Load prediction method for high-proportion new energy access power distribution network

Abstract

The invention relates to the technical field of power distribution network load prediction, and particularly discloses a load prediction method for a high-proportion new energy access power distribution network, and the method comprises the steps: building an omnibearing multi-source data collection system through the technology of the electric power Internet of Things, collecting multi-source data, processing the collected multi-source data, and constructing a power distribution network load prediction hybrid model. And outputting initial load prediction characteristics of the power distribution network, collecting uncertainty quantization parameters of load prediction of the power distribution network, obtaining uncertainty quantization results of load prediction of the power distribution network, and matching optimization load prediction results of the power distribution network based on the uncertainty quantization results of load prediction of the power distribution network and the initial load prediction characteristics of the power distribution network. According to the method, the problems that an existing load prediction model possibly cannot accurately describe the complex relation between the characteristics and the load of the power distribution network, the model generalization ability is limited, and the prediction precision is insufficient are solved, multi-source cooperative operation is achieved, and the overall operation efficiency and stability of the power distribution network are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network load forecasting, and specifically provides a load forecasting method for a distribution network with a high proportion of new energy access. Background Art

[0002] With the global emphasis on environmental protection and sustainable development, traditional fossil fuels are gradually being replaced by clean energy. The proportion of new energy such as solar energy and wind energy in the energy structure is continuously increasing. In recent years, the proportion of new energy power generation has continued to rise. This transformation of the energy structure requires the distribution network to be able to adapt to the access of a high proportion of new energy. The access of new energy makes the power source structure of the distribution network more complex and diverse, changing from traditional single-source power supply to coordinated power supply of multiple distributed new energies and traditional power sources. Different types of new energy power generation have different output characteristics, which greatly increases the uncertainty on the power source side of the distribution network. The development of smart grids requires the distribution network to have a higher level of intelligence, capable of realizing real-time monitoring, analysis, and prediction of the grid operation status. As an important part of smart grids, load forecasting needs to adapt to the new environment of high proportion of new energy access and provide accurate decision-making basis for the intelligent dispatching and control of the grid.

[0003] Currently, there are still some deficiencies in the research on load forecasting for distribution networks with a high proportion of new energy access. Specifically, new energy power generation has characteristics such as intermittency, volatility, and uncertainty. Existing load forecasting models may be difficult to accurately depict the complex relationship between these characteristics and the distribution network load, with limited model generalization ability and insufficient prediction accuracy. Current models often only consider single or partial uncertainty factors and are difficult to comprehensively quantify and analyze multi-source uncertainties, resulting in limited reliability of the prediction results. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a load forecasting method for a distribution network with a high proportion of new energy access, which can effectively solve the problems involved in the above background art.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A load forecasting method for a distribution network with a high proportion of new energy access, comprising the following steps: Using the power Internet of Things technology to build an all-round multi-source data acquisition system to collect multi-source data, where the multi-source data specifically includes new energy power generation data, distribution network historical load data, and distribution network environmental meteorological data; Processing the collected multi-source data, specifically including processing new energy power generation data, processing distribution network historical load data, and processing distribution network environmental meteorological data; Based on the processed multi-source data, obtaining new energy power generation characteristic data, distribution network historical load characteristic data, and distribution network environmental meteorological characteristic data, constructing a distribution network load forecasting hybrid model, and outputting the initial distribution network load forecasting characteristics; Collecting the uncertainty quantification parameters of the distribution network load forecasting, and based on the collected uncertainty quantification parameters of the distribution network load forecasting, obtaining the uncertainty quantification results of the distribution network load forecasting; Based on the uncertainty quantification results of the distribution network load forecasting and the initial distribution network load forecasting characteristics, constructing an objective optimization model to obtain an objective optimization signal, inputting the objective optimization signal into a trained linear regression model to obtain the distribution network load forecasting target index, and matching the distribution network optimized load forecasting result based on the distribution network load forecasting target index.

[0006] As a further method, the processing of new energy power generation data has the following specific analysis process: The new energy power generation data specifically includes the area S of the photovoltaic module, the light radiation intensity H, the photoelectric conversion efficiency η of the photovoltaic module, the air density ρ, the wind turbine swept area A, the wind speed V, and the wind energy utilization coefficient C of the wind turbine p ;

[0007] Calculate the theoretical power generation power characteristic P of a single photovoltaic module th :

[0008]

[0009] In the formula, e is the natural constant;

[0010] Calculate the theoretical power generation power characteristic P of the wind turbine wind ;

[0011] The theoretical power generation power characteristics of a single photovoltaic module and the wind turbine are recorded as new energy power generation characteristic data.

