Electric power system annual operation scene generation method and device considering different extreme weather events, electronic equipment and storage medium

Abstract

The invention discloses an electric power system annual operation scene generation method and device considering different extreme weather events, electronic equipment and a storage medium, and the method comprises the steps: obtaining a first new energy output and a first load of each time period of a historical typical year of an electric power system, and inputting the data into a deep belief prediction network to generate a new energy prediction output and load prediction annual power curve. Extracting new energy output and load under various extreme weather events from the historical data as second new energy output and second load; and inputting the second data into a universal output model to fit an abnormal new energy output curve and an abnormal load curve under the extreme weather event. And correcting the predicted output and load curve according to the abnormal curve, and generating a final new energy and load prediction annual power curve. And generating an annual operation scene of the power system in combination with the corrected new energy and load prediction curve. By implementing the method and the device, an annual operation scene which is more practical can be generated.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid operation scenario prediction, and particularly to a method, device, electronic device and storage medium for generating annual operation scenarios of a power system considering different extreme weather events. Background Art

[0002] With the rapid increase in the penetration rate of new energy sources such as wind power and photovoltaic power in the new power system, the impact of extreme weather events on the safe operation of the power system has become increasingly prominent. Extreme weather such as typhoons, high temperatures, and cold snaps not only cause drastic fluctuations in the output of new energy sources, but also result in abnormal surges in load demand. Traditional methods for generating annual operation scenarios are mostly based on historical meteorological statistical laws, and it is difficult to accurately reflect the coupling disturbance characteristics of both sides of power supply and demand under the superposition of multiple types of extreme weather, which easily leads to the deviation of power grid planning and operation strategies from the actual risk scenarios, and further causes problems such as curtailment of wind and light and power supply shortages.

[0003] When dealing with the impact of extreme weather on the power system, existing research has the following limitations: Firstly, most methods only analyze a single type of extreme weather, lacking a unified modeling framework for the correlation and differentiation of multiple types of weather events, resulting in low scenario generation efficiency and insufficient adaptability; Secondly, existing technologies focus on the impact of extreme weather on the output of the power supply side, but ignore its disturbance law on the load side, resulting in insufficient analysis of the disturbance law of the power system load side and inaccurate annual operation scenarios generated. Summary of the Invention

[0004] Embodiments of the present invention provide a method, device, electronic device and storage medium for generating annual operation scenarios of a power system considering different extreme weather events. By implementing the present invention, an annual operation scenario that is more in line with the actual extreme risk can be generated.

[0005] An embodiment of the present invention provides a method for generating annual operation scenarios of a power system considering different extreme weather events, including:

[0006] Obtain the first new energy output and the first load at each time period of the historical typical year of the power system;

[0007] Input the first new energy output and the first load into a preset deep belief prediction network, so that the deep belief prediction network generates an annual power curve of new energy predicted output and an annual power curve of predicted load according to the first new energy output and the first load;

[0008] Extract the first new energy output at each time period under various extreme weather events from the first new energy output at each time period of the historical typical year as the second new energy output; extract the first load at each time period under various extreme weather events from the first load at each time period of the historical typical year as the second load;

[0009] Input the second new energy output and the second load into a preset general output model, so that the general output model fits the abnormal new energy output curve and the abnormal load curve under various types of extreme weather events according to the second new energy output and the second load;

[0010] Revise the annual power curve of the predicted new energy output according to the abnormal new energy output curve to generate the final annual power curve of the predicted new energy output; revise the annual power curve of the predicted load according to the abnormal load curve to generate the final annual power curve of the predicted load;

[0011] Generate the annual operation scenario of the power system according to the final annual power curve of the predicted new energy output and the final annual power curve of the predicted load.

[0012] Furthermore, the deep belief prediction network is composed of stacking multiple restricted Boltzmann machines and a top-layer BP neural network;

[0013] The training of the deep belief prediction network includes:

[0014] Obtain the historical output and load data set of the power system for training; wherein, the historical output and load data set of the power system includes the second new energy output at each time period of the training historical years, the second load at each time period, and the corresponding labels; the labels include the new energy output labels at each time period of the year following the training historical years and the load labels at each time period.

[0015] According to the historical output and load data set of the power system, with the goal of maximizing a preset log-likelihood function, perform unsupervised training on the restricted Boltzmann machines in the deep belief prediction network to generate the trained restricted Boltzmann machines and update the deep belief prediction network; wherein, during the process of training the restricted Boltzmann machines, freeze the network parameters of the top-layer BP neural network.

[0016] Randomly divide the historical output load dataset of the power system into several batches of training samples according to a preset quantity, and sequentially input each batch of training samples into the deep belief prediction network to train the deep belief prediction network until the prediction training times are reached, generate the trained top-level BP neural network, and update the deep belief prediction network; wherein, when the deep belief prediction network model receives each batch of training samples, the trained restricted Boltzmann machine outputs the current probability distribution according to the current batch of training samples; the top-level BP neural network outputs the predicted new energy output and predicted load at the same moment of the next year for the current batch of training samples according to the current probability distribution; according to the predicted new energy output, predicted load and the corresponding labels, calculate the loss function value through cross-entropy loss; use the optimizer to update the current top-level BP neural network according to the loss function value.

[0017] Further, the construction of the general output model includes:

[0018] Obtain the third new energy output and the third load of the power system in the historical period under various extreme weather events.

[0019] Based on the polynomial fitting algorithm optimized by the least squares method, construct a general output model according to the third new energy output and the third load.

