New energy station digital intelligent operation system, control method, equipment and medium

Abstract

The invention provides a new energy station digital intelligent operation system, a control method, equipment and a medium, and relates to the technical field of new energy. According to the system, a marketization electricity price prediction result and a new energy output prediction result are output through an electricity price prediction module and a new energy output prediction module respectively, and an energy storage optimization scheduling module and a planning layout decision module work cooperatively. And combining the marketization electricity price prediction result and the new energy output prediction result to realize operation strategy optimization and capacity configuration decision of the new energy station, and solving and outputting a capacity configuration scheme corresponding to the new energy device and the energy storage device. According to the technical scheme, the multi-module collaborative new energy station digital intelligent operation system is constructed, so that full-link closed-loop optimization from prediction to scheduling to planning is realized, the adaptability of the system to power market fluctuation can be remarkably enhanced, and the operation income of the new energy station is ensured.

Description

Technical Field

[0001] The present disclosure relates to the field of new energy technology, and specifically to a digital and intelligent operation system for a new energy station and a control method, equipment, and medium. Background Art

[0002] With the continuous advancement of the "dual carbon" goals, the proportion of new energy stations in the power system has continued to rise, becoming an important support force for achieving a green and low-carbon transformation. Especially in the context of large-scale access to the power grid for renewable energy such as wind power and photovoltaics, the operation and management of new energy stations are facing a profound transformation from policy-driven to market-driven. Under the guidance of relevant policies, the fixed on-grid electricity price mechanism for renewable energy has been gradually abolished, and instead a price mechanism based on market transactions has been adopted to promote the full participation of new energy power in multiple trading links such as the electricity spot market, ancillary service market, and capacity market. This transformation means that new energy power generation projects will no longer rely on relevant subsidies, but will need to achieve economic benefits by actively participating in market transactions and responding to electricity price fluctuations.

[0003] However, as electricity pricing continues to become more market-oriented, a mismatch persists between the digital and intelligent operation of renewable energy stations and the complexity of power market mechanisms. On the one hand, renewable energy power is highly volatile and uncertain, particularly influenced by factors like meteorological conditions and geographic location, making its output characteristics difficult to accurately predict. On the other hand, electricity market prices are influenced by a multitude of intertwined factors, with both short-term volatility and long-term trends. This presents renewable energy stations with significant uncertainty risks when formulating trading strategies and operational decisions.

[0004] Furthermore, traditional operation and dispatch methods typically use fixed rules or linear models to make energy storage charging and discharging decisions, ignoring the deep nonlinear coupling between power market supply and demand dynamics and changes in renewable energy output. This results in insufficient dispatching flexibility and operational results that struggle to balance economic efficiency and system stability. Furthermore, during the planning phase for renewable energy sites, there is often an over-reliance on resource endowments as the basis for site selection, lacking systematic consideration of factors such as future market electricity price trends, load center distribution, and grid constraints. This results in some projects, despite having good resource conditions, facing low returns or even difficulty in profitability in actual operation.

[0005] In summary, the current operation and management of new energy power plants in the context of power marketization generally suffers from problems such as insufficient output forecast accuracy, lack of market responsiveness, lagging energy storage scheduling strategies, and a single dimension of planning and layout decision-making, making it difficult to achieve stable, efficient, and profit-oriented operational goals. Therefore, there is an urgent need to build a new digital and intelligent operation system that can adapt to market changes, integrate multi-source information, and possess intelligent analysis and strategy generation capabilities to enhance the viability and profitability of new energy power plants in the competitive power market.

[0006] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute prior art known to ordinary technicians in the field. Summary of the Invention

[0007] The purpose of the embodiments of the present disclosure is to provide a digital and intelligent operation system and control method, equipment, and medium for a new energy station, thereby significantly enhancing the system's adaptability to fluctuations in the electricity market and ensuring the operating benefits of the new energy station.

[0008] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by practice of the present disclosure.

[0009] According to a first aspect of an embodiment of the present disclosure, a digital and intelligent operation system for a new energy station is provided, comprising: The electricity price prediction module is used to input the acquired electricity price disturbance feature data into the electricity price prediction model based on the fusion of time series model and attention mechanism to predict the market electricity price in the future period and output the market electricity price prediction results at multiple time scales; A new energy output prediction module is used to input the acquired spatiotemporal output data of the station into a new energy output prediction module based on the fusion of time series and space features, so as to predict the power generation power of the new energy device of the new energy station and obtain the new energy output prediction result; an energy storage optimization scheduling module, connected to the electricity price prediction module and the new energy output prediction module, for constructing a diversified target model based on the market-based electricity price prediction results and the new energy output prediction results, and generating a charging and discharging strategy for the energy storage device by solving the diversified target model; A planning and layout decision module is coupled to the electricity price forecast module, the new energy output forecast module, and the energy storage optimization scheduling module, and is used to combine the site selection conditions of the new energy station, the market-based electricity price forecast results, the new energy output forecast results, and the energy storage device charging and discharging strategy to solve and output the capacity configuration plan corresponding to the new energy device and the energy storage device.

