponding to the multiple unified load prediction models is calculated according to the unified load prediction results corresponding to the multiple unified load prediction models and the sample label of each sample in the test set, including:

[0025] Calculate the unified load prediction error of the unified load prediction model at each sample in the test set according to the unified load prediction result corresponding to the unified load prediction model and the sample label of each sample in the test set;

[0026] Calculate an average centralized load prediction error according to the centralized load prediction error of the centralized load prediction model at each sample in the test set;

[0027] The centralized load prediction variance corresponding to the centralized load prediction model is calculated according to the centralized load prediction error of the centralized load prediction model at each sample in the test set and the average centralized load prediction error.

[0028] In some embodiments, calculating the unified load prediction variance corresponding to the unified load prediction model according to the unified load prediction error of the unified load prediction model at each sample in the test set and the average unified load prediction error includes:

[0029] According to the centralized load prediction error of the centralized load prediction model at each sample in the test set and the average centralized load prediction error, the centralized load prediction variance corresponding to the centralized load prediction model is calculated using the following formula:

[0030]

[0031] in, is the unified load prediction variance corresponding to the unified load prediction model; is the load forecasting error of the load forecasting model i at the jth sample in the test set, n is the number of samples in the test set, is the average coordinated load forecast error.

[0032] In some embodiments, the unified dispatching load prediction weights corresponding to the multiple unified dispatching load prediction models are calculated according to the unified dispatching load prediction variances corresponding to the multiple unified dispatching load prediction models, including:

[0033] Adding the inverses of the unified load prediction variances corresponding to the multiple unified load prediction models;

[0034] For any unified load forecasting model, the reciprocal of the unified load forecasting variance corresponding to the unified load forecasting model is divided by the addition result to obtain the unified load forecasting weight corresponding to the unified load forecasting model.

[0035] In a second aspect, an embodiment of the present invention provides a multi-model training method for centralized load forecasting, the method comprising:

[0036] Obtain regional weather forecast data, electricity price forecast data, and power station operation forecast data;

[0037] Using multiple unified dispatching load forecasting models in the comprehensive unified dispatching load forecasting model to process weather forecast data, electricity price forecast data, and power station operation forecast data to complete unified dispatching load forecasting and obtain unified dispatching load forecasting results corresponding to the multiple unified dispatching load forecasting models;

[0038] According to the unified dispatching load prediction weights corresponding to the multiple unified dispatching load prediction models, weighted summation is performed on the unified dispatching load prediction results corresponding to the multiple unified dispatching load prediction models to obtain a comprehensive unified dispatching load prediction result;

[0039] Among them, the comprehensive coordinated load forecasting model is obtained based on the multi-model training method as described above.

[0040] In a third aspect, an embodiment of the present invention provides a multi-model training device for centralized load forecasting, the device comprising:

[0041] An acquisition module is used to acquire a unified load forecasting data set, wherein the samples in the unified load forecasting data set are based on the historical meteorological data, historical electricity price data, and historical power station operation data of the region, and the historical unified load data of the region is used as the sample label;

[0042] A partitioning module is used to divide the centralized load forecasting data set into a training set, a validation set, and a test set;

[0043] A training module is used to train multiple different preset models according to a training set and a validation set to obtain multiple unified load forecasting models;

[0044] A prediction module is used to perform a unified load prediction on each sample in the test set using multiple unified load prediction models, and obtain unified load prediction results corresponding to the multiple unified load prediction models;

[0045] A calculation module, used to calculate the unified dispatching load prediction weights corresponding to the multiple unified dispatching load prediction models according to the unified dispatching load prediction results corresponding to the multiple unified dispatching load prediction models and the sample labels of each sample in the test set;

[0046] The integration module is used to perform weighted integration on the multiple centralized load prediction models according to the centralized load prediction weights corresponding to the multiple centralized load prediction models to obtain a comprehensive centralized load prediction model.

