A method and system for dispatching electric power resources based on electric power demand

Abstract

The present invention relates to the field of power dispatching, and specifically to a method and system for dispatching power resources based on power demand. A power resource dispatching system based on power demand, comprising: an impact degree analysis module, a power consumption prediction module, a power generation prediction module and a power resource dispatching module. The present invention comprehensively considers the impact of multiple factors such as temperature, humidity, rainfall, date type and time period on the power consumption of the load center in the commercial area, and constructs a power demand prediction model based on these factors, which can more accurately predict the power consumption of the load center in the next preset time period; this multi-factor comprehensive prediction method breaks through the limitations of traditional single factor or simple historical data prediction, helps to improve the operating efficiency and reliability of the power system, and provides more accurate decision-making basis for power market participants.

Description

Technical Field

[0001] The present invention relates to the field of electric power dispatching, and in particular to a method and system for dispatching electric power resources based on electricity demand. Background Art

[0002] The electricity demand in commercial areas is affected by many factors, including temperature, humidity, rainfall, date type, and time period; these factors interact with each other and jointly determine the electricity consumption of commercial areas at a specific time; for example, hot weather may lead to a significant increase in air-conditioning load, while rainfall may reduce outdoor commercial activities, thereby affecting electricity demand; however, existing power resource scheduling methods are usually unable to accurately quantify the impact of these factors on electricity demand, and thus cannot provide effective support for the rational scheduling of power resources.

[0003] In addition, with the gradual popularization of distributed energy, how to effectively integrate these renewable energy sources into power resource scheduling to improve energy utilization efficiency and reduce dependence on traditional energy has also become an important challenge facing the power industry; although distributed energy has many advantages, the intermittent and uncertain output, as well as coordination issues with traditional power grids, all need to be solved through innovative scheduling methods.

[0004] The present invention proposes a method and system for dispatching electric power resources based on electricity demand, aiming to achieve optimal allocation of electric power resources and meet the electricity demand in complex environments such as commercial areas through an innovative electricity demand prediction model and dispatching strategy. Summary of the invention

[0005] The present invention comprehensively considers the impact of multiple factors such as temperature, humidity, rainfall, date type and time period on the electricity consumption of the load center in the commercial area, and constructs an electricity demand prediction model based on these factors, which can more accurately predict the electricity consumption of the load center in the next preset time period; this multi-factor comprehensive prediction method breaks through the limitations of traditional single factor or simple historical data prediction, and can more comprehensively reflect the complexity and diversity of electricity demand in commercial areas; it helps to reduce the surplus or shortage of electricity supply, improve the operating efficiency and reliability of the power system, and also provides a more accurate decision-making basis for electricity market participants.

[0006] A method for dispatching electric power resources based on electric power demand, comprising:

[0007] For any one of several load centers in the commercial area, respectively calculate the influence of temperature, humidity, rainfall information, date type and time period information on the power consumption of the load center, and calculate and obtain a first weight, a second weight, a third weight, a fourth weight and a fifth weight based on the obtained influence degrees as a weight combination;

[0008] Construct an electricity demand forecasting model. For any load center, at any forecast time point, sort the historical electricity consumption of the load center in the previous T preset time periods in time to form electricity time series data, and obtain the forecast temperature, forecast humidity, forecast rainfall information and time period information for the next preset time period, and obtain the current date type at the same time; use the electricity time series data, forecast temperature, forecast humidity, forecast rainfall information, date type, time period information and weight combination obtained at the current forecast time point as the input of the electricity demand forecasting model, and output the forecast electricity consumption of the load center for the next preset time period;

[0009] Construct a renewable energy output prediction model, obtain the current solar irradiance at any prediction time point for any load center; use the solar irradiance, predicted temperature and predicted humidity obtained at the current prediction time point as inputs to the renewable energy output prediction model, and output the predicted renewable energy power generation of the load center in the next preset time period;

[0010] For any load center, the predicted power consumption and predicted renewable energy generation obtained at the current prediction time point are used to calculate the power required to be dispatched by the load center in the next preset time period;

[0011] The application obtains the required electricity volume of all load centers in the next preset time period and dispatches power resources.