[0010] In some example embodiments of the present disclosure, based on the aforementioned solution, the electricity price prediction model includes an electricity price time series feature extraction network based on a time series model structure and a feature enhancement representation network based on an attention mechanism; The electricity price prediction module includes: a first data fusion unit, configured to vectorize various multi-source heterogeneous data in the electricity price disturbance characteristic data to obtain an electricity price disturbance vector; The first model scheduling unit is used to schedule the trained electricity price prediction model to perform time series modeling on the electricity price disturbance vector through the electricity price time series feature extraction network, extract the electricity price time series features, and construct a weight matrix based on the electricity price time series features through the feature enhancement characterization network to achieve focused modeling of important disturbance factors and output the market-based electricity price prediction result.

[0011] In some example embodiments of the present disclosure, based on the aforementioned solution, the new energy output prediction module includes a spatial feature extraction network based on a convolutional neural network structure, an output time series feature extraction network based on a time series model structure, and a feature fusion network: The new energy output prediction module includes: The second data fusion unit is used to vectorize the multi-source heterogeneous data in the station output spatiotemporal data to obtain a meteorological space vector and a new energy output time series vector; The second model scheduling unit is used to schedule the trained new energy output prediction module to extract spatial features from the meteorological space vector through the spatial feature extraction network to obtain spatial features, and to perform time series modeling on the new energy output time series vector through the output time series feature extraction network to obtain time series features, and to fuse the spatial features and the time series features through the feature fusion network to output the new energy output prediction result.

[0012] In some example embodiments of the present disclosure, based on the aforementioned solution, the energy storage optimization scheduling module includes: an optimization model construction unit, configured to input the market-based electricity price forecast result and the renewable energy output forecast result as boundary conditions, and to construct a multi-objective optimization model with economic, technical, and social objectives in combination with physical constraints, grid constraints, and operational constraints as constraints; wherein the physical constraints, grid constraints, and operational constraints include any one or a combination of energy storage device type, initial capacity, battery cycle life, charge and discharge efficiency, equipment cost, grid load, and frequency regulation and standby requirements; The optimization solution unit is used to call the multi-objective optimization model, perform iterative calculations on the multi-objective optimization model based on a non-dominated sorting genetic algorithm or a multi-agent optimization algorithm, and output a charging and discharging strategy for the energy storage device that meets multiple constraints.

[0013] In some example embodiments of the present disclosure, based on the aforementioned solution, the planning and layout decision module includes: a strategy input unit, configured to obtain site selection condition information corresponding to the new energy site and receive resource attribute decision and market attribute decision inputs, wherein the site selection condition information includes at least land availability, grid access capacity, policy incentive information, and load center distance; A coupling modeling unit, configured to construct a capacity configuration optimization model for a new energy device and an energy storage device based on the site selection condition information, the resource attribute decision, the market attribute decision, the market-based electricity price forecast result, the new energy output forecast result, and the energy storage device charging and discharging strategy; The model solving unit is used to solve the capacity configuration optimization model through a competitive game analysis method or a Nash equilibrium solving method, and output an optimal capacity configuration plan for the new energy device and the energy storage device.

[0014] In some example embodiments of the present disclosure, based on the aforementioned solution, the new energy station digital intelligent operation system further includes a data acquisition and preprocessing module, which includes: a multi-source data interface unit, configured to connect to data interfaces corresponding to a meteorological monitoring system, an electricity market data platform, a station monitoring system, and a policy information repository, and collect the electricity price disturbance characteristic data and the station output spatiotemporal data; wherein the electricity price disturbance characteristic data includes any one or a combination of historical electricity price data, seasonal climate data, geographic location information, fuel cost data, power generation technology, power line data, and supply and demand balance data; and the station output spatiotemporal data includes any one or a combination of historical output data, local meteorological data, atmospheric boundary layer data, irradiance data, and wind speed data; a data cleaning unit, configured to perform data preprocessing on the electricity price disturbance characteristic data and the station output spatiotemporal data to obtain preprocessed electricity price disturbance characteristic data and station output spatiotemporal data, wherein the data preprocessing at least includes outlier removal, missing value filling, and time alignment processing; The data normalization unit is used to standardize the pre-processed electricity price disturbance characteristic data and the site output spatiotemporal data to obtain normalized electricity price disturbance characteristic data and site output spatiotemporal data, and transmit them to the electricity price prediction module and the new energy output prediction module respectively.