[0020] Further, the step of inputting the first new energy output and the first load into a preset deep belief prediction network so that the deep belief prediction network generates an annual power curve of the predicted new energy output and an annual power curve of the predicted load according to the first new energy output and the first load includes:

[0021] Input the first new energy output and the first load into a preset deep belief prediction network so that the deep belief prediction network generates the predicted new energy output corresponding to each time period and the predicted load corresponding to each time period according to the first new energy output and the first load.

[0022] Generate an annual power curve of the predicted new energy output according to the predicted new energy output corresponding to each time period.

[0023] Generate an annual power curve of the predicted load according to the predicted load corresponding to each time period.

[0024] Further, the step of correcting the annual power curve of the predicted load according to the abnormal load curve to generate the final annual power curve of the predicted load includes:

[0025] Obtain the typical frequencies of various extreme weather events occurring in each month of the power system.

[0026] Extract the data for each time period in the abnormal load curve to generate an abnormal load sequence for each type of extreme weather event;

[0027] For each type of extreme weather event, according to the typical frequency of the current type of extreme weather event occurring in each month and the corresponding abnormal load sequence, correct the predicted load annual power curve; wherein, in each correction, select the data of the time period with the smallest value in the abnormal load sequence and replace the data of a random time period in the corresponding month of the predicted load annual power curve until the number of replacements reaches the typical frequency of the current type of extreme weather event in the corresponding month.

[0028] Further, the obtaining of the typical frequency of each type of extreme weather event occurring in each month of the power system includes:

[0029] Obtain the frequency and monthly distribution of each type of extreme weather event occurring in the power system in recent years;

[0030] Calculate the average of the frequencies of each type of extreme weather event occurring in recent years to generate the annual average frequency of each type of extreme weather event;

[0031] According to the monthly distribution of each type of extreme weather event occurring in recent years and the frequency of each type of extreme weather event occurring in recent years, calculate the monthly proportion of each type of extreme weather event;

[0032] Calculate the monthly proportion and the annual average frequency to generate the typical frequency of each type of extreme weather event occurring in each month.

[0033] Further, the generating of the annual operation scenario of the power system according to the final new energy predicted output annual power curve and the final predicted load annual power curve includes:

[0034] Match the final new energy predicted output annual power curve and the final predicted load annual power curve on a time period by time period basis to generate the planned operation data for each time period;

[0035] According to the planned operation data, perform a power flow calculation to generate a power flow calculation result;

[0036] Eliminate the planned operation data whose power flow calculation results do not meet the preset operation constraints to generate the annual operation scenario of the power system.

[0037] Based on the above method item embodiments, the present invention correspondingly provides device item embodiments.

[0038] An embodiment of the present invention provides a device for generating an annual operation scenario of a power system considering different extreme weather events, including: a historical operation data acquisition module, an annual power curve generation module, an extreme weather data acquisition module, an extreme weather power curve generation module, an annual power curve correction module, and an annual operation scenario generation module;

[0039] The historical operation data acquisition module is configured to acquire the first new energy output and the first load of each time period of a historical typical year of the power system;

[0040] The annual power curve generation module is configured to input the first new energy output and the first load into a preset deep belief prediction network, so that the deep belief prediction network generates an annual power curve of the predicted new energy output and an annual power curve of the predicted load according to the first new energy output and the first load;

[0041] The extreme weather data acquisition module is configured to extract the first new energy output of each time period under various extreme weather events from the first new energy output of each time period of the historical typical year as the second new energy output; and extract the first load of each time period under various extreme weather events from the first load of each time period of the historical typical year as the second load;

[0042] The extreme weather power curve generation module is configured to input the second new energy output and the second load into a preset general output model, so that the general output model fits an abnormal new energy output curve and an abnormal load curve under various types of extreme weather events according to the second new energy output and the second load;

[0043] The annual power curve correction module is configured to correct the annual power curve of the predicted new energy output according to the abnormal new energy output curve to generate a final annual power curve of the predicted new energy output; and correct the annual power curve of the predicted load according to the abnormal load curve to generate a final annual power curve of the predicted load;

[0044] The annual operation scenario generation module is configured to generate the annual operation scenario of the power system according to the final annual power curve of the predicted new energy output and the final annual power curve of the predicted load.

[0045] Based on the above method item embodiment, the present invention correspondingly provides an electronic device item embodiment.

[0046] An embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method for generating an annual operation scenario of a power system considering different extreme weather events according to any one of the above method item embodiments can be implemented.

[0047] Based on the above method embodiments, the present invention correspondingly provides storage medium embodiments.

[0048] An embodiment of the present invention provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, it can implement the method for generating the annual operation scenario of the power system considering different extreme weather events described in any one of the above method embodiments.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] The embodiments of the present invention provide a method, device, electronic device and storage medium for generating the annual operation scenario of the power system considering different extreme weather events. The method obtains the new energy output and load data of each time period of the historical typical year of the power system, as well as the new energy output and load data under various extreme weather events, and uses a deep belief prediction network to generate the annual power curves of the new energy output and load. Subsequently, a general output model is used to fit the abnormal new energy output curves and load curves under different extreme weather events, and these abnormal curves are used to correct the original predicted annual power curves, and finally a more accurate annual operation scenario of the power system is generated.

[0051] By collecting the new energy output data and load data under various types of extreme weather events and using a general output model for fitting, the present invention uniformly constructs the abnormal disturbance laws on both the source and load sides under multiple types of weather events, solving the limitation of single weather modeling in the existing methods; at the same time, using the abnormal curve dynamic correction mechanism to synchronously optimize the prediction deviation of the new energy output and load, making up for the defect that the traditional model ignores the meteorological correlation on the load side, so as to generate an annual operation scenario that is more in line with the actual extreme risks. Description of the Drawings

[0052] Figure 1 is a schematic flowchart of a method for generating the annual operation scenario of the power system considering different extreme weather events provided by an embodiment of the present invention.