[0012] Preferably, the influence of temperature, humidity, rainfall information, date type and time period information on the power consumption of the load center is calculated respectively, and the specific operations are as follows:

[0013] For any load center, obtain the historical electricity consumption data of the load center for each day in the past n days, and obtain the average temperature, average humidity, rainfall information, date type and time period information of each day; wherein, the rainfall information includes the presence of rainfall and the absence of rainfall. If the rainfall information indicates the presence of rainfall, the rainfall information code is 1, and if the rainfall information indicates the absence of rainfall, the rainfall information code is 0; the date type includes working days and non-working days. If the date type is a working day, the date type code is 1, and if the date type is a non-working day, the date type code is 0;

[0014] The time period information of each day includes the power consumption of each preset time period of the load center in the k preset time periods of the day; the k preset time periods of each day are sequentially set to time period codes 1, 2, ..., k, and the time period codes corresponding to the preset time periods with the highest power consumption of each day are obtained respectively, and the number of time period codes a with the most repetitions among the obtained n time period codes is recorded;

[0015] The influence of average temperature, average humidity, rainfall information, date type and time period information on the current load center power consumption is calculated respectively R(p), p = 1, 2, 3, 4, 5; R(1) to R(5) represent the influence of average temperature, average humidity, rainfall information, date type and time period information respectively;

[0016] The calculation methods of R(1) to R(4) are as follows:

[0017] Using the formula Calculate the influence of average temperature, average humidity, rainfall information and date type on the current load center power consumption R(p); x(1) i to x(4) i Represents the average temperature, average humidity, rainfall information code and date type code of the i-th day in sequence; i = 1, 2, ..., n; arrive It represents the mean of the average temperature, the mean of the average humidity, the mean of the rainfall information code and the mean of the date type code in the past n days respectively; i represents the current load center's electricity consumption data for the i-th day, It represents the average power consumption of the current load center in the past n days;

[0018] R(5) is calculated as follows:

[0019] Using the formula Calculate the influence of the acquired time period information on the load center R(5).

[0020] Preferably, the first weight, the second weight, the third weight, the fourth weight and the fifth weight are calculated based on the acquired influence levels, and the specific operation is as follows:

[0021] For any load center, based on the influence of temperature, humidity, rainfall information, date type and time period information on the power consumption of the load center, the formula The first weight ω(1), the second weight ω(2), the third weight ω(3), the fourth weight ω(4) and the fifth weight ω(5) are calculated and obtained respectively.

[0022] Preferably, the electricity demand forecasting model is established based on LSTM and FCNN, including an input layer, an LSTM layer, a non-temporal feature extraction layer, a first fusion layer, a second fusion layer, a sixth fully connected layer and an output layer;

[0023] The input layer is used to input the power consumption time series data, predicted temperature, predicted humidity, predicted rainfall information, date type, time period information and weight combination obtained by any load center at the current prediction time point;

[0024] The LSTM layer is used to receive and process the electricity consumption time series data and extract the time series feature vector;

[0025] The non-time series feature extraction layer includes five parallel first fully connected layers, second fully connected layers, third fully connected layers, fourth fully connected layers and fifth fully connected layers; the first fully connected layer is used to receive predicted temperature data, extract features of the predicted temperature data, and obtain a first feature vector; the second fully connected layer is used to receive predicted humidity data, extract features of the predicted humidity data, and obtain a second feature vector; the third fully connected layer is used to receive predicted rainfall information, extract features of the predicted rainfall information, and obtain a third feature vector; the fourth fully connected layer is used to receive date types, extract features of date types, and obtain a fourth feature vector; the fifth fully connected layer is used to receive time period information, extract features of the time period information, and obtain a fifth feature vector;

[0026] The first fusion layer is used to perform weighted fusion on the first eigenvector, the second eigenvector, the third eigenvector, the fourth eigenvector and the fifth eigenvector output by the non-time series feature extraction layer by using the weight combination of the input power demand prediction model to obtain the non-time series feature vector, wherein the first weight, the second weight, the third weight, the fourth weight and the fifth weight correspond to the first eigenvector, the second eigenvector, the third eigenvector, the fourth eigenvector and the fifth eigenvector respectively;

[0027] The second fusion layer is used to concatenate the time series feature vector and the non-time series feature vector to obtain a comprehensive feature vector;

[0028] The sixth fully connected layer is used to further extract features from the comprehensive feature vector and learn more advanced feature representations;

[0029] The output layer is used to generate the predicted electricity consumption for the next preset time period.