[0015] In some example embodiments of the present disclosure, based on the aforementioned solution, the new energy station digital intelligent operation system further includes a market adaptation and dynamic update module, which includes: A data scheduling control unit is used to dynamically determine the data collection frequency and data collection priority according to the update rate, data importance and model call requirements of different types of electricity price disturbance characteristic data and station output spatiotemporal data, and distribute them to the multi-source data interface unit and data cache and backtracking unit; A data caching and backtracking unit, configured to cache and time-index the collected electricity price disturbance characteristic data and the site output spatiotemporal data in batches according to the received data collection frequency and the received data collection priority, so as to support the historical data backtracking requirements of different models; The data output management unit is used to extract the latest data of the target priority from the data cache and backtracking unit according to the data collection priority, and output it synchronously to the electricity price prediction module and the new energy output prediction module in the order of model calling.

[0016] According to a second aspect of an embodiment of the present disclosure, a control method for the digital intelligent operation system of a new energy station in the first aspect is provided, comprising: The electricity price prediction module inputs the acquired electricity price disturbance feature data into an electricity price prediction model based on the fusion of a time series model and an attention mechanism to predict the market-based electricity price for future periods and output market-based electricity price prediction results at multiple time scales. The obtained spatiotemporal data of the station output is input into the new energy output prediction module based on the fusion of time series and space features through the new energy output prediction module to predict the power generation of the new energy device of the new energy station and obtain the new energy output prediction result; Constructing a diversified target model based on the market-based electricity price forecast result and the new energy output forecast result through an energy storage optimization scheduling module, and generating a charging and discharging strategy for an energy storage device by solving the diversified target model; The planning and layout decision module combines the site selection conditions of the new energy station, the market-based electricity price forecast results, the new energy output forecast results and the energy storage device charging and discharging strategy to solve the capacity configuration plan corresponding to the output new energy device and the energy storage device.

[0017] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, including: a processor; and a memory, on which computer-readable instructions are stored. When the computer-readable instructions are executed by the processor, the digital and intelligent operation system of the new energy station in the first aspect is implemented.

[0018] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the digital and intelligent operation system of the new energy station in the first aspect is implemented.

[0019] The technical solutions provided by the embodiments of the present disclosure may have the following beneficial effects: The digital intelligent operation system of the new energy station in the example embodiment of the present disclosure, on the one hand, uses the output results of the electricity price prediction module and the new energy output prediction module as input to jointly drive the energy storage optimization scheduling module, and further feeds the energy storage device charging and discharging strategy as input to the planning and layout decision module to form a market-oriented decision cascade mechanism. The vertical information transmission and coupling between such modules enables each operation link to achieve coordinated and consistent strategy selection based on the same prediction. The system shows a high degree of consistency and goal unity at the overall level, which can significantly reduce the risk of resource mismatch caused by strategy conflicts; on the other hand, by dynamically inputting the prediction results into the energy storage optimization scheduling module, the energy storage strategy It can flexibly match market fluctuations at different time scales. At the same time, the dynamic adjustment results of the dispatch strategy in turn affect the capacity allocation plan, so that the newly built new energy sites have structural resistance to power market disturbances, and the system has the dual capabilities of feedforward prediction and feedback correction when facing real-time market changes, thereby improving the dispatch flexibility of the energy storage system; on the other hand, the generation of the capacity allocation plan depends on the coupled modeling of electricity price forecasts, output forecasts and energy storage dispatch results, so that the planning process can "simulate" future operating scenarios, thereby realizing a full-link feedback mechanism that reversely optimizes the early planning with operational efficiency, which can effectively improve the goal consistency and resource optimality from new energy site planning to real-time execution.

[0020] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification, are used to explain the principles of the present disclosure. Obviously, the drawings described below are only some embodiments of the present disclosure, and those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0022] Figure 1 The structural diagram of the digital and intelligent operation system of a new energy station according to some embodiments of the present disclosure is schematically shown.

[0023] Figure 2 The structural diagram of the electricity price prediction module according to some embodiments of the present disclosure is schematically shown.

[0024] Figure 3 The structure diagram of the new energy output prediction module according to some embodiments of the present disclosure is schematically shown.

[0025] Figure 4The schematic diagram shows the structure of the energy storage optimization scheduling module according to some embodiments of the present disclosure.

[0026] Figure 5 The structural diagram of the planning layout decision module according to some embodiments of the present disclosure is schematically shown.

[0027] Figure 6 The structural diagram of the data acquisition and preprocessing module according to some embodiments of the present disclosure is schematically shown.

