Fine prediction method for load and power supply space-time distribution based on artificial intelligence

Abstract

The invention, which relates to the technical field of the smart power grid, discloses an artificial intelligence-based fine prediction method for load and power supply spatial-temporal distribution, and the method comprises the steps: S1, building a multi-source spatial-temporal information hierarchical correlation model; s2, according to the multi-source data in the correlation model, carrying out public charging station candidate point distribution position optimization to obtain public charging station candidate point distribution positions; s3, obtaining a better public charging station configuration scheme by using a multi-target optimization model configured by the public charging station within the range of the public charging station candidate distribution position; and S4, taking the public charging station configuration scheme as input, simulating the electric taxi charging load spatial-temporal distribution by using a multi-agent system, and obtaining an electric taxi charging load spatial-temporal distribution prediction result under the fixed condition of the public charging station. The mutual influence of charging station configuration and charging load space-time distribution is fully considered.

Description

Technical Field

[0001] This invention relates to the field of smart grid technology, and in particular to a method for precise prediction of the spatiotemporal distribution of load and power supply based on artificial intelligence. Background Technology

[0002] With the continuous improvement of modern science and technology and the ever-increasing demands of people's production and life, my country's power industry is showing a diversified development trend in terms of user load types, electricity consumption characteristics, and upstream and downstream relationships. At the same time, the vigorous development of renewable energy power generation such as wind and solar power, and the continuous application of energy storage technology, have brought about significant changes in the power supply and load structure of the power distribution network. The high proportion of renewable energy connected to the grid has led to increased uncertainty in the power system, increasingly complex factors related to power supply and load, and unclear time-delay effects. This makes traditional spatiotemporal distribution prediction methods for the "source" and "load" ends of the distribution network no longer perfectly adequate, posing new challenges to the development of the power system.

[0003] In recent years, with the rapid development of information technology, artificial intelligence (AI) technology has also ushered in a period of rapid development. The power industry is accelerating its digital transformation, promoting industry reform, and improving the intelligence level of the power grid. Currently, the application of AI in the power grid field covers multiple areas such as power load forecasting, power grid fault prediction, large power grid dispatch simulation, and intelligent inspection, greatly improving data storage and analysis capabilities. Power companies have gradually formed a big data system for energy microgrids, including production data, marketing data, geographic information data, meteorological data, and related socio-economic data, providing a data foundation for achieving deep integration among regional energy sources, grids, loads, and storage, and realizing the goal of source-load interaction. However, smart grids based on big data technology are characterized by high dimensionality, complex computation, and diverse types, which limits the use of traditional analysis techniques for load and power source prediction and coordination. This makes it difficult to fully explore and utilize the advantages of big data, and to accurately predict power load demand and renewable energy output, as well as related spatiotemporal distribution factors. Therefore, it is of great significance to solve the problem of constructing accurate spatiotemporal distribution prediction technology for load types, energy storage and renewable energy with increasingly complex correlation factors in the context of big data, to achieve deep integration between regional power generation, grid, load and storage, to achieve the goal of power generation-load interaction, to ensure the safe and stable operation of the power system and improve social and economic benefits.

[0004] Against the backdrop of a growing energy crisis and environmental concerns, electric vehicles (EVs), as a low-carbon and environmentally friendly mode of transportation, have attracted significant attention from governments, energy companies, and automakers worldwide. Many countries have formulated relevant support policies, hoping to reduce carbon emissions in the transportation sector by promoting the development of the EV industry. With the continued growth in the number of EVs, the proportion of EV charging load in the total load of the power distribution network will become increasingly larger. However, due to the uncertainty and diversity of future EV demand and user behavior, the future charging load of large-scale EVs will exhibit uncertain characteristics such as temporal and spatial volatility, intermittency, and randomness. Therefore, it is necessary to establish a spatiotemporal distribution prediction model for EV charging load to lay the foundation for analyzing the impact of large-scale EV access on the power distribution network and formulating orderly control and management strategies for EVs.