[0030] Preferably, the specific operations for training the electricity demand prediction model are as follows:

[0031] Obtain a number of demand forecasting training samples, each of which contains the historical electricity consumption data of any load center in T+1 preset time periods sorted by time, the influencing factors corresponding to the last preset time period, and the weight combination of the influencing factors; divide all demand forecasting training samples into a demand forecasting training set and a demand forecasting verification set;

[0032] The electricity consumption data of the first T preset time periods in each demand forecast training set, the influencing factors corresponding to the last preset time period and the weight combination of the influencing factors are used as the input of the electricity demand forecast model, and the electricity consumption data of the last preset time period is used as the output of the electricity demand forecast model to train the electricity demand forecast model; then the demand forecast verification set is used to verify the electricity demand forecast model to obtain a first verification result; a first training condition is set to determine whether the first verification result meets the first training condition, and if so, the trained electricity demand forecast model is output; if not, the training set is continued to be used to train the electricity demand forecast model.

[0033] Preferably, the amount of electricity to be dispatched is calculated, and the specific operations are as follows:

[0034] For any load center, the predicted power consumption Y obtained at the current prediction time point is used forecast and forecast renewable energy generation F forecast , using the formula Y actual =Y forecast (1+g)-E forecast Calculate and obtain the amount of electricity Y that needs to be dispatched by the current load center in the next preset time period actual , where g is the preset safety factor.

[0035] Preferably, the renewable energy output prediction model is established based on FCNN, including a second input layer, a hidden layer, and a second output layer; the second input layer is used to receive the solar irradiance, predicted temperature, and predicted humidity obtained by any load center at the current prediction time point; the hidden layer is used to extract the features of the solar irradiance, predicted temperature, and predicted humidity; the second output layer is used to output the predicted renewable energy power generation of the current load center in the next preset time period;

[0036] The specific operations for training the renewable energy prediction model are as follows:

[0037] Acquire a number of energy prediction training samples, each of which contains the solar irradiance, temperature, humidity and renewable energy power generation of any load center in a preset time period; divide all acquired energy prediction training samples into an energy prediction training set and an energy prediction verification set; train a renewable energy prediction model using the energy prediction training set, and verify the renewable energy prediction model using the energy prediction verification set to obtain a second verification result; set a second training condition, and determine whether the second verification result meets the second training condition; if so, output the trained renewable energy prediction model; if not, continue to train the renewable energy prediction model using the energy prediction training set.

[0038] A power resource dispatching system based on power demand, comprising:

[0039] The impact degree analysis module includes an impact degree calculation unit and a weight combination acquisition unit; the impact degree calculation unit is used to calculate the impact degree of temperature, humidity, rainfall information, date type and time period information on the power consumption of each load center; the weight combination acquisition unit is used to calculate and acquire the first weight, the second weight, the third weight, the fourth weight and the fifth weight as the weight combination by using the acquired impact degrees;

[0040] The power consumption prediction module includes a data acquisition unit and a power demand prediction unit. The data acquisition unit is used to sort the historical power consumption of the load center in the previous T preset time periods in time to form power consumption time series data at any prediction time point, and obtain the predicted temperature, predicted humidity, predicted rainfall information and time period information of the next preset time period, and obtain the current date type at the same time; the power demand prediction unit is used to use the power consumption time series data, predicted temperature, predicted humidity, predicted rainfall information, date type, time period information and weight combination obtained at the current prediction time point as the input of the power demand prediction model, and output the predicted power consumption of the load center in the next preset time period;

[0041] The power generation prediction module is used to obtain the current solar irradiance at any prediction time point; the solar irradiance, predicted temperature and predicted humidity obtained at the current prediction time point are used as inputs of the renewable energy output prediction model, and the predicted renewable energy power generation of the load center in the next preset time period is output;

[0042] The power resource scheduling module is used to use the predicted power consumption and predicted renewable energy power generation obtained at the current prediction time point to calculate the power that needs to be scheduled at the load center in the next preset time period, and apply the acquired power that needs to be scheduled at all load centers in the next preset time period to schedule power resources.

[0043] The present invention has the following advantages:

[0044] The present invention comprehensively considers the impact of multiple factors such as temperature, humidity, rainfall, date type and time period on the electricity consumption of the load center in the commercial area, and constructs an electricity demand prediction model based on these factors, which can more accurately predict the electricity consumption of the load center in the next preset time period; this multi-factor comprehensive prediction method breaks through the limitations of traditional single factor or simple historical data prediction, and can more comprehensively reflect the complexity and diversity of electricity demand in commercial areas; it helps to reduce the surplus or shortage of electricity supply, improve the operating efficiency and reliability of the power system, and also provides a more accurate decision-making basis for electricity market participants.