[0028] Figure 7 The structural diagram of the market adaptation and dynamic update module according to some embodiments of the present disclosure is schematically shown.

[0029] Figure 8 A flow chart of a control method for a digital and intelligent operation system of a new energy station according to some embodiments of the present disclosure is schematically shown.

[0030] Figure 9 A schematic structural diagram of a computer system of an electronic device according to some embodiments of the present disclosure is schematically shown.

[0031] Figure 10 A schematic diagram of a computer-readable storage medium according to some embodiments of the present disclosure is schematically shown.

[0032] In the drawings, the same or corresponding reference numerals denote the same or corresponding parts. DETAILED DESCRIPTION

[0033] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with this specification. Rather, they are merely examples of apparatus and methods consistent with certain aspects of this specification, as detailed in the appended claims.

[0034] It should be understood that although the terms first, second, third, etc. may be used in this specification to describe various information, such information should not be limited to these terms. These terms are merely used to distinguish information of the same type from one another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information without departing from the scope of this specification. Depending on the context, the term "if" as used herein may be interpreted as "when," "when," or "in response to determining."

[0035] In addition, the described features, structures or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, well-known methods, devices, implementations or operations are not shown or described in detail to avoid blurring various aspects of the present disclosure.

[0036] Furthermore, the drawings are schematic illustrations only and are not necessarily drawn to scale. The block diagrams shown in the drawings are merely functional entities and do not necessarily correspond to physically separate entities. In other words, these functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0037] In this example embodiment, a new energy station digital intelligent operation system is first provided. Figure 1 As shown, the digital intelligent operation system 100 of the new energy station may include an electricity price prediction module 110, a new energy output prediction module 120, an energy storage optimization scheduling module 130, and a planning and layout decision module 140. Among them: The electricity price prediction module 110 can be used to input the acquired electricity price disturbance characteristic data into an electricity price prediction model based on a fusion of a time series model and an attention mechanism to predict the market-based electricity price for future periods and output market-based electricity price prediction results at multiple time scales. The electricity price disturbance characteristic data refers to a set of key factors that influence changes in market-based electricity prices, including but not limited to historical electricity price data, seasonal climate data, geographic location information, fuel cost data, power generation technology type, power line structure data, and power supply and demand balance data.

[0038] The electricity price forecasting model is a deep learning network architecture built by integrating time series modeling with an attention mechanism. Its core structure consists of a time series model for modeling the time series characteristics of electricity prices and an attention mechanism for enhancing the representation of key perturbations. The time series model can employ structures such as a recurrent neural network (RNN), a long short-term memory (LSTM), or a gated recurrent unit (GRU) to learn the trends and cyclical patterns of electricity prices over time. The LSTM, through its input, forget, and output gates, dynamically adjusts the contribution of historical states to the current output, effectively modeling the characteristics of electricity price series over longer time spans.

[0039] The attention mechanism introduces a self-attention mechanism to learn the relevance of different time segments to the current prediction point. By calculating the attention weights between the query, key, and value, it automatically focuses on key perturbation time periods, thereby enhancing the model's ability to model nonlinear perturbations such as price fluctuations and boundary effects. The attention mechanism in this embodiment can adopt a multi-head attention structure, combined with position encoding to achieve non-recursive information capture, significantly reducing computational complexity while improving the model's expressiveness.

[0040] The electricity price prediction module 110 can be implemented using deep learning frameworks such as TensorFlow and PyTorch. During the training phase, supervised learning can be performed using a loss function based on mean squared error (MSE) or weighted mean absolute error (WMAE) in conjunction with historically annotated datasets. Optionally, a graph neural network (GNN) can be used to enhance the modeling of structural correlations between perturbation factors, or external knowledge graphs can be integrated to improve the model's generalization capabilities.

[0041] The output after processing by the electricity price prediction module 110 is the market-based electricity price prediction result under multiple time scales, which can specifically include intraday prediction (15-minute or 1-hour granularity), day-ahead prediction (24-hour granularity) and monthly prediction, etc. The prediction results will serve as an important input source for the scheduling and optimization decisions of subsequent modules, thereby improving the system's ability to predict electricity price market fluctuations and regulatory foresight.

[0042] The new energy output prediction module 120 is used to input the acquired spatiotemporal data of the station output into a new energy output prediction model based on the fusion of time-series and spatial features to predict the power generation power of the new energy device at the new energy station and obtain the new energy output prediction result. For example, the new energy device can be a photovoltaic power generation device or a wind power generation device, but this embodiment is not limited to this.

