Optimization method, device and electronic equipment for comprehensive energy scheduling plan

Abstract

The present application proposes an optimization method, device and electronic equipment for an integrated energy scheduling plan, including: obtaining historical data samples corresponding to integrated energy, and an integrated energy scheduling plan prediction model trained by the historical data samples, inputting multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data in the historical training samples into the integrated energy scheduling plan prediction model to obtain multiple predicted scheduling plans, and combining the multiple historical scheduling plans in the historical training samples to construct an optimization function to optimize the model to obtain an optimized target integrated energy scheduling plan prediction model, thereby predicting the predicted photovoltaic power data, predicted wind power data and predicted thermal power data to be predicted to obtain a predicted optimized scheduling plan, thereby optimizing the integrated energy scheduling plan by combining the historical data samples, and improving the effectiveness and reliability of the integrated energy scheduling plan.

Description

Technical Field

[0001] The present application relates to the field of energy scheduling technology, and in particular to an optimization method, device, electronic device and storage medium for a comprehensive energy scheduling plan. Background Art

[0002] At present, the traditional integrated energy base scheduling plan only considers information such as power forecast data and equipment status data in the preset time period. According to the power forecast results, the output range of wind farms, photovoltaic power stations, and thermal power plants is determined as a constraint condition, and a wind, solar, and thermal power output allocation model is established. By setting an optimization objective function (such as an economic indicator), an integrated energy scheduling plan allocation scheme is formulated for the wind, solar, and thermal power ratios. However, due to the uncertainty of wind and solar power forecasts, the final output ratio often deviates from the plan formulated in the preset time period. The energy base scheduling plan in the relevant technology only considers information such as power forecast data and equipment status data in the preset time period, and does not utilize the information contained in a large amount of historical data. Therefore, it will affect the effectiveness and reliability of the preset scheduling plan. Summary of the Invention

[0003] The first aspect of the present application provides an optimization method for an integrated energy scheduling plan, the method comprising: obtaining historical data samples corresponding to the integrated energy, and an integrated energy scheduling plan prediction model trained by the historical data samples, wherein the historical data samples include multiple historical photovoltaic power data, multiple historical wind power data, multiple historical thermal power data and multiple historical scheduling plans of multiple same sampling periods; inputting the multiple historical photovoltaic power data, multiple historical wind power data and multiple historical thermal power data into the integrated energy scheduling plan prediction model to obtain multiple forecast scheduling plans corresponding to the integrated energy; constructing an optimization function for optimizing the integrated energy scheduling plan prediction model based on the multiple forecast scheduling plans and the multiple historical scheduling plans; optimizing the integrated energy scheduling plan prediction model according to the optimization function to obtain an optimized target integrated energy scheduling plan prediction model; inputting the predicted photovoltaic power data, predicted wind power data and predicted thermal power data to be predicted into the target integrated energy scheduling plan prediction model to obtain a predicted optimized scheduling plan.

[0004] In one embodiment of the present application, the obtaining of historical data samples corresponding to the comprehensive energy, and the comprehensive energy scheduling plan prediction model trained by the historical data samples, include: obtaining historical data samples corresponding to the comprehensive energy, and calculating multiple prediction scheduling plans corresponding to the comprehensive energy based on the multiple historical photovoltaic power data, multiple historical wind power data and multiple historical thermal power data in the historical data samples; using the multiple historical photovoltaic power data, multiple historical wind power data and multiple historical thermal power data as inputs of the LCA model, and using the multiple prediction scheduling plans as outputs of the LCA model for model training to obtain the comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy.

[0005] In one embodiment of the present application, the LCA model includes an input layer, a convolutional neural network CNN layer, a long short-term memory artificial neural network LSTM layer, an attention layer and an output layer, and the multiple historical photovoltaic power data, the multiple historical wind power data and the multiple historical thermal power data are used as inputs of the LCA model, and the multiple forecast scheduling plans are used as outputs of the LCA model for model training to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy, including: inputting the multiple historical photovoltaic power data, the multiple historical wind power data and the multiple historical thermal power data as training samples into the input layer, and obtaining input vectors corresponding to each of the training samples through the input layer; inputting the input vectors into the CNN layer, and performing feature extraction on the input vectors to screen Select a target feature vector; obtain the target feature vector, input the target feature vector into the LSTM layer to obtain a first output vector corresponding to each target feature vector; input the first output vector corresponding to each target feature vector into the attention layer, and filter the first output vector according to the attention weight parameter value of the first output vector in the attention layer to obtain a second output vector; input the second output vector into the output layer to determine multiple training scheduling plans corresponding to the comprehensive energy, and output multiple training scheduling plans corresponding to the comprehensive energy; adjust the model parameters of the LCA model according to the differences between the multiple training scheduling plans and the multiple prediction scheduling plans to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy.

[0006] In one embodiment of the present application, the CNN layer includes a convolutional layer and a dropout layer, the input vector is input into the CNN layer, and feature extraction is performed on the input vector to filter out a target feature vector, including: inputting the input vector into the convolutional layer to obtain multiple feature vectors of the input vector extracted by the convolutional layer; inputting the multiple feature vectors of the input vector into the dropout layer to filter out the target feature vector from the multiple features.

[0007] In one embodiment of the present application, after inputting the second output vector into the output layer to determine multiple training scheduling plans corresponding to the comprehensive energy, and outputting multiple training scheduling plans corresponding to the comprehensive energy, it also includes: obtaining multiple training scheduling plans corresponding to the comprehensive energy, and performing denormalization processing on the multiple training scheduling plans.