[0047] In a fourth aspect, an embodiment of the present invention provides a multi-model-based centralized load forecasting device, the device comprising:

[0048] The acquisition module is used to obtain regional weather forecast data, electricity price forecast data, and power station operation forecast data;

[0049] A prediction module is used to process meteorological forecast data, electricity price forecast data, and power station operation forecast data using multiple unified load forecast models in the comprehensive unified load forecast model to complete unified load forecasting and obtain unified load forecasting results corresponding to the multiple unified load forecasting models;

[0050] A summing module is used to perform weighted summation of the unified dispatching load prediction results corresponding to the multiple unified dispatching load prediction models according to the unified dispatching load prediction weights corresponding to the multiple unified dispatching load prediction models, so as to obtain a comprehensive unified dispatching load prediction result;

[0051] Among them, the comprehensive coordinated load forecasting model is obtained based on the multi-model training method as described above.

[0052] In a fifth aspect, an embodiment of the present invention provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; 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 method as described above.

[0053] In a sixth aspect, an embodiment of the present invention provides a non-transitory computer-readable storage medium storing computer instructions, where the computer instructions are used to cause a computer to execute the method described above.

[0054] In an embodiment of the present invention, a centralized load forecasting data set can be obtained, in which samples are based on regional historical meteorological data, historical electricity price data, and historical power station operation data, and regional historical centralized load data are used as sample labels; the centralized load forecasting data set is divided into a training set, a verification set, and a test set; a plurality of different preset models are trained according to the training set and the verification set; the centralized load forecasting models obtained through training are used to forecast the centralized load for each sample in the test set; the centralized load forecasting weights corresponding to the plurality of centralized load forecasting models are calculated according to the centralized load forecasting results corresponding to the plurality of centralized load forecasting models and the sample labels of each sample in the test set, so as to perform weighted integration of the plurality of centralized load forecasting models to obtain a comprehensive centralized load forecasting model, and then accurately forecast the centralized load based on the comprehensive centralized load forecasting model, so as to provide an effective reference for the formulation of power generation plans and the balance of power generation and consumption.

[0055] It should be understood that the contents described in the summary of the invention are not intended to limit the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The above and other features, advantages and aspects of the embodiments of the present invention will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. The accompanying drawings are provided for a better understanding of the present invention and do not constitute a limitation of the present invention. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:

[0057] Figure 1 A flowchart of a multi-model training method provided by an embodiment of the present invention is shown;

[0058] Figure 2 A flowchart of a method for predicting loads based on multiple models provided by an embodiment of the present invention is shown;

[0059] Figure 3 A structural diagram of a multi-model training device provided by an embodiment of the present invention is shown;

[0060] Figure 4 A structural diagram of a multi-model-based centralized load forecasting device provided by an embodiment of the present invention is shown;

[0061] Figure 5 A block diagram of an exemplary electronic device capable of implementing an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0062] 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 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.

[0063] In addition, the term "and / or" in this article is only a description of the association relationship between the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0064] In response to the problems arising in the background technology, an embodiment of the present invention provides a multi-model training method for centralized load forecasting. Specifically, a centralized load forecasting data set is obtained, wherein the samples are based on the historical meteorological data, historical electricity price data, and historical power station operation data of the region as the sample body, and the historical centralized load data of the region is used as the sample label; the centralized load forecasting data set is divided into a training set, a validation set, and a test set; a plurality of different preset models are trained according to the training set and the validation set; the centralized load forecasting is performed on each sample in the test set using the multiple centralized load forecasting models obtained through training; the centralized load forecasting weights corresponding to the multiple centralized load forecasting models are calculated according to the centralized load forecasting results corresponding to the multiple centralized load forecasting models and the sample labels of each sample in the test set, so as to perform weighted integration on the multiple centralized load forecasting models to obtain a comprehensive centralized load forecasting model, and then based on the comprehensive centralized load forecasting model, the centralized load is accurately predicted, so as to provide an effective reference basis for the formulation of power generation plans and the balance of power generation and consumption.

[0065] In conjunction with the accompanying drawings, a multi-model training method for centralized load forecasting provided by an embodiment of the present invention is described in detail below through specific embodiments.

[0066] Figure 1 A flowchart of a multi-model training method provided by an embodiment of the present invention is shown. Figure 1 As shown, the multi-model training method 100 may include the following steps:

[0067] S110, obtaining a centralized load forecasting data set.

[0068] Among them, the samples in the unified dispatching load forecasting dataset are based on the region’s historical meteorological data, historical electricity price data, and historical power station operation data, and the region’s historical unified dispatching load data are used as sample labels.