[0005] Based on this, this application proposes a refined prediction method for the spatiotemporal distribution of load and power supply based on artificial intelligence. Summary of the Invention

[0006] This invention provides a refined prediction method for the spatiotemporal distribution of load and power supply based on artificial intelligence. It alternately updates the configuration of public charging stations and the prediction results of the spatiotemporal distribution of charging load, and fully considers the mutual influence between the configuration of charging stations and the spatiotemporal distribution of charging load.

[0007] According to one aspect of this disclosure, a refined prediction method for the spatiotemporal distribution of load and power supply based on artificial intelligence is provided, the method comprising: S1, Establish a hierarchical correlation model for multi-source spatiotemporal information; S2, Based on the multi-source data in the association model, the candidate locations of public charging stations are optimized to obtain the candidate locations of public charging stations; S3. Within the range of candidate locations for public charging stations, a better public charging station configuration scheme is obtained using a multi-objective optimization model for public charging station configuration. S4. Taking the public charging station configuration scheme as input, the multi-agent system is used to simulate the spatiotemporal distribution of electric taxi charging load, and the spatiotemporal distribution prediction results of electric taxi charging load under fixed conditions of public charging stations are obtained. The results of the spatiotemporal distribution prediction of charging load include: the curve of charging load change over time when traffic nodes are fixed in space, the average number of electric taxis charging per hour at different traffic nodes, the average waiting time of electric taxis, the average charging power of electric taxis, the average charging price, and the profit of each public charging station operator.

[0008] In one possible implementation, the method further includes: S5, determine whether the load distribution uniformity of the charging load spatiotemporal distribution prediction result is greater than the threshold, and determine whether the total economic cost of the public charging station configuration scheme is less than the threshold. If one of the two conditions is not met or neither of them is met, return to step S3. If both conditions are met, determine the final electric taxi charging load spatiotemporal distribution prediction result.

[0009] In one possible implementation, establishing a multi-source spatiotemporal information hierarchical correlation model includes: (1) Multi-source data identification: Attribute identification of multi-source big data of smart distribution network, collection and judgment of the source, format, collection time and collection location of multi-source data; (2) Metadata construction: For each actual data object, the device ID, collection time and collection location coordinates and other identifier fields are mapped into a standard and unified metadata format; (3) Data layer networking: Multi-source data are divided into geographic information data layer, distribution network information data layer, load and photovoltaic power source data layer, traffic information data layer and socio-economic data layer respectively; (4) Multi-source spatiotemporal information hierarchical association: establish horizontal and vertical associations between different data nodes, edges or grids through a three-dimensional spatial coordinate system.

[0010] In one possible implementation, data layer networking includes: Based on the geographic coordinates in the metadata, all data is laid out across the data layers, and each data layer is abstracted into a network or grid. That is, a data network consisting of nodes and edges is generated in each data layer, or the data layer is divided into data grids of equal size. In each node, edge, or grid, the attribute data or time series data of the corresponding geographic location is stored.

[0011] In one possible implementation, the X and Y axes describe the spatial location of the data, while the Z axis describes the hierarchical structure of the data. For data nodes, edges, or grids within the same data layer, horizontal relationships are achieved through the same Z-axis coordinates, and distinctions are made based on geographical location differences. For data nodes, edges, or grids in different data layers, vertical relationships between data are achieved using geographic information as the core.

[0012] In one possible implementation, when selecting candidate locations for public charging stations, the relative importance of each traffic node is first calculated, and then traffic nodes with a relative importance greater than a threshold and associated with the distribution network bus are selected as candidate locations. The relative importance is defined as the number of times all taxi routes pass through a certain traffic node divided by the maximum number of times all routes pass through that traffic node.

[0013] In one possible implementation, multi-objective optimization of public charging station configuration includes optimizing the location and capacity of public charging stations; Economic optimization objective: To minimize the total economic cost of the public charging station configuration scheme. The total economic cost includes: the construction and operation cost of the public charging station, the power loss cost of the power distribution network, the charging cost of electric taxis, the time cost of electric taxis charging and waiting at public charging stations, and the opportunity loss cost for passengers. Load uniformity optimization objective: Minimize load unevenness in the distribution network; The greater the load unevenness, the more uneven the load is in the distribution network.