[0008] The present application proposes a method for optimizing a comprehensive energy scheduling plan, which obtains historical data samples corresponding to comprehensive energy and a comprehensive energy scheduling plan prediction model trained by the historical data samples, and inputs multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data in the historical training samples into the comprehensive energy scheduling plan prediction model to obtain multiple predicted scheduling plans, and combines the multiple historical scheduling plans in the historical training samples to construct an optimization function to optimize the model to obtain an optimized target comprehensive energy scheduling plan prediction model, thereby predicting the predicted photovoltaic power data, predicted wind power data, and predicted thermal power data to be predicted to obtain a predicted optimized scheduling plan. Thus, the optimization of the comprehensive energy scheduling plan is achieved by combining the historical data samples, thereby improving the effectiveness and reliability of the comprehensive energy scheduling plan.

[0009] The second aspect of the present application proposes an optimization device for an integrated energy scheduling plan, the device comprising: an acquisition module for acquiring historical data samples corresponding to the integrated energy, and an integrated energy scheduling plan prediction model trained by the historical data samples, wherein the historical data samples include multiple historical photovoltaic power data, multiple historical wind power data, multiple historical thermal power data and multiple historical scheduling plans of multiple same sampling periods; a generation module for inputting the multiple historical photovoltaic power data, multiple historical wind power data and multiple historical thermal power data into the integrated energy scheduling plan prediction model to obtain multiple prediction scheduling plans corresponding to the integrated energy; a construction module for constructing an optimization function for optimizing the integrated energy scheduling plan prediction model based on the multiple prediction scheduling plans and the multiple historical scheduling plans; an optimization module for optimizing the integrated energy scheduling plan prediction model according to the optimization function to obtain an optimized target integrated energy scheduling plan prediction model; a prediction module for inputting the predicted photovoltaic power data, predicted wind power data and predicted thermal power data to be predicted into the target integrated energy scheduling plan prediction model to obtain a predicted optimized scheduling plan.

[0010] In one embodiment of the present application, the acquisition module includes: an acquisition unit, used to acquire historical data samples corresponding to the comprehensive energy, and calculate multiple forecast scheduling plans corresponding to the comprehensive energy based on the multiple historical photovoltaic power data, multiple historical wind power data and multiple historical thermal power data in the historical data samples; a training unit, used to use the multiple historical photovoltaic power data, multiple historical wind power data and multiple historical thermal power data as inputs of the LCA model, and use the multiple forecast scheduling plans as outputs of the LCA model for model training, so as to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy.

[0011] In one embodiment of the present application, the LCA model includes an input layer, a convolutional neural network CNN layer, a long short-term memory artificial neural network LSTM layer, an attention layer and an output layer, and the training unit includes: an input subunit, which is used to input the multiple historical photovoltaic power data, the multiple historical wind power data and the multiple historical thermal power data as training samples into the input layer, and obtain the input vectors corresponding to the training samples through the input layer; an extraction subunit, which is used to input the input vector into the CNN layer and perform feature extraction on the input vector to screen out the target feature vector; a first generation subunit, which is used to obtain the target feature vector and input the target feature vector into the LSTM layer to obtain the target feature vector. a first output vector corresponding to each vector; a second generating subunit, used to input the first output vector corresponding to each target feature vector into the attention layer, and filter the first output vector according to the attention weight parameter value of the first output vector in the attention layer to obtain a second output vector; an output subunit, used to input the second output vector into the output layer to determine a plurality of training scheduling plans corresponding to the comprehensive energy, and output a plurality of training scheduling plans corresponding to the comprehensive energy; a third generating subunit, used to adjust the model parameters of the LCA model according to the differences between the plurality of training scheduling plans and the plurality of prediction scheduling plans to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy.

[0012] In one embodiment of the present application, the CNN layer includes a convolutional layer and a dropout layer, and the extraction unit is specifically used to: input the input vector into the convolutional layer to obtain multiple feature vectors of the input vector extracted by the convolutional layer; input the multiple feature vectors of the input vector into the dropout layer to filter out the target feature vector from the multiple features.

[0013] In one embodiment of the present application, the training unit further includes: a processing subunit, configured to obtain a plurality of training scheduling plans corresponding to the comprehensive energy, and perform denormalization processing on the plurality of training scheduling plans.

[0014] The present application proposes an optimization device for a comprehensive energy scheduling plan, which obtains historical data samples corresponding to comprehensive energy and a comprehensive energy scheduling plan prediction model trained by the historical data samples, and inputs multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data in the historical training samples into the comprehensive energy scheduling plan prediction model to obtain multiple prediction scheduling plans, and combines the multiple historical scheduling plans in the historical training samples to construct an optimization function to optimize the model to obtain an optimized target comprehensive energy scheduling plan prediction model, thereby predicting the predicted photovoltaic power data, predicted wind power data, and predicted thermal power data to be predicted to obtain a predicted optimized scheduling plan. Therefore, the optimization of the comprehensive energy scheduling plan is achieved by combining the historical data samples, and the effectiveness and reliability of the comprehensive energy scheduling plan are improved.

[0015] The third aspect of the present application proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, the optimization method of the integrated energy scheduling plan in the embodiment of the present application is implemented.

[0016] The fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, an optimization method for the integrated energy scheduling plan in the embodiment of the present application is implemented.