[0069] Exemplarily, in order to train a more general centralized load forecasting model, the region here may be multiple different regions; historical meteorological data may include: air pressure data, temperature data, humidity data, radiation data, wind data, etc.; electricity price data may be dynamic electricity price data; power plant operation data may be generator set operation data of power plants in the region; and centralized load data may be centralized power load demand data.

[0070] It is understandable that in order to achieve unified load forecasting of different time scales, the time scale of the samples in the unified load forecasting data set here can be adjusted as needed. For example, for short-term unified load forecasting, the time scale of the samples is short-term; for medium-term unified load forecasting, the time scale of the samples is medium-term; and for long-term unified load forecasting, the time scale of the samples is long-term.

[0071] In some embodiments, meteorological data, electricity price data, power station operation data and unified load data of a region in a historical time period can be obtained, and then samples can be generated based on the obtained meteorological data, electricity price data, power station operation data and unified load data to generate a unified load prediction data set.

[0072] For example, the meteorological data, electricity price data, power station operation data and unified load data of the region in the past month can be sampled to obtain the meteorological data, electricity price data, power station operation data and unified load data of the region in the past month, where the sampling interval can be 15 minutes, and then samples are generated based on the meteorological data, electricity price data, power station operation data and unified load data in the past month, and the unified load forecast data set is generated accordingly.

[0073] S120, dividing the centralized load forecasting data set into a training set, a validation set, and a test set.

[0074] It is worth noting that in order to improve the division effect of the training set and the test set, the centralized load forecasting data set can be divided in the following ways:

[0075] The centralized load forecasting data set is preprocessed, where the preprocessing includes: outlier removal (i.e., removing outliers in the original data), missing value filling (i.e., filling missing values ​​in the original data), normalization processing (i.e., linearly changing the original data to the interval [-1,1]), etc., and then the preprocessed centralized load forecasting data set is divided into a training set, a validation set, and a test set.

[0076] Optionally, the historical time period can be divided into a first historical time period, a second historical time period, and a third historical time period according to a preset division ratio, and the samples corresponding to the first historical time period in the preprocessed unified load forecast data set are divided into a training set, the samples corresponding to the second historical time period in the preprocessed unified load forecast data set are divided into a verification set, and the samples corresponding to the third historical time period in the preprocessed unified load forecast data set are divided into a test set.

[0077] For example, referring to the example in S110, the past month (30 days) can be divided into the first 20 days, the middle 5 days, and the last 5 days according to 4:1:1, and then the samples belonging to the first 20 days in the pre-processed unified load forecast data set are divided into the training set, the samples belonging to the middle 5 days in the pre-processed unified load forecast data set are divided into the validation set, and the samples belonging to the last 5 days in the pre-processed unified load forecast data set are divided into the test set.

[0078] S130, training a plurality of different preset models according to the training set and the validation set to obtain a plurality of coordinated load forecasting models.

[0079] Exemplarily, the preset models can be BP neural network, LSTM neural network, bidirectional LSTM neural network, GRU neural network, bidirectional GRU neural network, CNN neural network, extreme learning machine (ELM), etc., which are optimized by genetic algorithms with outstanding performance.

[0080] Taking the optimization of BP neural network by genetic algorithm as an example, the process of training the BP neural network optimized by genetic algorithm according to the training set and the verification set can be specifically as follows:

[0081] (1) Initialize the genetic algorithm to optimize the parameters of the BP neural network.

[0082] (2) Obtain the parameter population and calculate the fitness.

[0083] (3) The genetic algorithm is used to select, crossover and mutate parameters to obtain the optimal parameters (weights, biases, etc.).

[0084] (4) Substitute the obtained optimal parameters into the genetic algorithm optimized BP neural network, and then input the training set and the verification set into the genetic algorithm optimized BP neural network to train it. Among them, the evaluation index of the model can be set according to different power scenarios, which will not be elaborated here.

[0085] S140, using multiple centralized load prediction models to perform centralized load prediction on each sample in the test set, and obtaining centralized load prediction results corresponding to the multiple centralized load prediction models.

[0086] Specifically, for any unified load prediction model, the test set is input into the unified load prediction model, and the unified load prediction model performs unified load prediction on each sample in the test set to obtain the unified load prediction result of each sample, which is then used as the unified load prediction result corresponding to the unified load prediction model.