[0045] The present invention realizes accurate dispatch of electric power resources by calculating the amount of electricity that the load center needs to dispatch in the next preset time period; using the predicted electricity consumption and renewable energy power generation, the system can automatically calculate the amount of electricity that each load center needs to dispatch, thereby realizing optimal allocation of electric power resources; this prediction-based dispatching method can effectively reduce the waste of electric power resources, improve the operating efficiency of the power grid, and reduce the operating cost of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a schematic diagram of the structure of a power resource scheduling system based on power demand adopted in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to enable persons skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0048] Embodiment 1, a method for dispatching electric power resources based on electric power demand, comprising:

[0049] For any of the several load centers in the commercial area, the influence of temperature, humidity, rainfall information, date type and time period information on the power consumption of the load center is calculated respectively, and the first weight, second weight, third weight, fourth weight and fifth weight are calculated based on the obtained influence degrees as the weight combination; among them, temperature is one of the important factors affecting the power consumption of the commercial area; in high or low temperature weather, the use of air conditioning and heating equipment will significantly increase the power consumption; for example, in high temperature weather in summer, the frequency and power of air conditioning use will increase, resulting in an increase in power consumption; humidity will also indirectly affect the use of air conditioning and heating equipment; in a high humidity environment, people may use air conditioning more frequently to reduce humidity, thereby Increased electricity consumption; Rainfall may reduce outdoor commercial activities, such as the flow of people in shopping malls may decrease on rainy days, resulting in a decrease in electricity consumption; Date types include weekdays and non-working days; weekdays are usually busier with higher business activities and higher electricity consumption; non-working days such as weekends and holidays have relatively fewer business activities and lower electricity consumption; Electricity consumption in commercial areas will also vary at different times of the day; for example, electricity consumption is higher during business hours in the morning and afternoon, while it is lower at night; Through weight combination, it can be ensured that more attention is paid to factors with greater influence when predicting electricity consumption in the future; This helps to improve the accuracy and reliability of the forecast, allowing power companies to make power supply plans in advance and avoid oversupply or shortage of electricity;

[0050] Construct an electricity demand forecasting model. For any load center, at any forecast time point, sort the historical electricity consumption of the load center in the previous T preset time periods in time to form electricity time series data, and obtain the predicted temperature, predicted humidity, predicted rainfall information and time period information for the next preset time period, and obtain the current date type at the same time. The electricity time series data is one of the important inputs of the model. By collecting the historical electricity consumption of the load center in the previous T preset time periods and arranging them in chronological order, the changing trend and periodic characteristics of electricity consumption can be captured; and these environmental factors and time period information are the key factors affecting electricity consumption. By obtaining these forecast data, the model can consider the impact of these factors on electricity consumption, thereby generating more accurate forecast results; the electricity time series data, predicted temperature, predicted humidity, predicted rainfall information, date type, time period information and weight combination obtained at the current forecast time point are used as the input of the electricity demand forecasting model, and the predicted electricity consumption of the load center in the next preset time period is output; this forecast result is generated based on the input data and model algorithm, aiming to provide an accurate basis for power resource scheduling;

[0051] Construct a renewable energy output prediction model, obtain the current solar irradiance at any prediction time point for any load center; use the solar irradiance, predicted temperature and predicted humidity obtained at the current prediction time point as the input of the renewable energy output prediction model, and output the predicted renewable energy power generation of the load center in the next preset time period; by considering key environmental factors such as current solar irradiance, predicted temperature and humidity, the model provides a scientific basis for power resource scheduling, enabling power companies to more effectively integrate renewable energy, optimize power supply, improve energy utilization efficiency, reduce dependence on traditional energy, and promote sustainable development of energy;

[0052] For any load center, the predicted power consumption and predicted renewable energy generation obtained at the current prediction time point are used to calculate the power required to be dispatched by the load center in the next preset time period;

[0053] The application obtains the required electricity volume of all load centers in the next preset time period and dispatches power resources.

[0054] Calculate the influence of temperature, humidity, rainfall information, date type and time period information on the power consumption of the load center respectively. The specific operations are as follows:

[0055] For any load center, obtain the historical electricity consumption data of the load center for each day in the past n days, and obtain the average temperature, average humidity, rainfall information, date type and time period information of each day; wherein, the rainfall information includes the presence of rainfall and the absence of rainfall. If the rainfall information indicates the presence of rainfall, the rainfall information code is 1, and if the rainfall information indicates the absence of rainfall, the rainfall information code is 0; the date type includes working days and non-working days. If the date type is a working day, the date type code is 1, and if the date type is a non-working day, the date type code is 0;

[0056] The time period information of each day includes the power consumption of each preset time period of the load center in the k preset time periods of the day; the k preset time periods of each day are sequentially set to time period codes 1, 2, ..., k, and the time period codes corresponding to the preset time periods with the highest power consumption of each day are obtained respectively, and the number of time period codes a with the most repetitions among the obtained n time period codes is recorded;