[0014] Compared with the prior art, the beneficial effects of the present invention are: A method for predicting the spatiotemporal distribution of electric taxi charging load by integrating multi-source data, including road network, power grid, and vehicle operation data, is proposed. This method constructs a simulation model of the spatiotemporal distribution of electric taxi charging load based on a multi-agent system, considering multiple influencing factors and their interactions, such as passengers, taxis, public charging stations, transportation networks, and distribution networks. This model is then embedded into the iterative process of optimizing the configuration of public charging stations, achieving alternating updates of public charging station configuration schemes and the predicted spatiotemporal distribution of electric taxi charging load. The innovation of this method lies in: guided by the principles of "multidimensional expansion" and "data-driven" in a multi-domain interactive research paradigm, simulating the charging behavior of large-scale taxis through a data-driven multi-agent approach, effectively taking into account the behavioral and locational uncertainties of electric taxis as mobile loads during operation; and alternating updates of public charging station configuration and the predicted spatiotemporal distribution of charging load, fully considering the mutual influence between charging station configuration and the spatiotemporal distribution of charging load. Attached Figure Description

[0015] Figure 1 A flowchart illustrating an embodiment of the present disclosure shows a method for fine prediction of the spatiotemporal distribution of load and power supply based on artificial intelligence.

[0016] Figure 2 This diagram illustrates an embodiment of an artificial intelligence-based method for fine prediction of the spatiotemporal distribution of load and power supply. Detailed Implementation

[0017] Various exemplary embodiments, features, and aspects of this disclosure will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0018] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0019] Furthermore, to better illustrate this disclosure, numerous specific details are set forth in the following detailed description. Those skilled in the art will understand that this disclosure can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main points of this disclosure.

[0020] According to one aspect of this disclosure, a refined prediction method for the spatiotemporal distribution of load and power supply based on artificial intelligence is provided, the method comprising: S1, Establish a hierarchical correlation model for multi-source spatiotemporal information; By transforming distribution network topology data, geographic information, and user electricity consumption data, multiple data layers based on geographic information are generated. A multi-source spatiotemporal information correlation model is constructed using the strong correlations between different data layers, thereby achieving integrated organization of multi-source big data for smart distribution networks to meet the data needs of load and photovoltaic power generation spatiotemporal distribution prediction. Taking several regions in East China as the main analysis objects, a multi-source spatiotemporal information hierarchical correlation model is constructed for each region to provide data support for conventional load spatiotemporal distribution prediction, electric taxi charging load spatiotemporal distribution prediction, regional photovoltaic power generation total installed capacity time series prediction, and distributed photovoltaic power generation spatiotemporal distribution prediction. The model consists of data layers such as a geographic outline layer, a cell partitioning layer, a road, building and user layer, and a load distribution layer, with each data layer interconnected through geographic coordinates.

[0021] S2, Based on the multi-source data in the association model, the candidate locations of public charging stations are optimized to obtain the candidate locations of public charging stations; S3. Within the range of candidate locations for public charging stations, a better public charging station configuration scheme is obtained using a multi-objective optimization model for public charging station configuration. S4. Taking the public charging station configuration scheme as input, the multi-agent system is used to simulate the spatiotemporal distribution of electric taxi charging load, and the spatiotemporal distribution prediction results of electric taxi charging load under fixed conditions of public charging stations are obtained. The results of the spatiotemporal distribution prediction of charging load include: the curve of charging load change over time when traffic nodes are fixed in space, the average number of electric taxis charging per hour at different traffic nodes, the average waiting time of electric taxis, the average charging power of electric taxis, the average charging price, and the profit of each public charging station operator.

[0022] In one possible implementation, the method further includes: S5, determine whether the load distribution uniformity of the charging load spatiotemporal distribution prediction result is greater than the threshold, and determine whether the total economic cost of the public charging station configuration scheme is less than the threshold. If one of the two conditions is not met or neither of them is met, return to step S3. If both conditions are met, determine the final electric taxi charging load spatiotemporal distribution prediction result.