[0012] As a further method, the processing of the distribution network historical load data has the following specific analysis process: The distribution network historical load data specifically includes the load power value P at the i-th acquisition point in the historical statistical period of the distribution network i , the historical maximum load power P of the distribution network max , and the historical minimum load power P of the distribution network min, where \(i\) is the number of each collection point, \(i = 1, 2, 3, \cdots, n\), and \(n\) is the total number of collection points;

[0013] Obtain the historical average load \(P\) of the distribution network avg :

[0014]

[0015] Calculate the historical load signal \(P\) of the distribution network sin ;

[0016] Record the historical average load and the historical load signal of the distribution network as the historical load characteristic data of the distribution network.

[0017] As a further method, process the environmental meteorological data of the distribution network. The specific analysis process is as follows: The environmental meteorological data of the distribution network specifically includes the relative humidity \(RH\) of the distribution network environment, the temperature \(T\) of the distribution network environment, and the precipitation \(js\) of the distribution network environment;

[0018] Obtain the vapor pressure \(e\) of the distribution network environment sy :

[0019]

[0020] Calculate the humidity signal \(Sg\) of the distribution network environment d :

[0021]

[0022] In the formula, \(e\) is the natural constant;

[0023] Record the vapor pressure of the distribution network environment and the humidity signal of the distribution network environment as the environmental meteorological characteristic data of the distribution network.

[0024] As a further method, construct a hybrid model for distribution network load forecasting and output the initial load forecasting characteristics of the distribution network. The specific analysis process is as follows: Based on the new energy generation characteristic data, the historical load characteristic data of the distribution network, and the environmental meteorological characteristic data of the distribution network, construct a hybrid model for distribution network load forecasting to obtain the initial load forecasting characteristics of the distribution network. The initial load forecasting characteristics of the distribution network serve as the analysis basis for constructing the target optimization model;

[0025] The hybrid model for distribution network load forecasting, the specific analysis process is as follows:

[0026]

[0027] In the formula, \(\delta\) is the initial load forecasting characteristic of the distribution network, Energy new is the new energy generation characteristic factor, Load is the historical load characteristic factor of the distribution network, Envir is the environmental meteorological characteristic factor of the distribution network, \(P\) th is the theoretical power generation characteristic of a single photovoltaic module, \(P\)wind is the theoretical power generation characteristic of the wind turbine, P avg is the historical average load of the distribution network, P sin is the historical load signal of the distribution network, e sy is the environmental water vapor pressure of the distribution network, Sg d is the environmental humidity signal of the distribution network, μ1 is the set compensation factor for Energy new The compensation factor of, μ2 is the set compensation factor for Load, μ3 is the set compensation factor for Envir, and e is the natural constant.

[0028] As a further method, collect the quantization parameters of the uncertainty of the distribution network load forecast. Based on the collected quantization parameters of the uncertainty of the distribution network load forecast, obtain the quantization result of the uncertainty of the distribution network load forecast. The specific analysis process is as follows: Collect the quantization parameters of the uncertainty of the distribution network load forecast. The quantization parameters of the uncertainty of the distribution network load forecast specifically include the number of years the distribution network has been in use, the failure frequency of the distribution network equipment, and the area of the distribution network coverage area; Based on the collected quantization parameters of the uncertainty of the distribution network load forecast, comprehensively analyze to obtain the quantization result of the uncertainty of the distribution network load forecast. The quantization result of the uncertainty of the distribution network load forecast is used as the analysis basis for constructing the target optimization model.

[0029] As a further method, for the quantization result of the uncertainty of the distribution network load forecast, the specific analysis process is as follows:

[0030]

[0031] In the formula, Uncer is the quantization result of the uncertainty of the distribution network load forecast, Tn is the number of years the distribution network has been in use, Sgp is the failure frequency of the distribution network equipment, fg is the area of the distribution network coverage area, υ1 is the set compensation factor for Tn, υ2 is the set compensation factor for Sgp, and υ3 is the set compensation factor for fg.

[0032] As a further method, based on the quantization result of the uncertainty of the distribution network load forecast and the initial load forecast characteristics of the distribution network, construct a target optimization model to obtain a target optimization signal, and input the target optimization signal into the trained linear regression model to obtain the target index of the distribution network load forecast. The specific analysis process is as follows: Based on the quantization result of the uncertainty of the distribution network load forecast and the initial load forecast characteristics of the distribution network, construct a target optimization model and output a target optimization signal. The target optimization signal is used as the analysis basis for obtaining the target index of the distribution network load forecast; Input the target optimization signal into the trained linear regression model; The trained linear regression model is expressed as:

[0033] y = g0 + g1 * Myh + b;

[0034] Wherein, y is the target index of the distribution network load forecast, g0 is the intercept, g1 is the first slope, Myh is the target optimization signal, and b is the error term;

[0035] Output the target index of the distribution network load forecast, and the target index of the distribution network load forecast is used as the analysis basis for matching the optimized load forecast result of the distribution network.