[0043] Site output spatiotemporal data refers to a data set that characterizes the interaction between the operating status of new energy devices and the environment in which they operate. It can include historical output data, local meteorological data, atmospheric boundary layer data, irradiance data, wind speed data, etc. Among them, historical output data refers to the power generation records of various types of new energy devices (such as photovoltaic arrays and wind turbines) at a specific time scale; local meteorological data can include indicators such as temperature, humidity, wind speed, wind direction, and air pressure; atmospheric boundary layer data can reflect the impact of wind shear and thermal stability on the availability of wind resources; irradiance data can describe the intensity of solar radiation received per unit area of ​​the surface; and wind speed data can be collected wind speed variation data.

[0044] The renewable energy output prediction model can include a spatial feature extraction network, an output time series feature extraction network, and a feature fusion network. The spatial feature extraction network can employ a convolutional neural network (CNN) to extract spatial distribution patterns embedded in meteorological and geographic features. In practice, network structures of varying depths can be designed based on 2D or 3D convolution kernels to encode features of multivariate, multi-level spatial data and output a high-dimensional dense vector representation.

[0045] The output time series feature extraction network can be used to model the dynamic characteristics of historical power changes of new energy devices. The long short-term memory network (LSTM) or gated recurrent unit (GRU) structure is often used. This network structure can effectively identify time-dependent characteristics such as periodic trends, short-term fluctuations and operating inertia in new energy output, and enhance the model's ability to predict the future state of the power series.

[0046] The feature fusion network can be used to integrate spatial and temporal features, achieving deep fusion of cross-modal information. Fusion methods can include feature concatenation, attention-weighted fusion, or cross-network structures to ensure effective alignment of different information channels in a shared representation space, thereby improving final prediction accuracy. This embodiment does not impose any specific restrictions on the fusion method.

[0047] The implementation of the new energy output prediction module 120 can construct a multi-input multi-output neural network structure based on a deep learning framework, and use step-by-step prediction or sliding window methods to generate training samples during the training process.

[0048] The output renewable energy output forecast results can cover power generation estimation data at different time scales, providing dynamic boundary input for subsequent energy storage optimization scheduling modules, which helps to improve the dispatchability and operational safety margin of the energy system.

[0049] The energy storage optimization and scheduling module 130 is connected to the electricity price forecast module 110 and the renewable energy output forecast module 120. It is used to construct a diversified target model based on the market-based electricity price forecast results and the renewable energy output forecast results, and to generate energy storage device charging and discharging strategies by solving the diversified target model. This connection means that the energy storage optimization and scheduling module can receive the multi-timescale electricity price forecast results output by the electricity price forecast module and the generated power forecast value output by the renewable energy output forecast module in real time, and use these as input to establish an optimization decision model under operational constraints.

[0050] A multi-objective model is an optimization model that simultaneously considers multiple optimization objectives. It can include economic, technical, and social objectives. Economic objectives can include maximizing the return on investment, minimizing operating costs, or maximizing revenue over the energy storage system's lifecycle. Technical objectives can include extending battery life, balancing charge and discharge cycles, and enhancing the grid's frequency response capabilities. Social objectives can include minimizing carbon emissions, reducing wind and solar curtailment rates, and improving system stability.

[0051] When building a diversified target model, electricity price forecasts and renewable energy output forecasts can be used as boundary condition inputs, combined with multiple types of constraints. Constraints may include, but are not limited to: the electrochemical characteristics of the energy storage device type (e.g., lithium iron phosphate, flow batteries, sodium sulfur batteries, etc.), initial capacity settings, battery cycle life limits, charge and discharge efficiency parameters, equipment construction and operation and maintenance costs, grid load fluctuation range, grid-connected voltage and frequency constraints, and frequency regulation and backup scheduling requirements. Physical constraints reflect the operational boundaries of the energy storage system, grid constraints reflect grid connection conditions and scheduling feasibility, and operational constraints ensure the solution's time series feasibility.

[0052] The diverse objective models constructed by the energy storage optimization scheduling module 130 can be solved using a non-dominated sorting genetic algorithm (NSGA-II) or a multi-agent optimization algorithm (MAO). The NSGA is applicable to nonlinear optimization problems with multiple conflicting objectives and complex solution spaces. Its core features include fast non-dominated sorting, a congestion comparison operator, and an elitist strategy. MAO simulates the coordinated evolution of multiple agents under specific rules and is suitable for scenarios involving strategic competition or shared resource allocation between energy storage systems and the power market.

[0053] The solution process for diverse target models can be modeled and simulated using platforms such as Python and MATLAB, combining heuristic search with parallel computing to accelerate solution convergence. Of course, mixed-integer linear programming (MILP) can also be used to implement deterministic scheduling, or deep reinforcement learning algorithms can be used to achieve adaptive strategy evolution, although this embodiment does not impose specific limitations on this.