[0023] In one possible implementation, establishing a multi-source spatiotemporal information hierarchical correlation model includes: (1) Multi-source data identification: Attribute identification of multi-source big data of smart distribution network, collection and judgment of the source, format, collection time and collection location of multi-source data; (2) Metadata construction: For each actual data object, the device ID, collection time and collection location coordinates and other identifier fields are mapped into a standard and unified metadata format; (3) Data layer networking: Multi-source data are divided into geographic information data layer, distribution network information data layer, load and photovoltaic power source data layer, traffic information data layer and socio-economic data layer respectively; (4) Multi-source spatiotemporal information hierarchical association: establish horizontal and vertical associations between different data nodes, edges or grids through a three-dimensional spatial coordinate system.

[0024] In one possible implementation, data layer networking includes: Based on the geographic coordinates in the metadata, all data is laid out across the data layers, and each data layer is abstracted into a network or grid. That is, a data network consisting of nodes and edges is generated in each data layer, or the data layer is divided into data grids of equal size. In each node, edge, or grid, the attribute data or time series data of the corresponding geographic location is stored.

[0025] In one possible implementation, the X and Y axes describe the spatial location of the data, while the Z axis describes the hierarchical structure of the data. For data nodes, edges, or grids within the same data layer, horizontal relationships are achieved through the same Z-axis coordinates, and distinctions are made based on geographical location differences. For data nodes, edges, or grids in different data layers, vertical relationships between data are achieved using geographic information as the core.

[0026] The data in the load and photovoltaic power spatiotemporal distribution forecasting data architecture exists in the following forms: 1) In the data resource layer, various data exist in the form of dispersed data sources. A single data source may exist continuously in time or space, but various data sources are isolated from each other in time or space. The logical relationship of these multi-source data in the database is still centered on the equipment ledger; 2) In the data processing layer, various data exist in a hierarchical spatiotemporal distribution. Within the data layer, time-series or non-time-series data are decomposed into grids or network nodes for storage according to the highest achievable accuracy. Different data layers are interconnected through geographical coordinates, thereby realizing multi-source data deconstruction and spatiotemporal correlation; 3) In the data warehouse layer, data is stored according to business-oriented time granularity and spatial scale. For example, in conventional load spatiotemporal distribution forecasting, the time granularity and spatial scale of the data are standardized. At the spatial level, data is standardized according to cell size. At the time level, time-series data is standardized annually or monthly according to annual or monthly forecasting needs, thereby achieving clear and unified data deconstruction accuracy; 4) In the data application layer, the final prediction results of each application are stored, including the time-series change data of conventional load in each cell, the time-series change data of electric taxi charging load in each traffic node, the time-series change data of photovoltaic power installed capacity in the study area, and the time-series change data of distributed photovoltaic power in each cell.

[0027] In one possible implementation, when selecting candidate locations for public charging stations, the relative importance of each traffic node is first calculated, and then traffic nodes with a relative importance greater than a threshold and associated with the distribution network bus are selected as candidate locations. The relative importance is defined as the number of times all taxi routes pass through a certain traffic node divided by the maximum number of times all routes pass through that traffic node.

[0028] In one possible implementation, multi-objective optimization of public charging station configuration includes optimizing the location and capacity of public charging stations; Economic optimization objective: To minimize the total economic cost of the public charging station configuration scheme. The total economic cost includes: the construction and operation cost of the public charging station, the power loss cost of the power distribution network, the charging cost of electric taxis, the time cost of electric taxis charging and waiting at public charging stations, and the opportunity loss cost for passengers. Load uniformity optimization objective: Minimize load unevenness in the distribution network; The greater the load unevenness, the more uneven the load is in the distribution network.