[0057] The influence of average temperature, average humidity, rainfall information, date type and time period information on the current load center power consumption is calculated respectively R(p), p = 1, 2, 3, 4, 5; R(1) to R(5) represent the influence of average temperature, average humidity, rainfall information, date type and time period information respectively;

[0058] The calculation methods of R(1) to R(4) are as follows:

[0059] Using the formula Calculate the influence of average temperature, average humidity, rainfall information and date type on the current load center power consumption R(p); x(1) i to x(4) i Represents the average temperature, average humidity, rainfall information code and date type code of the i-th day in sequence; i = 1, 2, ..., n; arrive It represents the mean of the average temperature, the mean of the average humidity, the mean of the rainfall information code and the mean of the date type code in the past n days respectively; i represents the current load center's electricity consumption data for the i-th day, It represents the average power consumption of the current load center in the past n days;

[0060] R(5) is calculated as follows:

[0061] Using the formula Calculate the influence of the acquired time period information on the load center R(5).

[0062] The first weight, the second weight, the third weight, the fourth weight and the fifth weight are calculated based on the obtained influence levels. The specific operation is as follows:

[0063] For any load center, based on the influence of temperature, humidity, rainfall information, date type and time period information on the power consumption of the load center, the formula The first weight ω(1), the second weight ω(2), the third weight ω(3), the fourth weight ω(4) and the fifth weight ω(5) are calculated and obtained respectively.

[0064] The electricity demand forecasting model is built based on LSTM and FCNN, including input layer, LSTM layer, non-temporal feature extraction layer, first fusion layer, second fusion layer, sixth fully connected layer and output layer;

[0065] The input layer is used to input the power consumption time series data, predicted temperature, predicted humidity, predicted rainfall information, date type, time period information and weight combination obtained by any load center at the current prediction time point;

[0066] The LSTM layer is used to receive and process the electricity consumption time series data and extract the time series feature vector;

[0067] The non-time series feature extraction layer includes five parallel first fully connected layers, second fully connected layers, third fully connected layers, fourth fully connected layers and fifth fully connected layers; the first fully connected layer is used to receive predicted temperature data, extract features of the predicted temperature data, and obtain a first feature vector; the second fully connected layer is used to receive predicted humidity data, extract features of the predicted humidity data, and obtain a second feature vector; the third fully connected layer is used to receive predicted rainfall information, extract features of the predicted rainfall information, and obtain a third feature vector; the fourth fully connected layer is used to receive date types, extract features of date types, and obtain a fourth feature vector; the fifth fully connected layer is used to receive time period information, extract features of the time period information, and obtain a fifth feature vector;

[0068] The first fusion layer is used to perform weighted fusion on the first eigenvector, the second eigenvector, the third eigenvector, the fourth eigenvector and the fifth eigenvector output by the non-time series feature extraction layer by using the weight combination of the input power demand prediction model to obtain the non-time series feature vector, wherein the first weight, the second weight, the third weight, the fourth weight and the fifth weight correspond to the first eigenvector, the second eigenvector, the third eigenvector, the fourth eigenvector and the fifth eigenvector respectively;

[0069] The second fusion layer is used to concatenate the time series feature vector and the non-time series feature vector to obtain a comprehensive feature vector;

[0070] The sixth fully connected layer is used to further extract features from the comprehensive feature vector and learn more advanced feature representations;

[0071] The output layer is used to generate the predicted power consumption for the next preset time period;

[0072] Based on the long short-term memory network and the fully connected neural network, it can make full use of the powerful ability of LSTM in processing time series data, effectively capture the long-term dependence and short-term fluctuation of electricity consumption over time, and at the same time FCNN can further extract and integrate non-time series features, such as temperature, humidity, rainfall information, date type and time period information. Through this combination, the model can more comprehensively and accurately predict the electricity demand of commercial load centers, and provide solid data support for the precise scheduling of power resources, thereby improving the operating efficiency and reliability of the power system and reducing energy costs. At the same time, it also helps to better integrate distributed energy and promote the efficient use of renewable energy.