[0087] S150, calculating the unified load prediction weights corresponding to the multiple unified load prediction models according to the unified load prediction results corresponding to the multiple unified load prediction models and the sample labels of the samples in the test set.

[0088] In some embodiments, the unified load prediction variances corresponding to the multiple unified load prediction models can be calculated based on the unified load prediction results corresponding to the multiple unified load prediction models and the sample labels of each sample in the test set, and the unified load prediction weights corresponding to the multiple unified load prediction models can be calculated based on the unified load prediction variances corresponding to the multiple unified load prediction models.

[0089] For example, for any centralized load forecasting model, the centralized load forecasting variance corresponding to the centralized load forecasting model may be calculated in the following manner:

[0090] According to the centralized load prediction result corresponding to the centralized load prediction model and the sample label of each sample in the test set, the centralized load prediction error of the centralized load prediction model at each sample in the test set is calculated.

[0091] According to the centralized load prediction error of the centralized load prediction model at each sample in the test set, an average centralized load prediction error is calculated.

[0092] According to the centralized load prediction error of the centralized load prediction model at each sample in the test set and the average centralized load prediction error, the centralized load prediction variance corresponding to the centralized load prediction model is calculated. Optionally, the formula used here can be as follows:

[0093]

[0094] in, is the unified load prediction variance corresponding to the unified load prediction model; is the load forecasting error of the load forecasting model i at the jth sample in the test set, n is the number of samples in the test set, is the average coordinated load forecast error.

[0095] Exemplarily, the centralized load prediction weights corresponding to the multiple centralized load prediction models may be calculated in the following manner:

[0096] The inverses of the unified load prediction variances corresponding to the multiple unified load prediction models are added together, and for any unified load prediction model, the inverse of the unified load prediction variance corresponding to the unified load prediction model is divided by the addition result to obtain the unified load prediction weight corresponding to the unified load prediction model. Optionally, the formula used here can be as follows:

[0097]

[0098] Among them, ω i is the unified load forecasting weight corresponding to unified load forecasting model i, is the unified load forecast variance corresponding to the unified load forecast model i, and x is the number of unified load forecast models.

[0099] S160, performing weighted integration on the multiple centralized load prediction models according to the centralized load prediction weights corresponding to the multiple centralized load prediction models to obtain a comprehensive centralized load prediction model.

[0100] In an embodiment of the present invention, a centralized load prediction data set is obtained, in which samples are based on regional historical meteorological data, historical electricity price data, and historical power station operation data, and regional historical centralized load data are used as sample labels; the centralized load prediction data set is divided into a training set, a validation set, and a test set; a plurality of different preset models are trained according to the training set and the validation set; the plurality of centralized load prediction models obtained through training are used to perform centralized load prediction on each sample in the test set; the centralized load prediction weights corresponding to the plurality of centralized load prediction models are calculated according to the centralized load prediction results corresponding to the plurality of centralized load prediction models and the sample labels of each sample in the test set, and the plurality of centralized load prediction models are weightedly integrated to obtain a comprehensive centralized load prediction model.

[0101] Based on the multi-model training method 100 provided in the embodiment of the present invention, the embodiment of the present invention also provides a method for predicting load based on multi-models, such as Figure 2 As shown, the coordinated load forecasting method 200 may include the following steps:

[0102] S210, obtaining regional weather forecast data, electricity price forecast data, and power station operation forecast data.

[0103] S220, respectively use multiple unified dispatching load prediction models in the comprehensive unified dispatching load prediction model to process the weather forecast data, electricity price forecast data, and power station operation forecast data to complete the unified dispatching load forecast and obtain unified dispatching load forecast results corresponding to the multiple unified dispatching load prediction models.

[0104] Specifically, for any unified load forecasting model, the weather forecast data, electricity price forecast data, and power station operation forecast data are input into the unified load forecasting model, and the unified load forecasting model processes the data to complete the unified load forecast and obtain the unified load forecast result, which is then used as the unified load forecast result corresponding to the unified load forecasting model.