[0053] Figure 2 is a schematic structural diagram of a device for generating the annual operation scenario of the power system considering different extreme weather events provided by an embodiment of the present invention. Detailed Embodiments

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0055] As Figure 1 shown, an embodiment of the present invention provides a method for generating an annual operation scenario of a power system considering different extreme weather events, which at least includes the following steps:

[0056] Step S1: Obtain the first new energy output and the first load of each time period in the historical typical year of the power system;

[0057] Specifically, the historical typical year can be the previous year of the year to be predicted; it can also be several historical years of the year to be predicted, which is not limited here. In order to establish an accurate power system model and conduct reasonable load forecasting and new energy output evaluation, it is first necessary to obtain the new energy output and load data of the power system in each time period of the historical typical year. The new energy output mainly includes wind power and photovoltaic output, and the load data is usually provided by the actual electricity demand. The time period can be 1 hour or 15 minutes, which is not limited here.

[0058] Step S2: Input the first new energy output and the first load into a preset deep belief prediction network, so that the deep belief prediction network generates an annual power curve of the new energy predicted output and an annual power curve of the predicted load according to the first new energy output and the first load;

[0059] Specifically, after obtaining the first new energy output and load data of each time period in the historical typical year, the next step is to input these data into a deep learning model for more accurate prediction. In order to be able to reasonably and efficiently predict the future new energy output and load, especially in extreme weather conditions, using a deep belief network (Deep Belief Network, abbreviated as DBN) is an effective way.

[0060] The deep belief network (DBN) is a deep neural network based on unsupervised learning. It can capture complex patterns and potential features in data through layer-by-layer training. The DBN usually consists of multiple restricted Boltzmann machine (Restricted Boltzmann Machines, RBM) layers. Through layer-by-layer training, a deep model with high expression ability is finally obtained for processing power prediction tasks with time series properties.

[0061] In a preferred embodiment, the deep belief prediction network is composed of a stack of multiple restricted Boltzmann machines and a top-layer BP neural network;

[0062] The training of the deep belief prediction network includes:

[0063] Obtain a historical output load dataset of the power system for training; wherein, the historical output load dataset of the power system includes the second new energy output, the second load at each time period in the training historical years, and the corresponding labels; the labels include the new energy output labels and the load labels at each time period in the next year of the training historical years.

[0064] According to the historical output load dataset of the power system, with the goal of maximizing a preset log-likelihood function, perform unsupervised training on the restricted Boltzmann machine in the deep belief prediction network to generate a trained restricted Boltzmann machine, and update the deep belief prediction network; wherein, during the process of training the restricted Boltzmann machine, freeze the network parameters of the top-layer BP neural network.

[0065] Randomly divide the historical output load dataset of the power system into several batches of training samples according to a preset quantity, and sequentially input each batch of training samples into the deep belief prediction network to train the deep belief prediction network until the prediction training times are reached, generate a trained top-layer BP neural network, and update the deep belief prediction network; wherein, when the deep belief prediction network model receives each batch of training samples, it outputs the current probability distribution according to the current batch of training samples through the trained restricted Boltzmann machine; outputs the predicted new energy output and predicted load at the same moment in the next year of the current batch of training samples through the top-layer BP neural network according to the current probability distribution; calculates the loss function value through cross-entropy loss according to the predicted new energy output, predicted load, and the corresponding labels; and updates the current top-layer BP neural network by using an optimizer according to the loss function value.

[0066] It should be explained here that the deep belief network is mainly composed of stacking multiple restricted Boltzmann machines and a top-layer BP neural network.

[0067] The joint probability distribution energy density function of the visible layer v and the hidden layer h of the restricted Boltzmann machine is:

[0068]

[0069] wherein, θ = {ω, a, b} are the parameters of the restricted Boltzmann machine; ω is the weight of the connection between the visible layer and the hidden layer; ω mn is the weight of the connection between the m-th node of the visible layer and the n-th node of the hidden layer; a and b represent the bias values corresponding to the visible layer nodes and the hidden layer nodes respectively, a m is the bias value of the m-th node, b n is the bias value of the n-th node; v m is the probability of the m-th node of the visible layer, h n is the probability of the n-th node of the hidden layer; N represents the number of neurons in the visible layer; M represents the number of neurons in the hidden layer.

[0070] In a restricted Boltzmann machine, when the state of the hidden layer h is known, the probability that the m-th node of the visible layer is activated is:

[0071]

[0072] where σ is the activation function; common activation functions include the sigmoid function, the tanh function, etc.

[0073] Similarly, according to the structural characteristics of the restricted Boltzmann machine, when the state of the visible layer v is known, the probability that the j-th node of the hidden layer is activated is:

[0074]

[0075] By training the restricted Boltzmann machine layer by layer, seeking the RBM parameters θ = {ω, a, b} of each layer, and using the maximum log-likelihood function method to obtain the parameters:

[0076]

[0077] where L(θ) is the maximum likelihood function; g t is the value of the new energy output and the load at the t-th moment.

[0078] In a preferred embodiment, the step of inputting the first new energy output and the first load into a preset deep belief prediction network, so that the deep belief prediction network generates an annual power curve of the new energy predicted output and an annual power curve of the predicted load according to the first new energy output and the first load, includes:

[0079] Inputting the first new energy output and the first load into a preset deep belief prediction network, so that the deep belief prediction network generates the new energy predicted output corresponding to each time period and the predicted load corresponding to each time period according to the first new energy output and the first load;

[0080] Generating an annual power curve of the new energy predicted output according to the new energy predicted output corresponding to each time period;

[0081] Generating an annual power curve of the predicted load according to the predicted load corresponding to each time period.