[0036] As a further method, for the target optimization model, the specific analysis process is as follows:

[0037] Myh = δ + ln(1 + Uncer);

[0038] Wherein, Myh is the target optimization signal, δ is the initial load forecast feature of the distribution network, and Uncer is the quantification result of the uncertainty of the distribution network load forecast.

[0039] As a further method, based on the target index of the distribution network load forecast to match the optimized load forecast result of the distribution network, the specific analysis process is as follows: Obtain the mapping table of the target index of the distribution network load forecast - the optimized load forecast result of the distribution network pre-stored in the database, and by looking up the mapping table, according to the target index of the distribution network load forecast, find the matching optimized load forecast result of the distribution network.

[0040] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects:

[0041] (1) By providing a load forecasting method for a distribution network with a high proportion of new energy access, the present invention collects all-round multi-source data, including new energy power generation data, distribution network historical load data, distribution network environmental meteorological data, etc., and constructs a distribution network load forecasting hybrid model based on the processed multi-source data, which can comprehensively consider the influence of various factors on the distribution network load, make full use of the information contained in different data, and can comprehensively reflect various states and influencing factors of the distribution network operation, avoiding the limitations of a single data source.

[0042] (2) By constructing a target optimization model and obtaining a target optimization signal, and then inputting it into the trained linear regression model, the present invention can further optimize and adjust the initial load forecast feature, make full use of the advantages of various data and models, obtain a target index of the distribution network load forecast that is more in line with the actual situation, and thus match a more optimized load forecast result of the distribution network.

[0043] (3) By collecting the quantification parameters of the uncertainty of the distribution network load forecast and obtaining the quantification result of the uncertainty of the distribution network load forecast, the present invention can better understand and cope with these uncertainties, which helps to coordinate the relationship between new energy power generation, energy storage devices and traditional power sources, realize multi-source collaborative operation, and improve the overall operation efficiency and stability of the distribution network. Description of the Drawings

[0044] The present invention will be further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the following drawings without creative efforts.

[0045] Figure 1 It is a schematic flow chart of the method of the present invention. Specific embodiments

[0046] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with 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. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0047] Referring to Figure 1 As shown, the present invention provides a load forecasting method for a distribution network with a high proportion of new energy access, including: using the power Internet of Things technology to build an all-round multi-source data acquisition system to collect multi-source data, and the multi-source data specifically includes new energy power generation data, distribution network historical load data, and distribution network environmental meteorological data.

[0048] It is possible to collect various types of data such as new energy power generation data, distribution network historical load data, and distribution network environmental meteorological data. From the power generation side, information such as the area of photovoltaic modules, light radiation intensity, and wind energy utilization coefficient of wind turbines can be obtained to comprehensively grasp the new energy power generation status; on the load side, the load power values of each collection point in the historical statistical period of the distribution network are collected to accurately grasp the load situation; in terms of meteorological environment, data such as relative humidity, temperature, and precipitation are covered to provide support for analyzing the impact of meteorology on the power system and avoid information loss caused by a single data source.

[0049] The power Internet of Things technology can realize the real-time collection and transmission of data with the help of various sensors and communication devices.

[0050] The multi-source data provides rich inputs for the load forecasting model. Combining the new energy power generation data can better consider the impact of new energy output on the load, the historical load data can mine the load change law, and the environmental meteorological data can analyze the correlation between meteorological factors and the load. Different types of data reflect different aspects of the characteristics of the power system, enabling the load forecasting model to adapt to complex and changeable operating environments.

[0051] By collecting new energy power generation data and related environmental data, we can deeply understand the characteristics and laws of new energy power generation. Combining with load data, we can better coordinate new energy power generation and load demand, optimize the distribution and utilization of new energy power in the distribution network, improve the new energy consumption capacity, and promote the development of the energy structure towards cleaner and sustainable directions, such as reasonably arranging the charging and discharging of energy storage devices to balance the fluctuations of new energy power generation and load demand.

[0052] Process the collected multi-source data, specifically including processing new energy power generation data, processing historical load data of the distribution network, and processing environmental meteorological data of the distribution network.