[0029] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or technical improvements to the embodiments in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A refined prediction method for the spatiotemporal distribution of load and power supply based on artificial intelligence, characterized in that, The method includes: S1, Establish a hierarchical correlation model for multi-source spatiotemporal information; S2, Based on the multi-source data in the association model, the candidate locations of public charging stations are optimized to obtain the candidate locations of public charging stations; S3. Within the range of candidate locations for public charging stations, a better public charging station configuration scheme is obtained using a multi-objective optimization model for public charging station configuration. S4. Taking the public charging station configuration scheme as input, the multi-agent system is used to simulate the spatiotemporal distribution of electric taxi charging load, and the spatiotemporal distribution prediction results of electric taxi charging load under fixed conditions of public charging stations are obtained. The results of the spatiotemporal distribution prediction of charging load include: the curve of charging load change over time when traffic nodes are fixed in space, the average number of electric taxis charging per hour at different traffic nodes, the average waiting time of electric taxis, the average charging power of electric taxis, the average charging price, and the profit of each public charging station operator.

2. The method for fine prediction of load and power spatiotemporal distribution based on artificial intelligence according to claim 1, characterized in that, The method further includes: S5, determine whether the load distribution uniformity of the charging load spatiotemporal distribution prediction result is greater than the threshold, and determine whether the total economic cost of the public charging station configuration scheme is less than the threshold. If one of the two conditions is not met or neither of them is met, return to step S3. If both conditions are met, determine the final electric taxi charging load spatiotemporal distribution prediction result.

3. The method for fine prediction of load and power spatiotemporal distribution based on artificial intelligence according to claim 1, characterized in that, The establishment of a multi-source spatiotemporal information hierarchical association model includes: (1) Multi-source data identification: Attribute identification of multi-source big data of smart distribution network, collection and judgment of the source, format, collection time and collection location of multi-source data; (2) Metadata construction: For each actual data object, the device ID, collection time and collection location coordinates and other identifier fields are mapped into a standard and unified metadata format; (3) Data layer networking: Multi-source data are divided into geographic information data layer, distribution network information data layer, load and photovoltaic power source data layer, traffic information data layer and socio-economic data layer respectively; (4) Multi-source spatiotemporal information hierarchical association: establish horizontal and vertical associations between different data nodes, edges or grids through a three-dimensional spatial coordinate system.

4. The method for fine prediction of load and power spatiotemporal distribution based on artificial intelligence according to claim 3, characterized in that, Data layer networking includes: Based on the geographic coordinates in the metadata, all data is laid out across the data layers, and each data layer is abstracted into a network or grid. That is, a data network consisting of nodes and edges is generated in each data layer, or the data layer is divided into data grids of equal size. In each node, edge, or grid, the attribute data or time series data of the corresponding geographic location is stored.

5. The method for fine prediction of load and power spatiotemporal distribution based on artificial intelligence according to claim 3, characterized in that, The X and Y axes describe the spatial location of the data, while the Z axis describes the hierarchical structure of the data. For data nodes, edges, or grids within the same data layer, horizontal associations are achieved through the same Z-axis coordinates, and they are distinguished by differences in geographical location. For data nodes, edges, or grids in different data layers, vertical relationships between data are achieved with geographic information as the core.

6. The method for fine prediction of load and power spatiotemporal distribution based on artificial intelligence according to claim 1, characterized in that, When selecting candidate locations for public charging stations, the relative importance of each traffic node is first calculated, and then traffic nodes with a relative importance greater than a threshold and associated with the distribution network bus are selected as candidate locations. The relative importance is defined as the number of times all taxi routes pass through a certain traffic node divided by the maximum number of times all routes pass through that traffic node.

7. The method for fine prediction of load and power spatiotemporal distribution based on artificial intelligence according to claim 1, characterized in that, Multi-objective optimization of public charging station configuration includes optimizing the location and capacity of public charging stations; Economic optimization objective: To minimize the total economic cost of the public charging station configuration scheme. The total economic cost includes: the construction and operation cost of the public charging station, the power loss cost of the power distribution network, the charging cost of electric taxis, the time cost of electric taxis charging and waiting at public charging stations, and the opportunity loss cost for passengers. Load uniformity optimization objective: Minimize load unevenness in the distribution network; The greater the load unevenness, the more uneven the load is in the distribution network.

Images