[0082] Specifically, input the first new energy output and the first load into a preset deep belief prediction network, so that the deep belief prediction network generates the predicted new energy output corresponding to each time period and the predicted load corresponding to each time period according to the first new energy output and the first load. The deep belief network extracts features and learns patterns from the input time-series data through multiple layers of neurons, captures the temporal relationships and potential non-linear laws therein, and then predicts the new energy output and load demand in future time periods. The output of these time-period data will form the predicted values of wind power output, photovoltaic output, and load corresponding to each time period.

[0083] Generate an annual power curve of the predicted new energy output according to the predicted new energy output corresponding to each time period. By aggregating the predicted output data of each time period in a time series, the predicted power curves of wind power output and photovoltaic output for the whole year are obtained. This curve shows the output trend and fluctuations of new energy at different time points.

[0084] Generate an annual power curve of the predicted load according to the predicted load corresponding to each time period. Similarly, aggregate the load prediction values of each time period annually to obtain the annual load prediction power curve, which reflects the change trend of load demand at different time periods.

[0085] Step S3: Extract the first new energy output in each time period of various extreme weather events from the first new energy output in each time period of a historical typical year as the second new energy output; extract the first load in each time period of various extreme weather events from the first load in each time period of a historical typical year as the second load;

[0086] It should be noted here that the first new energy output in each time period of various extreme weather events is extracted from the first new energy output in each time period of a historical typical year as the second new energy output. The core of this step is to screen out the data in the time periods when specific extreme weather events occur through historical data, such as continuous high temperature, cold wave, typhoon, or extreme static stability weather, etc. The data in these time periods reflect the output characteristics of new energy (such as wind power, photovoltaic) under these extreme weather conditions. By screening the first new energy output in each time period of a historical typical year, the new energy output data that meet these extreme weather conditions are extracted to obtain a second new energy output sequence with strong representativeness, providing a basis for subsequent analysis of the change trends and laws of new energy under extreme weather.

[0087] Similarly, the first load in each time period of historical typical years is extracted for each time period under various extreme weather events as the second load. This step obtains the variation of the grid load under different extreme weather events through the screening and extraction of historical load data. For example, during a cold snap, the electricity demand may increase significantly, or during a typhoon, the load may show different fluctuation patterns. Through this screening, the load data that meets the extreme weather conditions is obtained as the second load. These second load data will provide data support for further analyzing the fluctuation and demand characteristics of the grid load under extreme weather.

[0088] Step S4: Input the second new energy output and the second load into a preset general output model, so that the general output model fits the abnormal new energy output curve and the abnormal load curve under various types of extreme weather events according to the second new energy output and the second load;

[0089] Specifically, input the second new energy output and the second load into a preset general output model, so that the general output model fits the abnormal new energy output curve and the abnormal load curve under various types of extreme weather events according to the second new energy output and the second load. The second new energy output and the second load are respectively derived from the actual observation data of historical typical years under different extreme weather conditions, representing the real impact of extreme weather events on the new energy output and load fluctuations. The general output model will fit these input data through optimization algorithms such as the least squares method to generate abnormal output curves related to different types of extreme weather events.

[0090] In this process, the general output model not only needs to consider the trend changes under normal weather conditions but also needs to combine the special changes brought about by extreme weather events. For example, the output characteristics of wind power and photovoltaic power generation under weather such as cold snaps or typhoons may be significantly different from those under normal weather, and the volatility of the load curve may also increase significantly under extreme climate conditions. Through the fitting of these historical data, the general output model can capture these changes and generate new energy output curves and load curves that can reflect abnormal fluctuations. These fitted curves will provide a more accurate basis for subsequent power prediction and grid dispatching, helping to prevent and address power system instability problems under extreme weather.

[0091] In a preferred embodiment, the construction of the general output model includes:

[0092] Obtain the third new energy output and the third load of the power system in historical time periods under various extreme weather events;

[0093] A polynomial fitting algorithm optimized based on the least squares method constructs a general output model according to the third new energy output and the third load.

[0094] Specifically, historical data of new energy output and load under different types of extreme weather events such as continuous high temperature, cold wave, typhoon, and extreme static stability weather in the regional power grid are collected, and the polynomial theory based on the least squares method is used to characterize the evolution law of new energy output and load during extreme weather events and construct a general output model.

[0095] The wind power output, photovoltaic output, and load power sequences under different extreme weather events in the regional power grid are specifically as follows:

[0096]

[0097] Among them, L s is the load power sequence under the collected extreme weather events; P w is the wind power sequence under the collected extreme weather events; P PV is the photovoltaic power sequence under the collected extreme weather events; is the value of the load power sequence at time t under the collected extreme weather events; is the power of the wind power sequence at time t under the collected extreme weather events; is the power of the photovoltaic power sequence at time t under the collected extreme weather events; T is the total time period, which is determined according to the actual occurrence time of different types of extreme weather events such as continuous high temperature, cold wave, typhoon, and extreme static stability weather.

[0098] The general output models of wind power output, photovoltaic output, and load based on the polynomial theory optimized by the least squares method are:

[0099]

[0100] Among them, x is a set that contains wind power output, photovoltaic output, and load elements, and x t is the value of any element in the set at a time period; and are respectively the values of wind power output, photovoltaic output, and load at a time period; T is the duration of the extreme weather event; a0 to a m are all polynomial coefficients; m is the number of terms of the polynomial function; p(x t ) is the fitting result of any element in the set x t at time t; y t is the actual value of any element in the set x t at time t.