[0105] S230, performing weighted summation on the unified dispatching load prediction results corresponding to the multiple unified dispatching load prediction models according to the unified dispatching load prediction weights corresponding to the multiple unified dispatching load prediction models, to obtain a comprehensive unified dispatching load prediction result.

[0106] Among them, the comprehensive coordinated load forecasting model is obtained based on the multi-model training method as described above.

[0107] It is worth noting that the formula used here can be shown as follows:

[0108]

[0109] in, is the comprehensive load forecast result, ωi is the unified load forecasting weight corresponding to unified load forecasting model i, is the unified load forecasting result corresponding to the unified load forecasting model i, and x is the number of unified load forecasting models.

[0110] In the embodiments of the present invention, the centralized load can be accurately predicted based on the comprehensive centralized load prediction model, so as to provide an effective reference for the formulation of power generation plans and the balance of power generation and consumption.

[0111] 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 optional embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0112] The above is an introduction to a method embodiment. The following is a further explanation of the solution of the present invention through an apparatus embodiment.

[0113] Figure 3 A structural diagram of a multi-model training device provided by an embodiment of the present invention is shown. Figure 3 As shown, the multi-model training device 300 may include:

[0114] The acquisition module 310 is used to obtain a centralized load forecasting data set, wherein the samples in the centralized load forecasting data set are based on the historical meteorological data, historical electricity price data, and historical power station operation data of the region, and the historical centralized load data of the region is used as the sample label.

[0115] The division module 320 is used to divide the centralized load forecasting data set into a training set, a validation set, and a test set.

[0116] The training module 330 is used to train multiple different preset models according to the training set and the verification set to obtain multiple coordinated load forecasting models.

[0117] The prediction module 340 is used to perform centralized load prediction on each sample in the test set using multiple centralized load prediction models to obtain centralized load prediction results corresponding to the multiple centralized load prediction models.

[0118] The calculation module 350 is used to calculate the unified load prediction weights corresponding to the multiple unified load prediction models according to the unified load prediction results corresponding to the multiple unified load prediction models and the sample labels of each sample in the test set.

[0119] The integration module 360 ​​is used to perform weighted integration on the multiple centralized load prediction models according to the centralized load prediction weights corresponding to the multiple centralized load prediction models to obtain a comprehensive centralized load prediction model.

[0120] Understandably, Figure 3 Each module / unit in the multi-model training device 300 shown has an implementation Figure 1 The functions of each step in the multi-model training method 100 shown can achieve its corresponding technical effects, and for the sake of brevity, they will not be repeated here.

[0121] Figure 4 FIG. 4 shows a structural diagram of a multi-model-based load forecasting device provided by an embodiment of the present invention. Figure 4 As shown, the centralized load forecasting device 400 may include:

[0122] The acquisition module 410 is used to acquire regional weather forecast data, electricity price forecast data, and power station operation forecast data.

[0123] The prediction module 420 is used to process the weather forecast data, electricity price forecast data, and power station forecast operation data using multiple unified load prediction models in the comprehensive unified load prediction model to complete the unified load prediction and obtain the unified load prediction results corresponding to the multiple unified load prediction models.

[0124] The summing module 430 is used to perform weighted summation of the centralized load prediction results corresponding to the multiple centralized load prediction models according to the centralized load prediction weights corresponding to the multiple centralized load prediction models to obtain a comprehensive centralized load prediction result.

[0125] Among them, the comprehensive coordinated load forecasting model is obtained based on the multi-model training method as described above.

[0126] Understandably, Figure 4 Each module / unit in the load forecasting device 400 shown in the figure has the function of realizing Figure 2 The functions of each step in the coordinated load forecasting method 200 shown and the ability to achieve the corresponding technical effects are not described in detail here for the sake of brevity.