[0073] The specific operations for training the electricity demand prediction model are as follows:

[0074] Obtain a number of demand forecasting training samples, each of which contains the historical electricity consumption data of any load center in T+1 preset time periods sorted by time, the influencing factors corresponding to the last preset time period, and the weight combination of the influencing factors; divide all demand forecasting training samples into a demand forecasting training set and a demand forecasting verification set;

[0075] The electricity consumption data of the first T preset time periods in each demand forecast training set, the influencing factors corresponding to the last preset time period and the weight combination of the influencing factors are used as the input of the electricity demand forecast model, and the electricity consumption data of the last preset time period is used as the output of the electricity demand forecast model to train the electricity demand forecast model; then the demand forecast verification set is used to verify the electricity demand forecast model to obtain a first verification result; a first training condition is set to determine whether the first verification result meets the first training condition, and if so, the trained electricity demand forecast model is output; if not, the training set is continued to be used to train the electricity demand forecast model.

[0076] Calculate the amount of electricity to be dispatched. The specific operations are as follows:

[0077] For any load center, the predicted power consumption Y obtained at the current prediction time point is used forecast and forecast renewable energy generation E forecast , using the formula Y actual =Y forecast (1+g)-E forecast Calculate and obtain the amount of electricity Y that needs to be dispatched by the current load center in the next preset time period actual , where g is the preset safety factor.

[0078] The renewable energy output prediction model is established based on FCNN, including the second input layer, the hidden layer and the second output layer; the second input layer is used to receive the solar irradiance, predicted temperature and predicted humidity obtained by any load center at the current prediction time point; the hidden layer is used to extract the features of solar irradiance, predicted temperature and predicted humidity; the second output layer is used to output the predicted renewable energy power generation of the current load center in the next preset time period;

[0079] The specific operations for training the renewable energy prediction model are as follows:

[0080] Acquire a number of energy prediction training samples, each of which contains the solar irradiance, temperature, humidity and renewable energy power generation of any load center in a preset time period; divide all acquired energy prediction training samples into an energy prediction training set and an energy prediction verification set; train a renewable energy prediction model using the energy prediction training set, and verify the renewable energy prediction model using the energy prediction verification set to obtain a second verification result; set a second training condition, and determine whether the second verification result meets the second training condition; if so, output the trained renewable energy prediction model; if not, continue to train the renewable energy prediction model using the energy prediction training set.

[0081] Embodiment 2, a power resource dispatching system based on power demand, such as Figure 1 As shown, including:

[0082] The impact degree analysis module includes an impact degree calculation unit and a weight combination acquisition unit; the impact degree calculation unit is used to calculate the impact degree of temperature, humidity, rainfall information, date type and time period information on the power consumption of each load center; the weight combination acquisition unit is used to calculate and acquire the first weight, the second weight, the third weight, the fourth weight and the fifth weight as the weight combination by using the acquired impact degrees;

[0083] The power consumption prediction module includes a data acquisition unit and a power demand prediction unit. The data acquisition unit is used to sort the historical power consumption of the load center in the previous T preset time periods in time to form power consumption time series data at any prediction time point, and obtain the predicted temperature, predicted humidity, predicted rainfall information and time period information of the next preset time period, and obtain the current date type at the same time; the power demand prediction unit is used to use the power consumption time series data, predicted temperature, predicted humidity, predicted rainfall information, date type, time period information and weight combination obtained at the current prediction time point as the input of the power demand prediction model, and output the predicted power consumption of the load center in the next preset time period;

[0084] The power generation prediction module is used to obtain the current solar irradiance at any prediction time point; the solar irradiance, predicted temperature and predicted humidity obtained at the current prediction time point are used as inputs of the renewable energy output prediction model, and the predicted renewable energy power generation of the load center in the next preset time period is output;

[0085] The power resource scheduling module is used to use the predicted power consumption and predicted renewable energy power generation obtained at the current prediction time point to calculate the power that needs to be scheduled at the load center in the next preset time period, and apply the acquired power that needs to be scheduled at all load centers in the next preset time period to schedule power resources.