[0101] The least - squares - based polynomial method can quickly and accurately fit the new - energy output curves and load curves under various types of extreme - weather times, and has the advantages of simple process, low workload, and high generality compared with the current combined method of mechanism analysis and data - driven approach.

[0102] Step S5: Modify the annual power curve of the predicted new - energy output according to the abnormal new - energy output curve to generate the final annual power curve of the predicted new - energy output; modify the annual power curve of the predicted load according to the abnormal load curve to generate the final annual power curve of the predicted load.

[0103] Specifically, modify the annual power curve of the predicted new - energy output according to the abnormal new - energy output curve to generate the final annual power curve of the predicted new - energy output. The abnormal new - energy output curve usually represents the abnormal fluctuations or unforeseen performance changes of the new - energy system under certain extreme weather or special working conditions. These abnormal fluctuations may be caused by factors such as sudden changes in wind speed, changes in light intensity, and equipment failures. When generating the annual power curve of the predicted new - energy output, the influencing factors of the abnormal new - energy output curve are taken into account, and the original annual prediction curve is corrected as necessary. By making fine - tuning to the predicted annual power curve, the corrected curve can more realistically reflect the fluctuation trend of the new - energy output under different extreme - weather scenarios, improving the accuracy and reliability of the prediction results.

[0104] Modify the annual power curve of the predicted load according to the abnormal load curve to generate the final annual power curve of the predicted load. The abnormal fluctuations of the load curve are usually closely related to factors such as seasonal changes, abnormal temperatures, changes in economic activities, and extreme weather. For example, cold snap weather may cause a sharp increase in winter load, and natural disasters such as typhoons may cause sudden increases or decreases in load. By analyzing the abnormal load curve, the deviations existing in the load prediction can be identified, and the annual power curve of the predicted load can be adjusted accordingly. The corrected load curve can more accurately predict the future power demand at each time period, providing more reliable data support for the load dispatching and operation of the power grid.

[0105] In a preferred embodiment, the step of modifying the annual power curve of the predicted load according to the abnormal load curve to generate the final annual power curve of the predicted load includes:

[0106] Obtain the typical frequencies of various extreme - weather events occurring in each month of the power system.

[0107] Extract the data of each time period in the abnormal load curve to generate abnormal load sequences under various types of extreme - weather events.

[0108] For each type of extreme weather event, the annual power curve of the predicted load is corrected according to the typical frequency of the occurrence of the current type of extreme weather event in each month and the corresponding abnormal load sequence. Among them, during each correction, the time period data with the smallest value in the abnormal load sequence is selected to replace the random time period data of the corresponding month in the annual power curve of the predicted load until the number of replacements reaches the typical frequency of the current type of extreme weather event in the corresponding month.

[0109] Specifically, after obtaining the typical frequencies of various extreme weather events occurring in each month of the power system, it is first necessary to extract the data of each time period in the abnormal load curve to generate the abnormal load sequences under various types of extreme weather events. At this time, the abnormal load sequence represents the change trend and fluctuation characteristics of the load during different extreme weather events. The abnormal load sequence under each type of extreme weather event contains the abnormal fluctuation information of the load within a specific time period, and this information can help us better predict the load change of the power grid when extreme weather events occur in the future.

[0110] Next, for each type of extreme weather event, the annual power curve of the predicted load is corrected according to the typical frequency of the occurrence of the current type of extreme weather event in each month and the corresponding abnormal load sequence. During the correction process, according to the typical frequency of extreme weather events in each month, part of the data of the corresponding month in the annual power curve of the predicted load will be gradually replaced. Each time a replacement is made, the time period data with the smallest value in the abnormal load sequence is selected and replaced with a random time period data of the corresponding month in the annual power curve of the predicted load. This replacement process will continue until the number of replacements reaches the typical frequency of the current type of extreme weather event in the corresponding month.

[0111] This correction method aims to make the annual power curve of the predicted load more conform to the actual extreme weather conditions. By introducing the extreme weather influence factors in the historical data, the accuracy and reliability of the load prediction are ensured. The finally generated corrected load curve will more reflect the actual situation of the power grid load fluctuation when extreme weather events occur, thus providing a more accurate decision-making basis for the dispatching and management of the power system.

[0112] In one embodiment, the minimum value of the fitting results of all extreme weather events of the same type is statistically calculated and arranged in descending order from largest to smallest, specifically as follows:

[0113]

[0114] s.t.

[0115]

[0116] s.t.

[0117] 2 ≤ g ≤ k

[0118] Among them, is the sequence arranged from high to low of the minimum values of the fitting results of all extreme weather events of the h-th type; k is the total number of occurrences of the extreme weather events of the h-th type in the historical data; is the set x i The minimum value of any element in all time periods of the g-th fitting result of the extreme weather event of the h-th type; i is the time period of the g-th fitting result of the extreme weather event of the h-th type in the historical data fitted by the general model; H is the total time period of the g-th fitting result of the extreme weather event of the h-th type in the historical data fitted by the general model; ψ is the minimum continuous duration of any extreme weather event.

[0119] The correction of the annual output prediction results of wind power, photovoltaic power and load for each type of extreme weather event will be carried out according to the above formula, that is, the fitting result of the minimum value in the sequence is preferentially embedded, and they are embedded in ascending order. In addition, different types of extreme weather events are all embedded according to the above method.