[0054] The charging and discharging strategy of the energy storage device can include a combination of parameters such as the energy charging amount, release amount, and standby state switching timing of the energy storage device at different time nodes. It has clear executability and market responsiveness, and provides dynamic adjustment capabilities for new energy stations, thereby achieving the unity of operational economy and system stability under the background of electricity price fluctuations and output uncertainty.

[0055] The planning and layout decision module 140 is coupled to the electricity price prediction module 110, the new energy output prediction module 120, and the energy storage optimization and scheduling module 130. It is used to combine the site selection criteria of the new energy station, the market-based electricity price prediction results, the new energy output prediction results, and the energy storage device charging and discharging strategy to solve and output the corresponding capacity configuration plan for the new energy devices and energy storage devices. Coupling means that the planning and layout decision module can simultaneously access the output results of the above three modules and use them as input parameters in the subsequent capacity configuration modeling and optimization analysis process, thus realizing a data-driven planning and design process.

[0056] The siting criteria for a new energy station refer to a collection of external information related to the proposed site's environment, resources, policies, and infrastructure. These factors can include land availability, grid access capacity, policy incentives, and distance to load centers. Land availability limits the deployment area and configuration of the device; grid access capacity reflects grid connection capabilities and grid adaptability; policy incentives include institutional support such as local subsidies, electricity pricing mechanisms, and green certificate quotas; and distance to load centers influences transmission losses, construction costs, and market proximity.

[0057] Under these multiple input conditions, planning and layout decision module 140 can implement capacity allocation decisions by constructing a capacity allocation optimization model for new energy devices and energy storage devices. This capacity allocation optimization model is a multi-objective, multi-constraint combined decision-making model. Its objective function can include minimizing investment costs, maximizing annual average profits, minimizing curtailment rates, and maximizing system flexibility. Constraints can include land area restrictions, grid-accessible capacity, maximum battery power and energy limits, and operation and maintenance budget constraints.

[0058] After the capacity configuration optimization model is constructed, it can be solved by calling the competitive game analysis method or the Nash equilibrium solution method. The competitive game analysis method is suitable for scenarios where multiple participants (such as new energy developers, power grid companies, and user-side load response units) are involved in resource or market occupation games. The modeling method can be based on strategy models such as non-cooperative games and Stackelberg games; the Nash equilibrium solution method can derive the optimal response combination of each party when multiple entities make interdependent decisions, so that the system reaches a stable equilibrium state. The construction method of the capacity configuration optimization model and the competitive game analysis method or the Nash equilibrium solution method are commonly used analysis methods in this field and will not be described in detail here.

[0059] In practical implementation, this can be achieved by building a distributed simulation platform, encapsulating prediction results, scenario parameters, economic models, and market mechanisms into a solution module, and using evolutionary games or iterative optimization methods to determine stable capacity allocations across multiple rounds of solution space. Furthermore, system dynamics methods can be used to simulate the dynamic evolution of large-scale station cluster configuration strategies, or machine learning methods can be introduced for empirical regression and strategy reinforcement.

[0060] The capacity configuration plan for new energy devices and energy storage devices can include key design parameters such as the optimal installed capacity of different types of equipment, energy storage system capacity distribution, deployment sequence and regional division, providing a structured support plan based on market feasibility and system benefits for the investment and construction of new energy projects, ensuring that the site design plan has good economy, system adaptability and long-term profitability.

[0061] Below, the digital and intelligent operation system of the new energy station in this example embodiment will be further explained.

[0062] In an exemplary embodiment of the present disclosure, an electricity price forecasting model can include an electricity price time series feature extraction network based on a time series model structure and a feature enhancement representation network based on an attention mechanism. The electricity price time series feature extraction network is used to learn the time series evolution law from the electricity price perturbation vector, while the feature enhancement representation network is used to dynamically weight key perturbation factors to enhance the model's ability to express important features. This combination of model structures enables the system to simultaneously consider both trend and mutation information in the electricity price series, thereby improving the stability and accuracy of multi-timescale electricity price forecast results.

[0063] refer to Figure 2 As shown, the electricity price prediction module 110 may include a first data fusion unit 111 and a first model scheduling unit 112, wherein: The first data fusion unit 111 is used to vectorize the various multi-source heterogeneous data in the electricity price disturbance characteristic data to obtain an electricity price disturbance vector; the electricity price disturbance characteristic data may be a data set from different data platforms with different formats, sampling granularities and update cycles. For example, the electricity price disturbance characteristic data may include historical electricity price records (such as the day-ahead average price, spot transaction price), meteorological factors (such as temperature, wind speed, air pressure, etc.), geographic location parameters (such as site latitude and longitude, altitude), fuel prices (such as coal, gas, and oil unit prices), power generation technology structure (such as the degree of wind-fire complementarity), power transmission capacity parameters (such as node voltage and current fluctuations), and regional load-supply curves, etc.