[0017] Other effects of the above optional manner will be described below in conjunction with specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of a method for optimizing a comprehensive energy scheduling plan provided in an embodiment of the present application;

[0019] Figure 2 is a graph of photovoltaic power data provided by an embodiment of the present application;

[0020] Figure 3 is a curve chart of wind power data provided in an embodiment of the present application;

[0021] Figure 4 This is a comparison chart of a predicted scheduling plan and a historical scheduling plan provided in an embodiment of the present application;

[0022] Figure 5 This is a flow chart of another method for optimizing a comprehensive energy scheduling plan provided in an embodiment of the present application;

[0023] Figure 6 This is a schematic diagram of the structure of an optimization device for a comprehensive energy scheduling plan provided in an embodiment of the present application;

[0024] Figure 7 This is a schematic structural diagram of another device for optimizing a comprehensive energy scheduling plan provided in an embodiment of the present application;

[0025] Figure 8 It is a block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] The following describes in detail embodiments of the present application. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0027] The following describes the optimization method, device and electronic device of the comprehensive energy scheduling plan of the embodiment of the present application with reference to the accompanying drawings.

[0028] Figure 1 The flowchart of the method for optimizing a comprehensive energy scheduling plan provided in an embodiment of the present application is shown in FIG. It should be noted that the execution subject of the method for optimizing a comprehensive energy scheduling plan provided in this embodiment is an optimization device for a comprehensive energy scheduling plan, which can be implemented by software and / or hardware. The optimization device for a comprehensive energy scheduling plan in this embodiment can be configured in an electronic device, which can include a server. This embodiment does not specifically limit the electronic device.

[0029] Figure 1 It is a flow chart of an optimization method for a comprehensive energy scheduling plan provided in an embodiment of the present application.

[0030] like Figure 1 As shown, the optimization method of the integrated energy scheduling plan may include:

[0031] Step 101: Obtain historical data samples corresponding to comprehensive energy, and a comprehensive energy scheduling plan prediction model trained by the historical data samples, wherein the historical data samples include multiple historical photovoltaic power data, multiple historical wind power data, multiple historical thermal power data, and multiple historical scheduling plans for multiple identical sampling periods.

[0032] In some embodiments, historical data samples corresponding to the integrated energy can be obtained from a database of the integrated energy base, but is not limited thereto.

[0033] Among them, the historical scheduling plan can be the output ratio of photovoltaic power, wind power and thermal power.

[0034] In some embodiments, after obtaining historical data samples corresponding to comprehensive energy, the output ratios of photovoltaic power, wind power and thermal power can be calculated through multiple historical photovoltaic power data, multiple historical wind power data and multiple historical thermal power data, thereby training a comprehensive energy scheduling plan prediction model.

[0035] Optionally, multiple identical sampling periods may be 1 hour, but are not limited thereto. Specifically, when the historical data sample is sampling data within one day and the sampling period is 1 hour apart, the multiple historical photovoltaic power data are The calculation method of the curve graph of multiple historical photovoltaic power data is as follows: The curve diagrams of multiple historical photovoltaic power data are as follows: Figure 2 As shown, n is equal to the number of sampled samples, and m is the amount of sampled data within 1 day.

[0036] On the other hand, when the historical data samples are sampled data within one day and the sampling period is 1 hour apart, multiple historical wind power data are The calculation method of the curve graph of multiple historical wind power data is as follows: The curve diagrams of multiple historical wind power data are as follows: Figure 3 As shown, n is equal to the number of sampled samples, and m is the amount of sampled data within 1 day.

[0037] In addition, similar to the above-mentioned multiple historical photovoltaic power data and multiple historical wind power data, when the historical data samples are sampled within one day and the sampling period is 1 hour apart, the multiple historical thermal power data are The calculation method of the curve graph of multiple historical thermal power data is as follows: Among them, n is equal to the number of sampling samples, and m is the amount of sampling data within 1 day.

[0038] Among them, if the interval between the sampling periods of the above-mentioned multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data is 15 minutes, then m is 96. The interval between the sampling periods can be determined by technical personnel based on actual business scenarios, and this embodiment does not make specific limitations on this.

[0039] Step 102 : Input multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data into a comprehensive energy scheduling plan prediction model to obtain multiple prediction scheduling plans corresponding to comprehensive energy.

[0040] In some embodiments, the historical photovoltaic power data Historical wind power data Historical thermal power data For 1 day, the forecast scheduling plan corresponding to comprehensive energy That is, the output ratio of photovoltaic power, wind power and thermal power on that day. Specifically, the forecast scheduling plan corresponding to the comprehensive energy The calculation formula can be:

[0041]

[0042] in, are the output ratios of photovoltaic power, wind power and thermal power on that day, respectively. The calculation formula can be:

[0043]

[0044]

[0045]

[0046] Among them, R' Pij , R' Wij , R' Tij are the outputs of photovoltaic power, wind power and thermal power respectively on that day, which are obtained by comprehensive analysis of DC transmission capacity, upper and lower limits of thermal power output, photovoltaic power and wind power, but are not limited to this.

[0047] In addition, optionally, multiple days of historical photovoltaic power data, multiple days of historical wind power data, and multiple days of historical thermal power data can be selected and input into the comprehensive energy scheduling plan prediction model to obtain multiple prediction scheduling plans corresponding to the comprehensive energy, thereby ensuring the accuracy of the data and the reliability of the optimization method of the comprehensive energy scheduling plan.

[0048] Step 103: construct an optimization function for optimizing the comprehensive energy scheduling plan prediction model based on the multiple prediction scheduling plans and the multiple historical scheduling plans.