[0120] In a preferred embodiment, the obtaining of the typical frequencies of various extreme weather events occurring in each month of the power system includes:

[0121] Obtain the frequencies and monthly distribution of various extreme weather events occurring in the power system in recent years;

[0122] Average the frequencies of various extreme weather events occurring in recent years to generate the annual average frequency of each type of extreme weather event;

[0123] According to the monthly distribution of various extreme weather events occurring in recent years and the frequencies of various extreme weather events occurring in recent years, calculate the monthly proportion of each type of extreme weather event;

[0124] Calculate the monthly proportion and the annual average frequency to generate the typical frequencies of various extreme weather events occurring in each month.

[0125] It should be noted here that in one embodiment, the frequencies of various extreme weather events are statistically calculated based on historical data, and the frequencies of the new energy output and load generation scenarios are determined according to the frequency means. Specifically:

[0126]

[0127] Among them, is the ceiling symbol, that is, the result automatically rounds up when there is a decimal point; h is the type of extreme weather event, and h{1, 2, 3, 4} respectively correspond to 4 types of continuous high temperature, cold wave, typhoon and extreme static stability weather; and are the frequencies of the h - type extreme weather events occurring in the wind power output, photovoltaic power output, and load generation scenarios respectively; Y is the number of annual historical data; and are the frequencies of the h - type extreme weather events occurring in the wind power output, photovoltaic power output, and load in the Y - th year respectively.

[0128] Based on the statistics of historical data and the distribution and proportion of different extreme weather events in different months in the year before the generation scenario, calculate the final frequency results for different months according to the generation scenario frequencies.

[0129]

[0130] Among them, and are the frequencies of the h - type extreme weather events occurring in the M - th month in the historical data of the year before the generation scenario respectively; and are the proportions of the h - type extreme weather events occurring in the wind power output, photovoltaic power output, and load in the M - th month in the historical data of the year before the generation scenario respectively.

[0131] The calculation formula for the final frequency results (i.e., typical frequencies) of different extreme weather events in different months for the wind power output, photovoltaic power output, and load sequence generation scenarios is:

[0132]

[0133] Among them, and are the frequencies of the h - type extreme weather events occurring in the M - th month for the new energy output and load generation scenarios respectively.

[0134] In a preferred embodiment, from the perspective of power planning, the specific time of occurrence of extreme weather events in a certain month does not affect the subsequent relevant applications of the annual operation scenario of the new power system (such as carrying out capacity planning for multi - type energy storage systems, etc.). Therefore, the embedding points can be selected through the following conditions:

[0135] Condition 1: The latter time period of the time period with the largest difference between this month and the initial time period of extreme weather is used as the embedding point of the fitting result of extreme weather events.

[0136] Condition 2: To ensure that the embedding results of various types of extreme weather events meet the above - mentioned monthly frequency calculation results, the embedding positions of various types of extreme weather events need to meet the following: The last time period of each type of extreme weather event after embedding shall not exceed the last time period of this month (i.e., the last hour of the last day).

[0137] If Condition 1 conflicts with Condition 2, then remove the selection result of the time period with the largest difference in Condition 1, and reselect the time period with the largest difference according to the calculation method of Condition 1 to find the embedding point. In addition, if the search results of the embedding points of multiple types of extreme weather events overlap in some time periods, the embedding results of the overlapping time periods are the maximum and minimum values of the fitting results of different extreme weather events in each time period.

[0138] Step S6: Generate the annual operation scenario of the power system according to the final annual power curve of the new energy predicted output and the final annual power curve of the predicted load.

[0139] In a preferred embodiment, the generating the annual operation scenario of the power system according to the final annual power curve of the new energy predicted output and the final annual power curve of the predicted load includes:

[0140] Match the final annual power curve of the new energy predicted output and the final annual power curve of the predicted load hour by hour to generate the planned operation data for each time period;

[0141] Calculate the power flow calculation according to the planned operation data to generate the power flow calculation result;

[0142] Remove the planned operation data whose power flow calculation results do not meet the preset operation constraints to generate the annual operation scenario of the power system.

[0143] Specifically, matching the final annual power curve of the new energy predicted output and the final annual power curve of the predicted load hour by hour means corresponding and pairing the new energy output and load data within each time period to ensure that the load demand and power generation capacity match within each time period. Through this matching process, the planned operation data for each time period can be generated, reflecting the power generation and load demand status of the power system at different time points.

[0144] Based on these planned operation data, power flow calculation can be performed to determine the power flow situation of each node in the power system. Power flow calculation is an important step in power system analysis, aiming to simulate and evaluate the power flow direction, power and voltage of each electrical equipment in the power grid under different operating conditions. Through power flow calculation, the power flow distribution of each line, substation, generator and load equipment in the power system can be obtained, and the operation of the power grid under different load and power generation conditions can be evaluated.

[0145] After obtaining the power flow calculation results, it is necessary to screen and eliminate them to ensure that the calculation results meet the preset operating constraints of the power system. Operating constraints usually include restrictions such as the voltage, power, frequency, and line carrying capacity of the power grid. Eliminating the planned operating data that does not meet the operating constraints means only retaining the data that meets the safe and stable operating conditions of the power system and eliminating any situations that may cause the power grid to operate unstably.

[0146] Finally, the planned operating data after power flow calculation and screening will generate the annual operating scenarios of the power system. These annual operating scenarios show the operating status of the power system under different extreme weather conditions and load demands, providing important data support for the dispatching, optimization, and risk assessment of the power grid. By analyzing the annual operating scenarios, the operators of the power system can predict and prevent potential operating risks and provide a basis for future power grid planning and optimization.