[0086] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A method for dispatching electric power resources based on electric power demand, characterized in that: include: For any one of several load centers in the commercial area, respectively calculate the influence of temperature, humidity, rainfall information, date type and time period information on the power consumption of the load center, and calculate and obtain a first weight, a second weight, a third weight, a fourth weight and a fifth weight based on the obtained influence degrees as a weight combination; Construct an electricity demand forecasting model. For any load center, at any forecast time point, sort the historical electricity consumption of the load center in the previous T preset time periods in time to form electricity time series data, and obtain the forecast temperature, forecast humidity, forecast rainfall information and time period information for the next preset time period, and obtain the current date type at the same time; use the electricity time series data, forecast temperature, forecast humidity, forecast rainfall information, date type, time period information and weight combination obtained at the current forecast time point as the input of the electricity demand forecasting model, and output the forecast electricity consumption of the load center for the next preset time period; Construct a renewable energy output prediction model, obtain the current solar irradiance at any prediction time point for any load center; use the solar irradiance, predicted temperature and predicted humidity obtained at the current prediction time point as inputs to the renewable energy output prediction model, and output the predicted renewable energy power generation of the load center in the next preset time period; For any load center, the predicted power consumption and predicted renewable energy generation obtained at the current prediction time point are used to calculate the power required to be dispatched by the load center in the next preset time period; The application obtains the required electricity of all load centers in the next preset time period and dispatches power resources; The electricity demand forecasting model is built based on LSTM and FCNN, including input layer, LSTM layer, non-temporal feature extraction layer, first fusion layer, second fusion layer, sixth fully connected layer and output layer; The input layer is used to input the power consumption time series data, predicted temperature, predicted humidity, predicted rainfall information, date type, time period information and weight combination obtained by any load center at the current prediction time point; The LSTM layer is used to receive and process the electricity consumption time series data and extract the time series feature vector; The non-time series feature extraction layer includes five parallel first fully connected layers, second fully connected layers, third fully connected layers, fourth fully connected layers and fifth fully connected layers; the first fully connected layer is used to receive predicted temperature data, extract features of the predicted temperature data, and obtain a first feature vector; the second fully connected layer is used to receive predicted humidity data, extract features of the predicted humidity data, and obtain a second feature vector; the third fully connected layer is used to receive predicted rainfall information, extract features of the predicted rainfall information, and obtain a third feature vector; the fourth fully connected layer is used to receive date types, extract features of date types, and obtain a fourth feature vector; the fifth fully connected layer is used to receive time period information, extract features of the time period information, and obtain a fifth feature vector; The first fusion layer is used to perform weighted fusion on the first eigenvector, the second eigenvector, the third eigenvector, the fourth eigenvector and the fifth eigenvector output by the non-time series feature extraction layer by using the weight combination of the input power demand prediction model to obtain the non-time series feature vector, wherein the first weight, the second weight, the third weight, the fourth weight and the fifth weight correspond to the first eigenvector, the second eigenvector, the third eigenvector, the fourth eigenvector and the fifth eigenvector respectively; The second fusion layer is used to concatenate the time series feature vector and the non-time series feature vector to obtain a comprehensive feature vector; The sixth fully connected layer is used to further extract features from the comprehensive feature vector and learn more advanced feature representations; The output layer is used to generate the predicted electricity consumption for the next preset time period.

2. A method for dispatching electric power resources based on electric power demand according to claim 1, characterized in that: Calculate the influence of temperature, humidity, rainfall information, date type and time period information on the power consumption of the load center respectively. The specific operations are as follows: For any load center, obtain the historical electricity consumption data of the load center for each day in the past n days, and obtain the average temperature, average humidity, rainfall information, date type and time period information of each day; wherein, the rainfall information includes the presence of rainfall and the absence of rainfall. If the rainfall information indicates the presence of rainfall, the rainfall information code is 1, and if the rainfall information indicates the absence of rainfall, the rainfall information code is 0; the date type includes working days and non-working days. If the date type is a working day, the date type code is 1, and if the date type is a non-working day, the date type code is 0; The time period information of each day includes the power consumption of each preset time period of the load center in the k preset time periods of the day; the k preset time periods of each day are sequentially set to time period codes 1, 2, ..., k, and the time period codes corresponding to the preset time periods with the highest power consumption of each day are obtained respectively, and the number of time period codes a with the most repetitions among the obtained n time period codes is recorded; The influence of average temperature, average humidity, rainfall information, date type and time period information on the current load center power consumption is calculated respectively R(p), p = 1, 2, 3, 4, 5; R(1) to R(5) represent the influence of average temperature, average humidity, rainfall information, date type and time period information respectively; The calculation methods of R(1) to R(4) are as follows: Using the formula Calculate the influence of average temperature, average humidity, rainfall information and date type on the current load center power consumption R(p); x(1) i to x(4) i Represents the average temperature, average humidity, rainfall information code and date type code of the i-th day in sequence; i = 1, 2, ..., n; arrive It represents the mean of the average temperature, the mean of the average humidity, the mean of the rainfall information code and the mean of the date type code in the past n days respectively; i represents the current load center's electricity consumption data for the i-th day, It represents the average power consumption of the current load center in the past n days; R(5) is calculated as follows: Using the formula Calculate the influence of the acquired time period information on the load center R(5).