[0064] In the first data fusion unit 111, the raw electricity price disturbance feature data can be preprocessed by type identification and standardization. Normalization methods (such as Z-score standardization and Min-Max scaling) are used for numerical data, and one-hot encoding or embedding is used for discretization of categorical data. Subsequently, different data sources are uniformly mapped into a common feature space to construct a unified data representation structure. Fusion methods can adopt simple concatenation, weighted fusion, or fusion compression strategies based on dimensionality reduction algorithms (such as principal component analysis (PCA), t-SNE, and autoencoder) to ensure that the data input dimension is controllable and the information is fully expressed.

[0065] The first model scheduling unit 112 can be used to schedule the trained electricity price prediction model. Scheduling refers to dynamically loading model parameters and computation paths based on the system task plan and call priority, completing model instantiation and inference tasks in edge computing nodes or central processing units. The electricity price prediction model consists of two collaborative sub-networks. First, the electricity price time series feature extraction network performs time series modeling on the electricity price perturbation vector and extracts electricity price time series features. The electricity price time series feature extraction network can be modeled using a gated recurrent structure (such as a long short-term memory (LSTM) network), recursively updating states within multi-layer recurrent units to extract implicit time-dependent patterns in the sequence.

[0066] Next, the first model scheduling unit 112 can construct a weight matrix based on the aforementioned electricity price time series characteristics through a feature enhancement representation network to achieve focused modeling of important disturbance factors. The feature enhancement representation network is a weighted focusing module built on an attention mechanism. It generates an attention distribution matrix through matching calculations and softmax normalization operations. The attention distribution matrix can represent the model's estimate of the importance of disturbance factors at each moment. The feature enhancement representation network can be constructed using single-head attention or multi-head attention mechanisms, and positional encoding can also be introduced to enhance temporal sequence perception capabilities. This embodiment does not impose any specific restrictions on this.

[0067] In some optional implementations, a Transformer encoder structure can be used instead of LSTM as the main body of time series feature extraction to improve modeling parallelism; or trend features of local time windows can be extracted based on a gated convolutional neural network to reduce the gradient dissipation problem caused by long-term dependence; in addition, the feature enhancement representation network can also use an adaptive gating mechanism (such as SE module, Dynamic Routing) instead of the traditional attention mechanism for dynamic feature regulation. This embodiment does not impose any special restrictions on the network structure in the electricity price prediction model.

[0068] By vectorizing the electricity price disturbance feature data and inputting it into the electricity price prediction model that integrates time series modeling and attention mechanism, while extracting the electricity price time series features, focused modeling of key disturbance factors can be achieved. This can effectively improve the trend capture and mutation response capabilities of the electricity price prediction model at multiple time scales, thereby enhancing the accuracy, stability and interpretability of the electricity price prediction results.

[0069] In an exemplary embodiment of the present disclosure, a new energy output prediction model may include a spatial feature extraction network based on a convolutional neural network structure, an output time series feature extraction network based on a time series model structure, and a feature fusion network.

[0070] refer to Figure 3 As shown, the new energy output prediction module 120 may include a second data fusion unit 121 and a second model scheduling unit 122, wherein: The second data fusion unit 121 is used to vectorize the various multi-source heterogeneous data in the station output spatiotemporal data to obtain a meteorological space vector and a new energy output time series vector.

[0071] Site output spatiotemporal data refers to a collection of data reflecting the environmental status and operating behavior of renewable energy sites. This data can include historical output data, local meteorological data, atmospheric boundary layer data, irradiance data, and wind speed data. Historical output data represents power records from the equipment's operating history, captured at minute or hourly granularity. Local meteorological data includes indicators such as temperature, humidity, wind speed, air pressure, and wind direction, with significant spatial variation. Atmospheric boundary layer data reflects the vertical structure of wind resources. Irradiance data represents the intensity of solar radiation received per unit area. Wind speed data reflects the driving factors of wind farm operation.

[0072] In specific implementations, the second data fusion unit can perform time synchronization processing and missing value filling operations on various types of raw data to ensure the alignment consistency of data from different sources on the time axis. The data is then decoupled and classified according to spatial characteristics and temporal characteristics, and meteorological space vectors and new energy output time series vectors are constructed respectively. Among them, the meteorological space vector is based on regional meteorological maps or raster data, and a two-dimensional or three-dimensional matrix representation is constructed through pixel interpolation or regional grouping methods; the new energy output time series vector is constructed through a sampling window mechanism to construct a sliding time series of fixed length. During the vectorization process, normalization, standardization, embedded mapping and other methods can be used to ensure data dimension consistency and expression stability.