[0049] In some embodiments, the historical photovoltaic power data Historical wind power data Historical thermal power data In the case of 1 day, calculate the forecast scheduling plan R for a single day i , and obtain the corresponding historical scheduling plan R for the day from the historical data sample i , where the historical scheduling plan R i The calculation method is:

[0050]

[0051] Among them, R Pij 、R Wij 、RTij They are the historical output proportions of photovoltaic power, wind power and thermal power on that day, among which the forecast dispatch plan for a single day And the historical data samples are used to obtain the historical scheduling plan R corresponding to the day i A comparison chart, such as Figure 4 shown.

[0052] Therefore, through multiple forecast scheduling plans based on different days and multiple historical schedules R i , construct an optimization function for optimizing the comprehensive energy scheduling plan prediction model. Specifically, the optimization function can be constructed in the following way:

[0053]

[0054] Where n is or R i The total number of samples.

[0055] Step 104 : Optimizing the integrated energy scheduling plan prediction model according to the optimization function to obtain an optimized target integrated energy scheduling plan prediction model.

[0056] In some embodiments, the minimum value of the optimization function can be used as the target or R i Optimization is performed, and then the comprehensive energy scheduling plan prediction model is optimized, so as to obtain the target comprehensive energy scheduling plan prediction model corresponding to the minimum value of the optimization function.

[0057] Step 105 : inputting the predicted photovoltaic power data, the predicted wind power data, and the predicted thermal power data to be predicted into the target comprehensive energy scheduling plan prediction model to obtain a predicted optimized scheduling plan.

[0058] In some embodiments, the integrated energy scheduling plan prediction model is optimized by combining historical data samples, so that the predicted photovoltaic power data, predicted wind power data and predicted thermal power data are processed according to the optimized target integrated energy scheduling plan prediction model to obtain an optimized scheduling plan with prediction, thereby improving the rationality and effectiveness of the optimized scheduling plan, and further improving the stability and safety of the operation of the integrated energy base.

[0059] The present application proposes a method for optimizing a comprehensive energy scheduling plan, which obtains historical data samples corresponding to comprehensive energy and a comprehensive energy scheduling plan prediction model trained by the historical data samples, and inputs multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data in the historical training samples into the comprehensive energy scheduling plan prediction model to obtain multiple predicted scheduling plans, and combines the multiple historical scheduling plans in the historical training samples to construct an optimization function to optimize the model to obtain an optimized target comprehensive energy scheduling plan prediction model, thereby predicting the predicted photovoltaic power data, predicted wind power data, and predicted thermal power data to be predicted to obtain a predicted optimized scheduling plan. Thus, the optimization of the comprehensive energy scheduling plan is achieved by combining the historical data samples, thereby improving the effectiveness and reliability of the comprehensive energy scheduling plan.

[0060] Figure 5 This is a flow chart of another method for optimizing a comprehensive energy scheduling plan provided by an embodiment of the present application. The method may include:

[0061] Step 501 , obtain historical data samples corresponding to comprehensive energy, and calculate multiple forecast scheduling plans corresponding to comprehensive energy based on multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data in the historical data samples.

[0062] One implementation method of obtaining historical data samples corresponding to comprehensive energy and calculating multiple forecast scheduling plans corresponding to comprehensive energy based on multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data in the historical data samples can be to obtain the respective outputs of photovoltaic power, wind power, and thermal power through comprehensive analysis of DC transmission capacity, upper and lower limits of thermal power output, photovoltaic power, and wind power, and then calculate the respective ratios of photovoltaic power, wind power, and thermal power, thereby generating a forecast scheduling plan corresponding to comprehensive energy based on the respective ratios of photovoltaic power, wind power, and thermal power.

[0063] In step 502, a plurality of historical photovoltaic power data, a plurality of historical wind power data, and a plurality of historical thermal power data are used as inputs of the LCA model, and a plurality of forecast scheduling plans are used as outputs of the LCA model for model training to obtain a comprehensive energy scheduling plan forecasting model corresponding to comprehensive energy.

[0064] Among them, it should be noted that, in this embodiment, a hybrid (LSTM+CNN+attention, LCA) model is described as an example, including an input layer, a convolutional neural network (CNN) layer, a long short-term memory artificial neural network (LSTM) layer, an attention layer and an output layer. Specifically, multiple historical photovoltaic power data, multiple historical wind power data and multiple historical thermal power data are used as inputs of the LCA model, and multiple forecast scheduling plans are used as outputs of the LCA model for model training to obtain a comprehensive energy scheduling plan prediction model corresponding to comprehensive energy. The implementation method can be to input multiple historical photovoltaic power data, multiple historical wind power data and multiple historical thermal power data as training samples into the input layer, obtain input vectors corresponding to each training sample through the input layer, input the input vectors into the CNN layer, and perform feature extraction on the input vectors to screen out Target feature vector, obtain the target feature vector, input the target feature vector into the LSTM layer to obtain the first output vector corresponding to each target feature vector, input the first output vector corresponding to each target feature vector into the attention layer, and filter the first output vector according to the attention weight parameter value of the first output vector in the attention layer to obtain a second output vector, input the second output vector into the output layer to determine multiple training scheduling plans corresponding to the comprehensive energy, and output multiple training scheduling plans corresponding to the comprehensive energy, and adjust the model parameters of the LCA model according to the differences between the multiple training scheduling plans and the multiple prediction scheduling plans to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy.