[0127] Figure 5A block diagram of an exemplary electronic device capable of implementing an embodiment of the present invention is shown. Electronic device 500 is intended to represent 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 device 500 may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0128] like Figure 5 As shown, the electronic device 500 may include a computing unit 501, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic device 500 can also be stored. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0129] Multiple components in the electronic device 500 are connected to the I / O interface 505, including: an input unit 506, such as a keyboard, a mouse, etc.; an output unit 507, such as various types of displays, speakers, etc.; a storage unit 508, such as a disk, an optical disk, etc.; and a communication unit 509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 509 allows the electronic device 500 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0130] The computing unit 501 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 501 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 running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 501 performs the various methods and processes described above, such as method 100 or 200. For example, in some embodiments, the method 100 or 200 may be implemented as a computer program product, including a computer program, which is tangibly contained in a computer-readable medium, such as a storage unit 508. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 500 via the ROM 502 and / or the communication unit 509. When the computer program is loaded into the RAM 503 and executed by the computing unit 501, one or more steps of the method 100 or 200 described above may be performed. Alternatively, in other embodiments, the computing unit 501 may be configured to execute the method 100 or 200 in any other appropriate manner (eg, by means of firmware).

[0131] The various embodiments described above herein can 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 chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs, which may be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general programmable processor, which 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.

[0132] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0133] In the context of the present invention, a computer-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a computer-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0134] It should be noted that the present invention also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute method 100 or 200 and achieve the corresponding technical effects achieved by the embodiments of the present invention executing the methods thereof. For the sake of concise description, they will not be repeated here.

[0135] In addition, the present invention also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the method 100 or 200 is implemented.

[0136] To provide interaction with a user, the above-described embodiments may be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types 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 input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0137] The embodiments described above can 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 computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0138] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0139] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps 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, and this document does not limit this.

[0140] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-model training method for centralized load forecasting, characterized in that: The method comprises: Obtain a centralized load forecasting data set, wherein the samples in the centralized load forecasting data set are based on the historical meteorological data, historical electricity price data, and historical power station operation data of the region, and the historical centralized load data of the region is used as the sample label; Dividing the centralized load forecasting data set into a training set, a validation set, and a test set; Training a plurality of different preset models according to the training set and the verification set to obtain a plurality of coordinated load forecasting models; Using the multiple centralized load prediction models respectively to perform centralized load prediction on each sample in the test set, and obtaining centralized load prediction results corresponding to the multiple centralized load prediction models; Calculating the unified dispatching load prediction weights corresponding to the multiple unified dispatching load prediction models according to the unified dispatching load prediction results corresponding to the multiple unified dispatching load prediction models and the sample labels of the samples in the test set; According to the centralized load prediction weights corresponding to the multiple centralized load prediction models, the multiple centralized load prediction models are weighted integrated to obtain a comprehensive centralized load prediction model.

2. The method according to claim 1, characterized in that Obtaining the unified dispatching load forecast data set includes: Obtain regional meteorological data, electricity price data, power station operation data, and coordinated load data within a historical period; Samples are generated based on the acquired meteorological data, electricity price data, power station operation data and unified load data, and the unified load forecast data set is generated based on this.

3. The method according to claim 2, characterized in that Dividing the centralized load forecasting data set into a training set, a validation set, and a test set includes: Preprocessing the centralized load forecasting data set, wherein the preprocessing includes: outlier removal, missing value filling, and normalization processing; Dividing the historical time period into a first historical time period, a second historical time period, and a third historical time period according to a preset division ratio; Dividing the samples corresponding to the first historical time period in the preprocessed centralized load forecasting data set into a training set; Dividing the samples corresponding to the second historical time period in the preprocessed centralized load forecasting data set into a validation set; The samples corresponding to the third historical time period in the preprocessed coordinated load forecasting data set are divided into a test set.

4. The method according to claim 1, characterized in that: According to the centralized load prediction results corresponding to the multiple centralized load prediction models and the sample labels of the samples in the test set, calculating the centralized load prediction weights corresponding to the multiple centralized load prediction models includes: Calculating the unified load prediction variance corresponding to the multiple unified load prediction models according to the unified load prediction results corresponding to the multiple unified load prediction models and the sample label of each sample in the test set; The unified dispatching load prediction weights corresponding to the multiple unified dispatching load prediction models are calculated according to the unified dispatching load prediction variances corresponding to the multiple unified dispatching load prediction models.