3. The method for dispatching electric power resources based on electric power demand according to claim 2, characterized in that: The first weight, the second weight, the third weight, the fourth weight and the fifth weight are calculated based on the obtained influence levels. The specific operation is as follows: For any load center, based on the influence of temperature, humidity, rainfall information, date type and time period information on the power consumption of the load center, the formula The first weight ω(1), the second weight ω(2), the third weight ω(3), the fourth weight ω(4) and the fifth weight ω(5) are calculated and obtained respectively.

4. The method for dispatching electric power resources based on electric power demand according to claim 3, characterized in that: The specific operations for training the electricity demand prediction model are as follows: Obtain several demand forecasting training samples, each of which contains the historical electricity consumption data of any load center in T+1 preset time periods sorted by time, the influencing factors corresponding to the last preset time period, and the weight combination of the influencing factors; divide all demand forecasting training samples into a demand forecasting training set and a demand forecasting verification set, The power consumption data of the first T preset time periods in each demand forecast training set, the influencing factors corresponding to the last preset time period, and the weight combination of the influencing factors are used as the input of the power demand forecast model, and the power consumption data of the last preset time period is used as the output of the power demand forecast model to train the power demand forecast model; then the power demand forecast model is verified by using the demand forecast verification set to obtain the first verification result; A first training condition is set to determine whether the first verification result satisfies the first training condition. If so, the trained electricity demand prediction model is output; if not, the training set is continued to be used to train the electricity demand prediction model.

5. The method for dispatching electric power resources based on electric power demand according to claim 4, characterized in that: Calculate the amount of electricity to be dispatched. The specific operations are as follows: For any load center, the predicted power consumption Y obtained at the current prediction time point is used forecast and forecast renewable energy generation E forecast , using the formula Y actual =Y forecast (1+g)-E forecast Calculate and obtain the amount of electricity Y that needs to be dispatched by the current load center in the next preset time period actual , where g is the preset safety factor.

6. A method for dispatching electric power resources based on electric power demand according to claim 5, characterized in that: The renewable energy output prediction model is established based on FCNN, including the second input layer, the hidden layer and the second output layer; the second input layer is used to receive the solar irradiance, predicted temperature and predicted humidity obtained by any load center at the current prediction time point; the hidden layer is used to extract the features of solar irradiance, predicted temperature and predicted humidity; the second output layer is used to output the predicted renewable energy power generation of the current load center in the next preset time period; The specific operations for training the renewable energy prediction model are as follows: Acquire a number of energy prediction training samples, each of which contains the solar irradiance, temperature, humidity and renewable energy power generation of any load center in a preset time period; divide all acquired energy prediction training samples into an energy prediction training set and an energy prediction verification set; train a renewable energy prediction model using the energy prediction training set, and verify the renewable energy prediction model using the energy prediction verification set to obtain a second verification result; set a second training condition, and determine whether the second verification result meets the second training condition; if so, output the trained renewable energy prediction model; if not, continue to train the renewable energy prediction model using the energy prediction training set.

7. A power resource dispatching system based on power demand, characterized in that: The system is applied to a method for dispatching electric power resources based on electric power demand as described in any one of claims 1 to 6 above, comprising: The impact degree analysis module includes an impact degree calculation unit and a weight combination acquisition unit; the impact degree calculation unit is used to calculate the impact degree of temperature, humidity, rainfall information, date type and time period information on the power consumption of each load center; the weight combination acquisition unit is used to calculate and acquire the first weight, the second weight, the third weight, the fourth weight and the fifth weight as the weight combination by using the acquired impact degrees; The power consumption prediction module includes a data acquisition unit and a power demand prediction unit. The data acquisition unit is used to sort the historical power consumption of the load center in the previous T preset time periods in time to form power consumption time series data at any prediction time point, and obtain the predicted temperature, predicted humidity, predicted rainfall information and time period information of the next preset time period, and obtain the current date type at the same time; the power demand prediction unit is used to use the power consumption time series data, predicted temperature, predicted humidity, predicted rainfall information, date type, time period information and weight combination obtained at the current prediction time point as the input of the power demand prediction model, and output the predicted power consumption of the load center in the next preset time period; The power generation prediction module is used to obtain the current solar irradiance at any prediction time point; the solar irradiance, predicted temperature and predicted humidity obtained at the current prediction time point are used as inputs of the renewable energy output prediction model, and the predicted renewable energy power generation of the load center in the next preset time period is output; The power resource scheduling module is used to use the predicted power consumption and predicted renewable energy power generation obtained at the current prediction time point to calculate the power that needs to be scheduled at the load center in the next preset time period, and apply the obtained power that needs to be scheduled for all load centers in the next preset time period to schedule power resources.

Images