[0065] Specifically, after obtaining a training sample consisting of multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data, the training sample is input into the LCA model through the input layer to obtain an input vector converted from the training sample. For example, if the length of the batch input training sample is m, the input vector is

[0066] In some embodiments, in order to accurately filter out the target feature vector, the CNN layer includes a convolution layer and a dropout layer. The input vector is input into the CNN layer, and features are extracted from the input vector to filter out the target feature vector. One implementation method is: input the input vector into the convolution layer, obtain multiple feature vectors of the input vector extracted by the convolution layer, and input the multiple feature vectors of the input vector into the dropout layer to filter out the target feature vector from multiple features.

[0067] In some exemplary embodiments, when the training sample dimension is 1-dimensional, the convolution layer is selected as a one-dimensional convolution, then the size of the convolution kernel is 3, and the RELU activation function is used in combination to obtain multiple feature vectors of the input vector.

[0068] In some exemplary embodiments, with a dropout layer set to 0.2, half of the hidden neurons in the dropout layer network can be temporarily and randomly deleted, while the input and output neurons remain unchanged. The input neurons are then forward propagated through the modified network, and the resulting loss is backpropagated through the modified network. After this process is completed for a small batch of training samples, the parameters (w, b) corresponding to the neurons that were not deleted are updated using stochastic gradient descent.

[0069] It is understandable that after obtaining the updated corresponding parameters (w, b), in order to avoid the problem of overfitting of the trained prediction model, in some embodiments, this process can be repeated continuously: the deleted neurons are restored, wherein the deleted neurons remain the same, and the neurons that have not been deleted have been updated, and a subset of half the size is randomly selected from the hidden neurons and temporarily deleted, and the parameters of the deleted neurons are backed up. For a small batch of training samples, the loss is first forward propagated and then back-propagated, and the parameters (w, b) are updated according to the stochastic gradient descent method to solve the problem of overfitting of different networks.

[0070] Among them, w is the parameter weight in the neural network, and b is the bias in the neural network.

[0071] In other embodiments, when the dropout layer is set to 0.2, if the number of neurons is n, 0.2n neurons may be deleted. One implementation of deleting neurons is to change the activation function value of the neuron in the network to 0 with probability p. If the length of the output vector is i, then the target feature vector is H c =[h c1 …h c1 …] T , where the dropout neurons calculate the activation function value in the network One way to calculate is:

[0072]

[0073]

[0074]

[0075]

[0076] Among them, the Bernoulli function generates a probability vector r, that is, a vector of 0 and 1 is randomly generated.

[0077] In some embodiments, the obtained target feature vector is input into the LSTM layer, and the behavioral characteristics of the training sample are learned through the LSTM layer and the bidirectional long short-term memory artificial neural network (biLSTM) layer structure. If the length of the first output vector is j, the first output vector of the LSTM layer is H L =[h L1 …h Lε …h L ] J , calculate H L One way to calculate is:

[0078]

[0079] H l =max(dropout(L))+b r

[0080] Among them, the LSTM layer needs to be connected to the dropout layer and the maxpooling layer, Max is the maximum value function in the maxpooling layer, br is the bias of the pooling layer, L is the output of the LSTM layer, W C ,b c are the weight and bias of the LSTM layer respectively.

[0081] In some embodiments, after the first output vectors corresponding to the target feature vectors are input into the attention layer, the attention weight parameter values of the first output vectors are distributed according to the weight distribution principle in the attention layer to obtain the attention weight parameter values of the first output vectors, and the first output vectors are screened according to the attention weight parameter values of the first output vectors to obtain the second output vector.

[0082] In some exemplary embodiments, if the length of the second output vector is k, the second output vector S' is

[0083] In some embodiments, after the second output vector is input to the output layer, the output layer obtains multiple training scheduling plans corresponding to the comprehensive energy through the fully connected layer. Assuming that the compensation predicted by the output layer is n, the multiple training scheduling plans Y corresponding to the comprehensive energy are An exemplary method for calculating Y is:

[0084] Y=f(W r ·s+b r )

[0085] Among them, W r is the output layer weight, b r is the output layer bias, and f is the activation function of the fully connected layer.

[0086] In other embodiments, after obtaining multiple training scheduling plans corresponding to the comprehensive energy, in order to accurately train the model, in some embodiments, the multiple training scheduling plans corresponding to the comprehensive energy can also be denormalized to obtain multiple training scheduling plans in a standard format after denormalization.

[0087] An exemplary method of denormalization is as follows:

[0088]

[0089] in, are multiple training scheduling plans corresponding to the comprehensive energy before denormalization predicted by the LCA model, y* are multiple training scheduling plans in the standard format after denormalization, and y min 、y max are the minimum and maximum output values of photovoltaic power, wind power and thermal power in the historical output data before normalization.

[0090] Step 503 : Input multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data into the comprehensive energy scheduling plan prediction model to obtain multiple prediction scheduling plans corresponding to the comprehensive energy.

[0091] Step 504 : constructing an optimization function for optimizing the comprehensive energy scheduling plan prediction model based on the multiple prediction scheduling plans and the multiple historical scheduling plans.

[0092] Step 505 : Optimizing the integrated energy scheduling plan prediction model according to the optimization function to obtain an optimized target integrated energy scheduling plan prediction model.

[0093] Step 506 : Input the predicted photovoltaic power data, predicted wind power data, and predicted thermal power data to be predicted into the target comprehensive energy scheduling plan prediction model to obtain a predicted optimized scheduling plan.

[0094] It should be noted that, for the specific implementation of steps 503 to 506, reference may be made to the relevant description in the above embodiment.