[0147] Extreme weather events have the typical characteristics of low probability and high impact, and are particularly critical for the new power system. Compared with the existing technologies, the present invention proposes a discrimination method for extreme weather events from the perspective of abnormal new energy output and conducts category division, including the load within the scope of the impact of some extreme weather events. A general model of new energy output and load under different extreme weather events is constructed based on the polynomial theory, and a multi-stage generation method for the annual system operating scenarios considering multiple extreme weather events is proposed by using the deep belief network and the embedding theory.

[0148] As Figure 2 shown, an embodiment of the present invention provides a device for generating the annual operating scenarios of a power system considering different extreme weather events, including: a historical operating data acquisition module, an annual power curve generation module, an extreme weather data acquisition module, an extreme weather power curve generation module, an annual power curve correction module, and an annual operating scenario generation module;

[0149] The historical operating data acquisition module is used to acquire the first new energy output and the first load of each time period in the historical typical year of the power system;

[0150] The annual power curve generation module is used to input the first new energy output and the first load into a preset deep belief prediction network, so that the deep belief prediction network generates an annual power curve of the predicted new energy output and an annual power curve of the predicted load according to the first new energy output and the first load;

[0151] The extreme weather data acquisition module is used to extract the first new energy output at each time period under various extreme weather events from the first new energy output at each time period of historical typical years as the second new energy output; and extract the first load at each time period under various extreme weather events from the first load at each time period of historical typical years as the second load.

[0152] The extreme weather power curve generation module is used to input the second new energy output and the second load into a preset general output model, so that the general output model fits the abnormal new energy output curve and the abnormal load curve under various types of extreme weather events according to the second new energy output and the second load.

[0153] The annual power curve correction module is used to correct the annual power curve of the new energy predicted output according to the abnormal new energy output curve to generate the final annual power curve of the new energy predicted output; and correct the annual power curve of the predicted load according to the abnormal load curve to generate the final annual power curve of the predicted load.

[0154] The annual operation scenario generation module is used to generate the annual operation scenario of the power system according to the final annual power curve of the new energy predicted output and the final annual power curve of the predicted load.

[0155] It should be noted that the embodiments of the device described above correspond to the above embodiments of the present invention and can implement the method for generating the annual operation scenario of the power system considering different extreme weather events described in any one of the above of the present invention. In addition, the embodiments of the above device are merely illustrative. The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that they have a communication connection, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement without creative work.

[0156] Based on the above method embodiment of the present invention, an embodiment of an electronic device is correspondingly provided.

[0157] An embodiment of the present invention provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the method for generating an annual operation scenario of a power system considering different extreme weather events according to any one of the present invention, or when the processor executes the computer program, it implements the functions of each module in the above device embodiments.

[0158] Exemplarily, the computer program can be divided into one or more modules. The one or more modules are stored in the memory and executed by the processor to complete the present invention. The one or more modules can be a series of computer program instruction segments capable of completing specific functions, and these instruction segments are used to describe the execution process of the computer program in the terminal device.

[0159] The terminal device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory.

[0160] The so-called processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the terminal device, and connects various parts of the entire terminal device through various interfaces and lines.

[0161] The memory can be used to store the computer programs and / or modules. By running or executing the computer programs and / or modules stored in the memory and invoking the data stored in the memory, the processor realizes various functions of the terminal device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function, etc.; the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage devices.

[0162] Based on the above method item embodiments, the present invention correspondingly provides storage medium item embodiments;

[0163] Another embodiment of the present invention provides a storage medium. The storage medium includes a stored computer program. When the computer program runs, it controls the device where the storage medium is located to execute any one of the above methods for generating an annual operation scenario of a power system considering different extreme weather events in the present invention.

[0164] Among them, the above storage medium is a computer-readable storage medium. The computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0165] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0166] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A method for generating annual operation scenarios of a power system considering different extreme weather events, characterized in that: include: Obtain the top renewable energy output and top load of each period in a typical historical year of the power system; Inputting the first new energy output and the first load into a preset deep belief prediction network, so that the deep belief prediction network generates a new energy predicted output annual power curve and a predicted load annual power curve according to the first new energy output and the first load; Extract the first renewable energy output in each period under various extreme weather events from the first renewable energy output in each period of a typical historical year as the second renewable energy output; Extract the first load of each period under various extreme weather events from the first load of each period in historical typical years as the second load; Inputting the second new energy output and the second load into a preset universal output model, so that the universal output model fits the abnormal new energy output curve and the abnormal load curve under various types of extreme weather events according to the second new energy output and the second load; Correcting the predicted annual power curve of new energy output according to the abnormal new energy output curve to generate a final predicted annual power curve of new energy output; Correcting the predicted load annual power curve according to the abnormal load curve to generate a final predicted load annual power curve; The annual operation scenario of the power system is generated according to the final predicted annual power curve of new energy output and the final predicted annual power curve of load.