[0053] Process the new energy power generation data. The specific analysis process is as follows: The new energy power generation data specifically includes the area S of the photovoltaic module, the light radiation intensity H, the photoelectric conversion efficiency η of the photovoltaic module, the air density ρ, the wind turbine swept area A, the wind speed V, and the wind energy utilization coefficient C of the wind turbine p ;

[0054] Calculate the theoretical power generation power characteristic P of a single photovoltaic module th :

[0055]

[0056] In the formula, e is the natural constant;

[0057] Calculate the theoretical power generation power characteristic P of the wind turbine wind :

[0058]

[0059] The new energy power generation characteristic data specifically includes the theoretical power generation power characteristic of a single photovoltaic module and the theoretical power generation power characteristic of the wind turbine.

[0060] Calculating the theoretical power generation power characteristics of photovoltaic and wind turbines respectively can accurately reflect the physical characteristics of new energy power generation. For example, the theoretical power generation power of the photovoltaic module is related to the module area, light radiation intensity, and photoelectric conversion efficiency. Through this calculation, the influence of factors such as light conditions on the photovoltaic output can be quantified; the theoretical power generation power of the wind turbine takes into account the air density, wind turbine swept area, wind speed, and wind energy utilization coefficient, etc., and can effectively reflect the role of factors such as wind speed on the wind power output, thus providing more accurate new energy power generation information input for subsequent distribution network load forecasting and improving the forecasting accuracy.

[0061] The new energy power generation characteristic data provides key basic data for constructing a distribution network load forecasting hybrid model, etc. Based on these characteristic data, the model can better explore the internal relationship between new energy power generation and the distribution network load and construct a more accurate load forecasting model.

[0062] Process the historical load data of the distribution network. The specific analysis process is as follows: The historical load data of the distribution network specifically includes the load power value \(P\) of the \(i\)-th collection point in the historical statistical period of the distribution network i , the historical maximum load power \(P\) max , and the historical minimum load power \(P\) min . Here, \(i\) is the number of each collection point, \(i = 1, 2, 3,\cdots, n\), and \(n\) is the total number of collection points;

[0063] Obtain the historical average load \(P\) avg :

[0064]

[0065] Calculate the historical load signal \(P\) sin :

[0066]

[0067] The historical load characteristic data of the distribution network specifically includes the historical average load and the historical load signal of the distribution network.

[0068] Calculating the historical average load of the distribution network can intuitively reflect the overall average level of the distribution network load during the statistical period. Through this indicator, the long-term electricity consumption scale of this area can be understood. For example, when planning the expansion of the power grid or adding new power supply facilities, the future load growth trend and the approximate demand scale can be judged based on the historical average load. The calculation of the historical load signal of the distribution network comprehensively considers the historical maximum and minimum load powers and the average load. It can not only reflect the fluctuation range of the load, but also measure the relative degree of load fluctuation in combination with the average load and the maximum load. This is of great significance for analyzing the change law of the load, determining the peak and valley periods of the load, and formulating reasonable peak shaving strategies.

[0069] The characteristic data such as the historical average load and the historical load signal of the distribution network obtained through processing are important inputs for constructing the load forecasting model. These data contain the historical change information of the load, and the model can learn the change pattern and trend of the load based on this, improving the accuracy of load forecasting.

[0070] Process the environmental meteorological data of the distribution network. The specific analysis process is as follows: The environmental meteorological data of the distribution network specifically includes the relative humidity \(RH\) of the distribution network environment, the temperature \(T\) of the distribution network environment, and the precipitation \(js\) of the distribution network environment;

[0071] Obtain the vapor pressure \(e\) of the distribution network environment sy :

[0072]

[0073] Calculate the humidity signal \(Sg\) of the distribution network environment d :

[0074]

[0075] The specific data of the environmental meteorological characteristics of the distribution network include the water vapor pressure of the distribution network environment and the humidity signal of the distribution network environment.

[0076] Calculating the water vapor pressure and temperature signal of the distribution network environment can explore the potential correlations between meteorological factors such as relative humidity, temperature, precipitation and the distribution network load. For example, an increase in temperature may lead to an increase in the electricity consumption of cooling equipment such as residential air conditioners, and relative humidity and precipitation may also affect the electricity consumption of industrial production. By quantifying the characteristics of these meteorological factors, we can better understand how meteorological conditions affect the load changes of the distribution network and provide more comprehensive considerations for load forecasting.

[0077] Incorporating the processed environmental meteorological characteristic data into the load forecasting model can enable the model to more accurately capture the load fluctuations caused by meteorological changes. When the model considers meteorological factors such as water vapor pressure and temperature signals, it can more accurately predict the load demand under different seasons and weather conditions. Especially under extreme meteorological conditions, the accuracy of load forecasting can be significantly improved.