[0095] The present application proposes a method for optimizing a comprehensive energy scheduling plan. The method obtains historical data samples corresponding to the comprehensive energy, and calculates multiple forecast scheduling plans corresponding to the comprehensive energy based on multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data in the historical data samples. The multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data are used as inputs of an LCA model, and the multiple forecast scheduling plans are used as outputs of the LCA model for model training to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy. The multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data in the historical training samples are input into the comprehensive energy scheduling plan prediction model to obtain multiple forecast scheduling plans. An optimization function is constructed to optimize the model in combination with the multiple historical scheduling plans in the historical training samples to obtain an optimized target comprehensive energy scheduling plan prediction model, thereby predicting the predicted photovoltaic power data, predicted wind power data, and predicted thermal power data to be predicted to obtain a predicted optimized scheduling plan. Thus, the optimization of the comprehensive energy scheduling plan is achieved by combining the historical data samples, the availability of the historical data samples is improved, and the effective optimization of the comprehensive energy scheduling plan is achieved.

[0096] Figure 6 It is a structural diagram of an optimization device for a comprehensive energy scheduling plan provided in an embodiment of the present application.

[0097] like Figure 6 As shown, the optimization device 600 for the integrated energy scheduling plan includes an acquisition module 601, a generation module 602, a construction module 603, an optimization module 604 and a prediction module 605, wherein:

[0098] Acquisition module 601 is used to obtain historical data samples corresponding to the comprehensive energy, and a comprehensive energy scheduling plan prediction model trained by the historical data samples, wherein the historical data samples include multiple historical photovoltaic power data, multiple historical wind power data, multiple historical thermal power data and multiple historical scheduling plans for multiple identical sampling periods.

[0099] The generation module 602 is used to input the multiple historical photovoltaic power data, the multiple historical wind power data, and the multiple historical thermal power data into the comprehensive energy scheduling plan prediction model to obtain multiple prediction scheduling plans corresponding to the comprehensive energy.

[0100] The construction module 603 is used to construct an optimization function for optimizing the comprehensive energy scheduling plan prediction model based on the multiple prediction scheduling plans and the multiple historical scheduling plans.

[0101] The optimization module 604 is configured to optimize the integrated energy scheduling plan prediction model according to the optimization function to obtain an optimized target integrated energy scheduling plan prediction model.

[0102] The prediction module 605 is used to input the predicted photovoltaic power data, predicted wind power data and predicted thermal power data to be predicted into the target comprehensive energy scheduling plan prediction model to obtain a predicted optimized scheduling plan.

[0103] The present application proposes an optimization device for a comprehensive energy scheduling plan, which obtains historical data samples corresponding to comprehensive energy and a comprehensive energy scheduling plan prediction model trained by the historical data samples, and inputs multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data in the historical training samples into the comprehensive energy scheduling plan prediction model to obtain multiple prediction scheduling plans, and combines the multiple historical scheduling plans in the historical training samples to construct an optimization function to optimize the model to obtain an optimized target comprehensive energy scheduling plan prediction model, thereby predicting the predicted photovoltaic power data, predicted wind power data, and predicted thermal power data to be predicted to obtain a predicted optimized scheduling plan. Therefore, the optimization of the comprehensive energy scheduling plan is achieved by combining the historical data samples, and the effectiveness and reliability of the comprehensive energy scheduling plan are improved.

[0104] In one embodiment of the present application, Figure 7 This is a schematic diagram of the structure of another device for optimizing the integrated energy scheduling plan provided in an embodiment of the present application. Figure 7 As shown, the optimization device 700 for the comprehensive energy scheduling plan can also include: an acquisition module 701, a generation module 702, a construction module 703, an optimization module 704 and a prediction module 705, wherein the acquisition module 701 includes an acquisition unit 7011 and a training unit 7012, and the training unit 7012 includes an input subunit 70121, an extraction subunit 70122, a first generation subunit 70123, a second generation subunit 70124, an output subunit 70125, a third generation subunit 70126 and a processing subunit 70127.

[0105] For detailed descriptions of the acquisition module 701, the generation module 702, the construction module 703, the optimization module 704, and the prediction module 705, please refer to Figure 6 The description of the acquisition module 601, the generation module 602, the construction module 603, the optimization module 604 and the prediction module 605 in the illustrated embodiment will not be described here.

[0106] In one embodiment of the present application, Figure 7 The acquisition module 701 includes:

[0107] The acquisition unit 7011 is used to obtain historical data samples corresponding to the comprehensive energy, and calculate multiple forecast scheduling plans corresponding to the comprehensive energy based on the multiple historical photovoltaic power data, multiple historical wind power data and multiple historical thermal power data in the historical data samples.

[0108] The training unit 7012 is used to use the multiple historical photovoltaic power data, the multiple historical wind power data and the multiple historical thermal power data as the input of the LCA model, and use the multiple prediction scheduling plans as the output of the LCA model for model training to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy.

[0109] In one embodiment of the present application, Figure 7 The LCA model includes an input layer, a convolutional neural network (CNN) layer, a long short-term memory (LSTM) artificial neural network (ANN) layer, an attention layer, and an output layer. The training unit 7012 includes:

[0110] The input subunit 70121 is used to input the multiple historical photovoltaic power data, the multiple historical wind power data and the multiple historical thermal power data as training samples into the input layer, and obtain the input vectors corresponding to the training samples through the input layer.

[0111] The extraction subunit 70122 is used to input the input vector into the CNN layer and perform feature extraction on the input vector to filter out the target feature vector.

[0112] The first generating subunit 70123 is used to obtain the target feature vectors and input the target feature vectors into the LSTM layer to obtain the first output vectors corresponding to the target feature vectors.