5. The method according to claim 4, characterized in that According to the centralized load prediction results corresponding to the multiple centralized load prediction models and the sample labels of the samples in the test set, calculating the centralized load prediction variances corresponding to the multiple centralized load prediction models includes: For any centralized load forecasting model, the centralized load forecasting variance corresponding to the centralized load forecasting model is calculated in the following manner: Calculate the unified load prediction error of the unified load prediction model at each sample in the test set according to the unified load prediction result corresponding to the unified load prediction model and the sample label of each sample in the test set; Calculate an average centralized load prediction error according to the centralized load prediction error of the centralized load prediction model at each sample in the test set; The centralized load prediction variance corresponding to the centralized load prediction model is calculated according to the centralized load prediction error of the centralized load prediction model at each sample in the test set and the average centralized load prediction error.

6. The method according to claim 5, characterized in that Calculating the unified load prediction variance corresponding to the unified load prediction model according to the unified load prediction error of the unified load prediction model at each sample in the test set and the average unified load prediction error comprises: According to the centralized load prediction error of the centralized load prediction model at each sample in the test set and the average centralized load prediction error, the centralized load prediction variance corresponding to the centralized load prediction model is calculated using the following formula: in, is the unified load prediction variance corresponding to the unified load prediction model; is the load forecasting error of the load forecasting model i at the jth sample in the test set, n is the number of samples in the test set, is the average coordinated load forecast error.

7. The method according to claim 5, characterized in that Calculating the unified dispatching load prediction weights corresponding to the multiple unified dispatching load prediction models according to the unified dispatching load prediction variances corresponding to the multiple unified dispatching load prediction models includes: Adding the inverses of the centralized load prediction variances corresponding to the multiple centralized load prediction models; For any centralized load prediction model, the inverse of the centralized load prediction variance corresponding to the centralized load prediction model is divided by the addition result to obtain the centralized load prediction weight corresponding to the centralized load prediction model.

8. A method for predicting loads based on multiple models, characterized in that: The method comprises: Obtain regional weather forecast data, electricity price forecast data, and power station operation forecast data; Respectively use multiple coordinated load forecasting models in the comprehensive coordinated load forecasting model to process the weather forecast data, the electricity price forecast data, and the power station operation forecast data to complete the coordinated load forecasting and obtain coordinated load forecasting results corresponding to the multiple coordinated load forecasting models; According to the centralized load prediction weights corresponding to the multiple centralized load prediction models, weighted summation is performed on the centralized load prediction results corresponding to the multiple centralized load prediction models to obtain a comprehensive centralized load prediction result; Wherein, the comprehensive coordinated load forecasting model is obtained based on the multi-model training method described in any one of claims 1-7.

9. A multi-model training device for centralized load forecasting, characterized in that: The device comprises: An acquisition module is used to acquire a centralized load forecasting data set, wherein the samples in the centralized load forecasting data set are based on the historical meteorological data, historical electricity price data, and historical power station operation data of the region, and the historical centralized load data of the region is used as the sample label; A partitioning module, used to divide the centralized load forecasting data set into a training set, a validation set, and a test set; A training module, used for training a plurality of different preset models according to the training set and the verification set to obtain a plurality of coordinated load forecasting models; A prediction module, used to perform a centralized load prediction on each sample in the test set using the multiple centralized load prediction models respectively, and obtain the centralized load prediction results corresponding to the multiple centralized load prediction models; A calculation module, used to calculate the unified dispatching load prediction weights corresponding to the multiple unified dispatching load prediction models according to the unified dispatching load prediction results corresponding to the multiple unified dispatching load prediction models and the sample labels of each sample in the test set; The integration module is used to perform weighted integration on the multiple centralized load prediction models according to the centralized load prediction weights corresponding to the multiple centralized load prediction models to obtain a comprehensive centralized load prediction model.

10. A multi-model-based load forecasting device, characterized in that: The device comprises: The acquisition module is used to obtain regional weather forecast data, electricity price forecast data, and power station operation forecast data; A prediction module, used to process the weather forecast data, the electricity price forecast data, and the power station operation forecast data using multiple unified load forecast models in the comprehensive unified load forecast model to complete unified load forecasting and obtain unified load forecasting results corresponding to the multiple unified load forecasting models; A summing module, configured to perform weighted summation of the centralized load prediction results corresponding to the multiple centralized load prediction models according to the centralized load prediction weights corresponding to the multiple centralized load prediction models, to obtain a comprehensive centralized load prediction result; Wherein, the comprehensive coordinated load forecasting model is obtained based on the multi-model training method described in any one of claims 1-7.

Images