[0078] Environmental meteorological conditions not only affect the distribution network load, but also have an important impact on new energy power generation (such as photovoltaic power generation and wind power generation). The characteristic data obtained by processing environmental meteorological data can be used to jointly analyze the relationship between new energy power generation and the distribution network load. For example, when the light intensity decreases due to factors such as precipitation, combined with the distribution network load situation, reasonably arrange the power generation plans of traditional power sources and new energy power sources to ensure the stability and reliability of power supply.

[0079] Based on the processed multi-source data, obtain the characteristic data of new energy power generation, the historical load characteristic data of the distribution network, and the environmental meteorological characteristic data of the distribution network, construct a hybrid model for distribution network load forecasting, and output the initial load forecasting characteristics of the distribution network.

[0080] Specifically, to construct a hybrid model for distribution network load forecasting and output the initial load forecasting characteristics of the distribution network, the specific analysis process is as follows: Based on the characteristic data of new energy power generation, the historical load characteristic data of the distribution network, and the environmental meteorological characteristic data of the distribution network, construct a hybrid model for distribution network load forecasting to obtain the initial load forecasting characteristics of the distribution network. The initial load forecasting characteristics of the distribution network are used as the analysis basis for constructing the target optimization model;

[0081] The specific analysis process of the hybrid model for distribution network load forecasting is as follows:

[0082]

[0083] In the formula, δ is the initial load forecasting characteristic of the distribution network, Energy newEnergy is the characteristic factor of new energy power generation, Load is the historical load characteristic factor of the distribution network, Envir is the environmental meteorological characteristic factor of the distribution network, P th is the theoretical power generation characteristic of a single photovoltaic module, P wind is the theoretical power generation characteristic of a wind turbine generator, P avg is the historical average load of the distribution network, P sin is the historical load signal of the distribution network, e sy is the environmental water vapor pressure of the distribution network, Sg d is the environmental humidity signal of the distribution network, μ1 is the set compensation factor of Energy new The compensation factor, μ2 is the set compensation factor of Load, and μ3 is the set compensation factor of Envir.

[0084] The model is constructed based on multi-source characteristic data such as new energy power generation, historical load of the distribution network, and environmental meteorology. The new energy power generation characteristic factor considers the theoretical power of photovoltaic and wind power generation. The historical load characteristic factor covers the average load and load signal. The environmental meteorological characteristic factor includes water vapor pressure and temperature signal, etc. It can comprehensively reflect the influence of different factors on the load of the distribution network, avoid the limitations of single-factor analysis, and provide a richer and more comprehensive information basis for load forecasting.

[0085] The model structure of multi-source data fusion and comprehensive consideration of multiple factors enables the model to more accurately capture the laws and trends of load changes. Compared with the prediction methods that only rely on single data or simple models, this hybrid model can better cope with the complexity and uncertainty of the distribution network load after the high proportion of new energy access, effectively improve the accuracy of load forecasting, and provide a more reliable basis for power grid dispatching and operation.

[0086] The initial load forecasting characteristics of the distribution network output are used as the analysis basis for constructing the target optimization model, providing a basis for subsequent optimization and adjustment of the load forecasting results. Through further analysis and processing of these initial characteristics, the uncertainty and reliability of load forecasting can be more accurately evaluated, so as to formulate more reasonable optimization strategies and improve the stability of power grid operation.

[0087] Collect the uncertainty quantification parameters of the distribution network load forecasting. Based on the collected uncertainty quantification parameters of the distribution network load forecasting, obtain the uncertainty quantification results of the distribution network load forecasting.

[0088] Specifically, collect the uncertainty quantification parameters of the distribution network load prediction. Based on the collected uncertainty quantification parameters of the distribution network load prediction, obtain the uncertainty quantification result of the distribution network load prediction. The specific analysis process is as follows: Collect the uncertainty quantification parameters of the distribution network load prediction. The uncertainty quantification parameters of the distribution network load prediction specifically include the number of years since the distribution network was put into use, the equipment failure frequency of the distribution network, and the area covered by the distribution network. Based on the collected uncertainty quantification parameters of the distribution network load prediction, comprehensively analyze to obtain the uncertainty quantification result of the distribution network load prediction. The uncertainty quantification result of the distribution network load prediction is used as the analysis basis for constructing the target optimization model.