[0113] The second generating subunit 70124 is used to input the first output vectors corresponding to each of the target feature vectors into the attention layer, and filter the first output vectors according to the attention weight parameter value of the first output vector in the attention layer to obtain the second output vector.

[0114] The output subunit 70125 is used to input the second output vector into the output layer to determine multiple training scheduling plans corresponding to the comprehensive energy, and output the multiple training scheduling plans corresponding to the comprehensive energy.

[0115] The third generating subunit 70126 is used to adjust the model parameters of the LCA model according to the differences between the multiple training scheduling plans and the multiple prediction scheduling plans to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy.

[0116] In one embodiment of the present application, Figure 7 The CNN layer includes a convolutional layer and a dropout layer, and the extraction unit 70122 is specifically used to:

[0117] The input vector is input into the convolution layer to obtain a plurality of feature vectors of the input vector extracted by the convolution layer.

[0118] The plurality of feature vectors of the input vector are input to the dropout layer to filter out a target feature vector from the plurality of features.

[0119] In one embodiment of the present application, Figure 7 As shown, the training unit 7012 further includes:

[0120] The processing sub-unit 70127 is used to obtain multiple training scheduling plans corresponding to the comprehensive energy and perform denormalization processing on the multiple training scheduling plans.

[0121] The present application proposes an optimization device for a comprehensive energy scheduling plan, which obtains historical data samples corresponding to comprehensive energy and a comprehensive energy scheduling plan prediction model trained by the historical data samples, and inputs multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data in the historical training samples into the comprehensive energy scheduling plan prediction model to obtain multiple prediction scheduling plans, and combines the multiple historical scheduling plans in the historical training samples to construct an optimization function to optimize the model to obtain an optimized target comprehensive energy scheduling plan prediction model, thereby predicting the predicted photovoltaic power data, predicted wind power data, and predicted thermal power data to be predicted to obtain a predicted optimized scheduling plan. Therefore, the optimization of the comprehensive energy scheduling plan is achieved by combining the historical data samples, and the effectiveness and reliability of the comprehensive energy scheduling plan are improved.

[0122] like Figure 8 , which is a block diagram of an electronic device according to an embodiment of the present application.

[0123] like Figure 8 As shown, the electronic device includes:

[0124] Memory 801 , processor 802 , and computer instructions stored in the memory 801 and executable on the processor 802 .

[0125] When the processor 802 executes the instructions, the optimization method of the comprehensive energy scheduling plan provided in the above embodiment is implemented.

[0126] Furthermore, the electronic device further includes:

[0127] The communication interface 803 is used for communication between the memory 801 and the processor 802 .

[0128] The memory 801 is used to store computer instructions that can be executed on the processor 802 .

[0129] The memory 801 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0130] The processor 802 is configured to implement the method for optimizing the integrated energy scheduling plan of the above embodiment when executing the program.

[0131] If the memory 801, processor 802, and communication interface 803 are implemented independently, the communication interface 803, memory 801, and processor 802 can be connected to each other via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0132] Optionally, in a specific implementation, if the memory 801, the processor 802 and the communication interface 803 are integrated on a chip, the memory 801, the processor 802 and the communication interface 803 can communicate with each other through an internal interface.

[0133] The processor 802 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0134] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of such features. Throughout the description of this application, "plurality" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0135] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0136] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for optimizing a comprehensive energy scheduling plan, characterized in that: The method comprises: Obtaining historical data samples corresponding to the comprehensive energy and a comprehensive energy scheduling plan prediction model trained by the historical data samples, wherein the historical data samples include multiple historical photovoltaic power data, multiple historical wind power data, multiple historical thermal power data, and multiple historical scheduling plans for multiple same sampling periods; Inputting the plurality of historical photovoltaic power data, the plurality of historical wind power data, and the plurality of historical thermal power data into the comprehensive energy scheduling plan prediction model to obtain a plurality of prediction scheduling plans corresponding to the comprehensive energy; Based on the multiple forecast scheduling plans and the multiple historical scheduling plans, constructing an optimization function for optimizing the comprehensive energy scheduling plan prediction model; Optimizing the integrated energy scheduling plan prediction model according to the optimization function to obtain an optimized target integrated energy scheduling plan prediction model; Inputting the predicted photovoltaic power data, the predicted wind power data, and the predicted thermal power data to be predicted into the target comprehensive energy scheduling plan prediction model to obtain a predicted optimized scheduling plan; The acquiring of historical data samples corresponding to the comprehensive energy and a comprehensive energy scheduling plan prediction model trained by the historical data samples includes: Obtaining historical data samples corresponding to the comprehensive energy, and calculating multiple forecast scheduling plans corresponding to the comprehensive energy based on the multiple historical photovoltaic power data, multiple historical wind power data, and multiple historical thermal power data in the historical data samples; The multiple historical photovoltaic power data, multiple historical wind power data and multiple historical thermal power data are used as inputs of the LCA model, and the multiple forecast scheduling plans are used as outputs of the LCA model for model training to obtain a comprehensive energy scheduling plan forecasting model corresponding to the comprehensive energy.