2. The method for generating annual operation scenarios of a power system considering different extreme weather events according to claim 1, characterized in that: The deep belief prediction network is composed of a stack of multiple restricted Boltzmann machines and a top-level BP neural network; The training of the deep belief prediction network includes: Acquire a power system historical output load data set for training; wherein the power system historical output load data set includes the second renewable energy output of each period of the training historical year, the second load of each period, and the corresponding label; the label includes the renewable energy output label of each period of the next year of the training historical year and the load label of each period; According to the historical output load data set of the power system, with the goal of maximizing a preset log-likelihood function, unsupervised training is performed on the restricted Boltzmann machine in the deep belief prediction network to generate a trained restricted Boltzmann machine, and the deep belief prediction network is updated; wherein, in the process of training the restricted Boltzmann machine, the network parameters of the top-level BP neural network are frozen; The historical output and load data set of the power system is randomly divided into several batches of training samples according to a preset number, and the training samples of each batch are input into the deep belief prediction network in turn, and the deep belief prediction network is trained until the prediction training times are reached, and the trained top-level BP neural network is generated, and the deep belief prediction network is updated; wherein, when the deep belief prediction network model receives each batch of training samples, it outputs the current probability distribution according to the training samples of the current batch through the trained restricted Boltzmann machine; outputs the new energy predicted output and predicted load of the current batch of training samples at the same time of the next year according to the current probability distribution through the top-level BP neural network; calculates the loss function value through the cross entropy loss according to the new energy predicted output, predicted load and corresponding labels; and uses the optimizer to update the current top-level BP neural network according to the loss function value.

3. The method for generating annual operation scenarios of a power system considering different extreme weather events as claimed in claim 2, characterized in that: The construction of the universal output model includes: Obtain the third renewable energy output and third load of the power system during various extreme weather events in historical periods; Based on a polynomial fitting algorithm optimized by the least squares method, a universal output model is constructed according to the third new energy output and the third load.

4. The method for generating annual operation scenarios of a power system considering different extreme weather events as claimed in claim 3, characterized in that: The step of inputting the first new energy output and the first load into a preset deep belief prediction network, so that the deep belief prediction network generates a new energy predicted output annual power curve and a predicted load annual power curve according to the first new energy output and the first load, includes: Inputting the first new energy output and the first load into a preset deep belief prediction network, so that the deep belief prediction network generates a predicted new energy output corresponding to each time period and a predicted load corresponding to each time period according to the first new energy output and the first load; Generate an annual power curve of predicted output of new energy according to the predicted output of new energy corresponding to each time period; According to the predicted load corresponding to each time period, a predicted load annual power curve is generated.

5. The method for generating annual operation scenarios of a power system considering different extreme weather events as claimed in claim 4, characterized in that: The step of correcting the predicted load annual power curve according to the abnormal load curve to generate a final predicted load annual power curve includes: Obtain the typical frequency of various extreme weather events in the power system in each month; Extracting data from each period of the abnormal load curve to generate abnormal load sequences under various types of extreme weather events; For each type of extreme weather event, the predicted load annual power curve is revised according to the typical frequency of extreme weather events of the current type occurring in each month and the corresponding abnormal load sequence; wherein, during each revision, the time period data with the smallest value in the abnormal load sequence is selected to replace a random time period data of the corresponding month in the predicted load annual power curve, until the number of replacements reaches the typical frequency of extreme weather events of the current type occurring in the corresponding month.

6. The method for generating annual operation scenarios of a power system considering different extreme weather events as claimed in claim 5, characterized in that: The typical frequency of various extreme weather events occurring in the power system in each month is obtained, including: Obtain the frequency and monthly distribution of various types of extreme weather events in the power system in recent years; The frequency of each type of extreme weather event in recent years is averaged to generate the annual average frequency of each type of extreme weather event; Based on the monthly distribution of each type of extreme weather event in recent years and the frequency of each type of extreme weather event in recent years, calculate the monthly proportion of each type of extreme weather event; The monthly proportions and the annual average frequencies are calculated to generate typical frequencies of various extreme weather events occurring in each month.

7. The method for generating annual operation scenarios of a power system considering different extreme weather events as claimed in claim 6, characterized in that: The generating of the annual operation scenario of the power system according to the final predicted annual power curve of new energy output and the final predicted annual power curve of load includes: Match the final predicted annual power curve of new energy output and the final predicted annual power curve of load in each period to generate the planned operation data for each period; Calculate power flow calculations based on the planned operation data and generate power flow calculation results; The planned operation data whose power flow calculation results do not satisfy the preset operation constraints are eliminated to generate the annual operation scenario of the power system.

8. A device for generating annual operation scenarios of a power system taking into account different extreme weather events, characterized in that: include: Historical operation data acquisition module, annual power curve generation module, extreme weather data acquisition module, extreme weather power curve generation module, annual power curve correction module and annual operation scenario generation module; The historical operation data acquisition module is used to obtain the first new energy output and the first load of each period of a typical historical year of the power system; The annual power curve generation module is used to input the first new energy output and the first load into a preset deep belief prediction network, so that the deep belief prediction network generates a new energy predicted output annual power curve and a predicted load annual power curve according to the first new energy output and the first load; The extreme weather data acquisition module is used to extract the first new energy output of each period under various extreme weather events from the first new energy output of each period in a typical historical year as the second new energy output; Extract the first load of each period under various extreme weather events from the first load of each period in historical typical years as the second load; The extreme weather power curve generation module is used to input the second new energy output and the second load into a preset universal output model, so that the universal output model fits the abnormal new energy output curve and the abnormal load curve under various types of extreme weather events according to the second new energy output and the second load; The annual power curve correction module is used to correct the new energy predicted output annual power curve according to the abnormal new energy output curve to generate a final new energy predicted output annual power curve; Correcting the predicted load annual power curve according to the abnormal load curve to generate a final predicted load annual power curve; The annual operation scenario generation module is used to generate the annual operation scenario of the power system according to the final new energy predicted output annual power curve and the final predicted load annual power curve.

9. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for generating annual operation scenarios of an electric power system taking into account different extreme weather events as described in any one of claims 1 to 7 can be implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it can implement the method for generating annual operation scenarios of a power system taking into account different extreme weather events as described in any one of claims 1 to 7.

Images