[0089] For the uncertainty quantification result of the distribution network load prediction, the specific analysis process is as follows:

[0090]

[0091] In the formula, Uncer is the uncertainty quantification result of the distribution network load prediction, Tn is the number of years since the distribution network was put into use, Sgp is the equipment failure frequency of the distribution network, fg is the area covered by the distribution network, υ1 is the compensation factor for the set Tn, υ2 is the compensation factor for the set Sgp, and υ3 is the compensation factor for the set fg.

[0092] It covers parameters such as the number of years since the distribution network was put into use, the equipment failure frequency, and the covered area, comprehensively reflecting the impact of the distribution network's own state and scale on load prediction. For example, the aging of distribution network equipment with a long service life may lead to unstable power supply and affect load prediction; a high equipment failure frequency will bring uncertainty to power supply; a large covered area means complex load distribution and increased prediction difficulty. By quantifying these factors, the reliability of the prediction result can be accurately evaluated.

[0093] In the operation and dispatch of the power grid, dispatchers can formulate multiple response plans in advance based on the uncertainty quantification result. When the predicted uncertainty is high, more reserve power generation capacity can be reserved, or the monitoring and guarantee of important loads can be strengthened, and the power grid operation mode can be flexibly adjusted to reduce the risks brought by load fluctuations and inaccurate predictions.

[0094] As one of the uncertainty quantification parameters, the equipment failure frequency can help power enterprises reasonably arrange equipment maintenance plans. For distribution network areas with a high failure frequency, increase the maintenance frequency, reserve spare parts in advance, conduct targeted equipment repairs and upgrades, reduce the impact of equipment failures on load prediction and power grid operation, and improve the reliability of equipment and the operation efficiency of the power grid.

[0095] During the power grid planning stage, the results of uncertainty quantification can serve as an important reference to help planners judge the uncertainty of the current distribution network status and future load growth. For distribution networks with a long service life, high equipment failure frequencies, and large coverage areas, planners can use this to evaluate whether it is necessary to increase the grid capacity, optimize the grid structure, or update equipment to ensure that the grid can adapt to load changes in the future, improve the scientific nature and rationality of the planning, and avoid waste or insufficiency of resources.

[0096] Based on the results of uncertainty quantification of distribution network load forecasting and the initial load forecasting characteristics of the distribution network, a target optimization model is constructed to obtain a target optimization signal. The target optimization signal is input into a trained linear regression model to obtain the target index of distribution network load forecasting. Based on the target index of distribution network load forecasting, the optimized load forecasting results of the distribution network are matched.

[0097] Specifically, based on the results of uncertainty quantification of distribution network load forecasting and the initial load forecasting characteristics of the distribution network, a target optimization model is constructed to obtain a target optimization signal. The target optimization signal is input into a trained linear regression model to obtain the target index of distribution network load forecasting. The specific analysis process is as follows: Based on the results of uncertainty quantification of distribution network load forecasting and the initial load forecasting characteristics of the distribution network, a target optimization model is constructed to output a target optimization signal, and the target optimization signal serves as the analysis basis for obtaining the target index of distribution network load forecasting; the target optimization signal is input into a trained linear regression model; the trained linear regression model can be expressed as:

[0098] y = g0 + g1 * Myh + b;

[0099] In the formula, y is the target index of distribution network load forecasting, g0 is the intercept, g1 is the first slope, Myh is the target optimization signal, and b is the error term;

[0100] Output the target index of distribution network load forecasting, and the target index of distribution network load forecasting serves as the analysis basis for matching the optimized load forecasting results of the distribution network.

[0101] The target optimization model, the specific analysis process is as follows:

[0102] Myh = δ + ln(1 + Uncer);

[0103] In the formula, Myh is the target optimization signal, δ is the initial load forecasting characteristic of the distribution network, and Uncer is the result of uncertainty quantification of distribution network load forecasting.

[0104] Match the optimized load forecasting results of the distribution network based on the load forecasting target indicators of the distribution network. The specific analysis process is as follows: Obtain the mapping table of the load forecasting target indicators - optimized load forecasting results of the distribution network pre-stored in the database. By searching the mapping table and according to the load forecasting target indicators of the distribution network, find the matching optimized load forecasting results.

[0105] The target optimization model synthesizes the initial load forecasting characteristics of the distribution network and the quantification results of load forecasting uncertainty. The former includes multi-source data information such as new energy generation, historical load, and environmental meteorology, reflecting the internal law of load change; the latter considers the uncertainty brought by factors such as the state and scale of the distribution network itself. By integrating these two types of information, the model can capture various factors affecting the load more comprehensively, reduce the prediction error caused by missing or one-sided information, and thus improve the accuracy of load forecasting.