2. The method according to claim 1, wherein The LCA model includes an input layer, a convolutional neural network (CNN) layer, a long short-term memory (LSTM) artificial neural network (ANN) layer, an attention layer, and an output layer. The LCA model uses the multiple historical photovoltaic power data, the multiple historical wind power data, and the multiple historical thermal power data as inputs to the LCA model, and uses the multiple forecast scheduling plans as outputs of the LCA model for model training to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy, including: Inputting the plurality of historical photovoltaic power data, the plurality of historical wind power data, and the plurality of historical thermal power data as training samples into the input layer, and obtaining input vectors corresponding to the respective training samples through the input layer; Inputting the input vector into the CNN layer, and performing feature extraction on the input vector to filter out a target feature vector; Obtain the target feature vectors, and input the target feature vectors into the LSTM layer to obtain first output vectors corresponding to the target feature vectors; Inputting the first output vectors corresponding to the target feature vectors into the attention layer, and filtering the first output vectors according to the attention weight parameter values of the first output vectors in the attention layer to obtain second output vectors; Inputting the second output vector into the output layer to determine a plurality of training scheduling plans corresponding to the comprehensive energy, and outputting the plurality of training scheduling plans corresponding to the comprehensive energy; According to the differences between the multiple training scheduling plans and the multiple prediction scheduling plans, the model parameters of the LCA model are adjusted to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy.

3. The method according to claim 2, wherein The CNN layer includes a convolution layer and a dropout layer, inputs the input vector into the CNN layer, and performs feature extraction on the input vector to filter out a target feature vector, including: Inputting the input vector into the convolution layer to obtain multiple feature vectors of the input vector extracted by the convolution layer; The plurality of feature vectors of the input vector are input to the dropout layer to filter out a target feature vector from the plurality of features.

4. The method according to claim 2, wherein After inputting the second output vector into the output layer to determine a plurality of training scheduling plans corresponding to the comprehensive energy, and outputting the plurality of training scheduling plans corresponding to the comprehensive energy, the method further includes: A plurality of training scheduling plans corresponding to the comprehensive energy are obtained, and a denormalization process is performed on the plurality of training scheduling plans.

5. An optimization device for a comprehensive energy scheduling plan, characterized in that: The device comprises: an acquisition module, configured to acquire historical data samples corresponding to the comprehensive energy, and a comprehensive energy scheduling plan prediction model trained by the historical data samples, wherein the historical data samples include a plurality of historical photovoltaic power data, a plurality of historical wind power data, a plurality of historical thermal power data, and a plurality of historical scheduling plans for a plurality of the same sampling periods; A generation module, configured to input the plurality of historical photovoltaic power data, the plurality of historical wind power data, and the plurality of historical thermal power data into the comprehensive energy scheduling plan prediction model to obtain a plurality of prediction scheduling plans corresponding to the comprehensive energy; A construction module, configured to construct an optimization function for optimizing the comprehensive energy scheduling plan prediction model based on the multiple prediction scheduling plans and the multiple historical scheduling plans; An optimization module, configured to optimize the integrated energy scheduling plan prediction model according to the optimization function to obtain an optimized target integrated energy scheduling plan prediction model; A prediction module, configured to input the predicted photovoltaic power data, the predicted wind power data, and the predicted thermal power data to be predicted into the target comprehensive energy scheduling plan prediction model to obtain a predicted optimized scheduling plan; The acquisition module includes: an acquisition unit, configured to acquire historical data samples corresponding to the comprehensive energy, and calculate multiple forecast scheduling plans corresponding to the comprehensive energy based on the multiple historical photovoltaic power data, the multiple historical wind power data, and the multiple historical thermal power data in the historical data samples; A training unit is used to use the multiple historical photovoltaic power data, the multiple historical wind power data and the multiple historical thermal power data as inputs of the LCA model, and use the multiple forecast scheduling plans as outputs of the LCA model for model training to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy.

6. The device according to claim 5, characterized in that The LCA model includes an input layer, a convolutional neural network (CNN) layer, a long short-term memory (LSTM) artificial neural network (ANN) layer, an attention layer, and an output layer. The training unit includes: an input subunit, configured to input the plurality of historical photovoltaic power data, the plurality of historical wind power data, and the plurality of historical thermal power data as training samples into the input layer, and obtain input vectors corresponding to the respective training samples through the input layer; an extraction subunit, configured to input the input vector into the CNN layer and perform feature extraction on the input vector to screen out a target feature vector; a first generating subunit, configured to obtain the target feature vectors, and input the target feature vectors into the LSTM layer to obtain first output vectors corresponding to the target feature vectors; a second generating subunit, configured to input the first output vectors corresponding to the respective target feature vectors into the attention layer, and filter the first output vectors according to the attention weight parameter values of the first output vectors in the attention layer to obtain a second output vector; an output subunit, configured to input the second output vector into the output layer to determine a plurality of training scheduling plans corresponding to the comprehensive energy, and output the plurality of training scheduling plans corresponding to the comprehensive energy; The third generating subunit is used to adjust the model parameters of the LCA model according to the differences between the multiple training scheduling plans and the multiple prediction scheduling plans to obtain a comprehensive energy scheduling plan prediction model corresponding to the comprehensive energy.

7. The device according to claim 6, characterized in that The CNN layer includes a convolutional layer and a dropout layer, and the extraction subunit is specifically used to: Inputting the input vector into the convolution layer to obtain multiple feature vectors of the input vector extracted by the convolution layer; The plurality of feature vectors of the input vector are input to the dropout layer to filter out a target feature vector from the plurality of features.

8. The device according to claim 6, wherein The training unit further includes: The processing subunit is used to obtain multiple training scheduling plans corresponding to the comprehensive energy and perform denormalization processing on the multiple training scheduling plans.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for optimizing the integrated energy scheduling plan as described in any one of claims 1 to 4 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the optimization method of the integrated energy scheduling plan as described in any one of claims 1 to 4 is implemented.

Images