[0106] The target optimization signal is further processed by the trained linear regression model to obtain the load forecasting target indicators, and this process optimizes and adjusts the initial forecasting results. The linear regression model can fit and predict the target optimization signal according to the rules learned from historical data, making the final load forecasting target indicators closer to the actual load situation and compensating for the possible deviation in the initial forecasting.

[0107] Match the optimized load forecasting results through the pre-stored mapping table of the load forecasting target indicators - optimized load forecasting results of the distribution network. This method is based on the summary of a large amount of historical data and experience. It can quickly and accurately find the corresponding optimized load forecasting results according to the calculated load forecasting target indicators, reduce the uncertainty brought by manual judgment and estimation, and further improve the reliability and stability of load forecasting.

[0108] Accurate and reliable load forecasting results provide an important decision-making basis for grid dispatchers. Dispatchers can reasonably arrange power generation resources, optimize the grid operation mode, and improve the operation efficiency of the power system according to the optimized load forecasting results. For example, arrange the power generation equipment to increase output in advance during peak load periods, and reasonably adjust the equipment operation status during low load periods to reduce the power generation cost and grid loss.

[0109] Accurate load forecasting also helps to more reasonably plan the construction and expansion of the grid. By accurately estimating the future load, the appropriate substation capacity, transmission line specifications, and layout can be determined, avoiding over-construction or under-construction, improving the effectiveness and rationality of grid investment, ensuring that the grid can meet the future load growth demand, and promoting the sustainable development of the grid.

[0110] The above content is only an example and explanation of the structure of the present invention. Those skilled in the art to which the present technology pertains may make various modifications or supplements to the described specific embodiments or use similar methods for substitution, as long as they do not deviate from the structure of the invention or exceed the scope defined by this claims, and shall fall within the protection scope of the present invention.

Claims

1. A load forecasting method for a high proportion of new energy access to a distribution network, characterized in that: The following steps are involved: Use the power Internet of Things technology to build a comprehensive multi-source data collection system to collect multi-source data, including new energy power generation data, distribution network historical load data, and distribution network environmental meteorological data; Process the collected multi-source data, including processing the new energy power generation data, processing the distribution network historical load data, and processing the distribution network environmental meteorological data; Based on the processed multi-source data, the new energy power generation characteristic data, the distribution network historical load characteristic data, and the distribution network environmental meteorological characteristic data are obtained to build a distribution network load forecasting hybrid model and output the distribution network initial load forecasting characteristics; Collecting distribution network load forecasting uncertainty quantitative parameters, and obtaining distribution network load forecasting uncertainty quantitative results based on the collected distribution network load forecasting uncertainty quantitative parameters; Based on the uncertainty quantification results of distribution network load forecasting and the initial load forecasting characteristics of the distribution network, a target optimization model is constructed to obtain the target optimization signal. The target optimization signal is input into the trained linear regression model to obtain the target index of distribution network load forecasting. The distribution network optimized load forecasting result is matched based on the distribution network load forecasting target index.

2. The load forecasting method for a high proportion of new energy access to a distribution network according to claim 1 is characterized in that: The specific analysis process of processing the renewable energy power generation data is as follows: The new energy power generation data specifically includes the area S of the photovoltaic module, the light radiation intensity H, the photoelectric conversion efficiency η of the photovoltaic module, the air density ρ, the wind rotor swept area A, the wind speed V, and the wind energy utilization coefficient C of the wind turbine p ; Calculate the theoretical power generation characteristics P of a single photovoltaic module th : In the formula, e is a natural constant; Calculate the theoretical power characteristics P of wind turbines wind ; The theoretical power generation characteristics of a single photovoltaic module and the theoretical power generation characteristics of a wind turbine generator set are recorded as new energy power generation characteristic data.

3. The load forecasting method for a high proportion of new energy access to a distribution network according to claim 1 is characterized in that: The specific analysis process of processing the historical load data of the distribution network is as follows: The distribution network historical load data specifically includes the load power value P of the ith collection point in the distribution network historical statistical period. i , the historical maximum load power of the distribution network P max , the historical minimum load power of the distribution network P min , i is the number of each collection point, i = 1, 2, 3, ..., n, n is the total number of collection points; Get the historical average load P of the distribution network avg : Calculate the distribution network historical load signal P sin ; The historical average load of the distribution network and the historical load signal of the distribution network are recorded as the historical load characteristic data of the distribution network.

4. The load forecasting method for a high proportion of new energy access to a distribution network according to claim 1 is characterized in that: The specific analysis process of processing the distribution network environment meteorological data is as follows: The distribution network environment meteorological data specifically includes the distribution network environment relative humidity RH, the distribution network environment temperature T, and the distribution network environment precipitation js; Get the ambient water vapor pressure e of the distribution network sy : Calculate the distribution network environmental humidity signal Sg d : In the formula, e is a natural constant; The distribution network environment water vapor pressure and distribution network environment humidity signals are recorded as distribution network environment meteorological characteristic data.

5. The load forecasting method for a high proportion of new energy access to a distribution network according to claim 1 is characterized in that: The hybrid model for load forecasting of the distribution network is constructed to output the initial load forecasting characteristics of the distribution network. The specific analysis process is as follows: Based on the new energy power generation characteristic data, the distribution network historical load characteristic data, and the distribution network environmental meteorological characteristic data, a distribution network load forecasting hybrid model is constructed to obtain the initial load forecasting characteristics of the distribution network. The initial load forecasting characteristics of the distribution network are used as the analysis basis for constructing the target optimization model. The hybrid model for distribution network load forecasting has the following specific analysis process: Where δ is the initial load forecasting characteristic of the distribution network, Energy new is the characteristic factor of renewable energy power generation, Load is the characteristic factor of distribution network historical load, Envir is the characteristic factor of distribution network environment and meteorology, P th is the theoretical power generation characteristic of a single photovoltaic module, P wind is the theoretical power generation characteristic of the wind turbine generator set, P avg is the historical average load of the distribution network, P sin is the historical load signal of the distribution network, e sy is the ambient water vapor pressure of the distribution network, Sg d is the environmental humidity signal of the distribution network, μ1 is the set Energy new μ2 is the compensation factor of the set Load, μ3 is the compensation factor of the set Envir, and e is a natural constant.

6. The load forecasting method for a high proportion of new energy access to a distribution network according to claim 1 is characterized in that: The distribution network load forecast uncertainty quantification parameters are collected, and based on the collected distribution network load forecast uncertainty quantification parameters, the distribution network load forecast uncertainty quantification results are obtained. The specific analysis process is: Collect the quantitative parameters of distribution network load forecast uncertainty, which specifically include the number of years the distribution network has been in use, the frequency of distribution network equipment failures, and the area covered by the distribution network; Based on the collected distribution network load forecasting uncertainty quantitative parameters, a comprehensive analysis is performed to obtain the distribution network load forecasting uncertainty quantitative results, which are used as the analysis basis for constructing the target optimization model.

7. A load forecasting method for high-proportion new energy access to a distribution network according to claim 6, characterized in that: The distribution network load forecast uncertainty quantification result, the specific analysis process is as follows: Where Uncer is the quantitative result of the uncertainty of distribution network load forecasting, Tn is the number of years the distribution network has been in service, Sgp is the failure frequency of distribution network equipment, fg is the area covered by the distribution network, υ1 is the compensation factor of the set Tn, υ2 is the compensation factor of the set Sgp, and υ3 is the compensation factor of the set fg.

8. The load forecasting method for a high proportion of new energy access to a distribution network according to claim 1 is characterized in that: Based on the uncertainty quantification results of the distribution network load forecast and the initial load forecast characteristics of the distribution network, a target optimization model is constructed to obtain a target optimization signal, and the target optimization signal is input into the trained linear regression model to obtain the distribution network load forecast target index. The specific analysis process is as follows: Based on the uncertainty quantification results of distribution network load forecasting and the initial load forecasting characteristics of the distribution network, a target optimization model is constructed and a target optimization signal is output. The target optimization signal is used as the analysis basis for obtaining the target indicators of distribution network load forecasting. Input the target optimization signal into the trained linear regression model; The trained linear regression model is expressed as: y=g0+g1*Myh+b; In the formula, y is the target index of distribution network load forecasting, g0 is the intercept, g1 is the first slope, Myh is the target optimization signal, and b is the error term; The output distribution network load forecast target index is used as the analysis basis for matching the distribution network optimized load forecast result.

9. A method for load forecasting when a high proportion of new energy is connected to a distribution network according to claim 8, characterized in that: The target optimization model, the specific analysis process is: Myh = δ + ln (1 + Uncer); Where Myh is the target optimization signal, δ is the initial load forecasting feature of the distribution network, and Uncer is the uncertainty quantification result of the distribution network load forecasting.

10. A method for load forecasting when a high proportion of new energy is connected to a distribution network according to claim 8, characterized in that: The load forecasting target index of the distribution network is matched with the distribution network optimization load forecasting result, and the specific analysis process is as follows: Obtain the distribution network load forecast target index-distribution network optimized load forecast result mapping table pre-stored in the database, and find the matching distribution network optimized load forecast result according to the distribution network load forecast target index by searching the mapping table.

Images