[0073] The second model scheduling unit 122 can be used to schedule the trained new energy output prediction model, so as to extract spatial features of the meteorological space vector through the spatial feature extraction network to obtain spatial features, and to perform time series modeling on the new energy output time series vector through the output time series feature extraction network to obtain time series features, and to fuse the spatial features and time series features through the feature fusion network to output the new energy output prediction result.

[0074] The spatial feature extraction network can be constructed based on a convolutional neural network (CNN) structure, which can include an input layer, convolutional layers, activation layers, pooling layers, and fully connected layers. The convolutional layer uses multiple convolution kernels to perform local perception operations on the input data, extracting the spatial correlation and pattern characteristics of regional meteorological variables. The activation layer often uses rectified linear units (ReLU) to introduce nonlinear expression capabilities. Pooling layers (such as max pooling and average pooling) can be used to reduce feature dimensionality and enhance feature abstraction. Finally, the fully connected layer outputs a spatial feature vector, which serves as one of the inputs to the subsequent fusion network.

[0075] The time series feature extraction network can be constructed based on a recurrent neural network structure, preferably a long short-term memory network or a gated recurrent unit. By performing time series analysis on historical renewable energy power data, dynamic features such as periodic changes in power output, short-term inertial fluctuations, and sudden disturbance responses can be identified. The time series feature extraction network structure models long-term dependencies through its hidden states and gating mechanism, thereby improving the extraction of nonlinear output features of renewable energy sources.

[0076] The feature fusion network can be used to achieve unified encoding and joint modeling of two types of heterogeneous features. The feature fusion network can include structures such as a feature concatenation module, an attention weighting module, and a fusion perception module. The feature concatenation module connects spatial features and temporal features in the feature dimension to form a joint feature representation; the attention weighting module can enhance the expressive power of key features by calculating the importance weights of each channel; and the fusion perception module can further model cross-modal interaction relationships based on a multilayer perceptron (MLP) or Transformer structure to generate the final output prediction vector.

[0077] In some optional implementations, a residual network (ResNet) or a densely connected network (DenseNet) can be used in the spatial feature extraction network to improve the transmission capability of deep features; a bidirectional LSTM can be used in the temporal feature extraction network to enhance the ability to model the association between previous and next moments; and a cross-modal attention mechanism or a graph convolutional network can be introduced in the feature fusion stage to further explore the structural connections between spatial regions.

[0078] The new energy output forecast results are output in the form of hourly power forecast values ​​for a period of time in the future, which can be used to support subsequent energy storage optimization scheduling and system operation strategy formulation. The output of the new energy output forecast model through the above network structure can effectively improve the prediction accuracy, enhance the generalization ability of the model, and have good adaptability to multi-source data.

[0079] By constructing a new energy output prediction model that includes a spatial feature extraction network, an output time series feature extraction network, and a feature fusion network, it is possible to effectively model the meteorological spatial characteristics and the time series variation patterns of new energy output respectively, and achieve efficient integration of the two types of features through a fusion mechanism, thereby improving the new energy output prediction model's ability to fit the relationship between complex meteorological disturbances and nonlinear power response, significantly improving the accuracy and robustness of new energy output prediction, and enhancing the predictability and controllability of new energy station operation and scheduling.

[0080] In an exemplary embodiment of the present disclosure, referring to Figure 4 As shown, the energy storage optimization scheduling module 130 may include an optimization model building unit 131 and an optimization solution unit 132, wherein: The optimization model construction unit 131 can be used to input the market-based electricity price forecast results and the new energy output forecast results as boundary conditions, and combine physical constraints, grid constraints and operation constraints as constraints to construct a multi-objective optimization model with economic, technical and social goals.

[0081] Boundary condition inputs can include forecasted electricity price and output sequences for several future time periods. These inputs have multi-timescale and dynamic characteristics, limiting the input domain of the multi-objective optimization model and constraining the operational context of the energy storage strategy. Economic objectives can include maximizing energy storage device revenue, maximizing peak-valley arbitrage income, and minimizing the payback period. Technical objectives involve extending the cycle life of energy storage batteries, maximizing load balancing rates, and optimizing grid fluctuation suppression capabilities. Social objectives can encompass minimizing carbon emissions, minimizing curtailment rates, and increasing regional energy self-sufficiency.

[0082] Physical constraints refer to the operating restrictions based on the energy storage device itself, including parameters such as the type of energy storage device (such as lithium batteries, sodium-sulfur batteries, flywheel energy storage, compressed air energy storage, etc.), initial capacity (expressed in rated power and energy capacity), battery cycle life (expressed in equivalent full charge and discharge times or total operating time), and charge and discharge efficiency (usually 80%-95%, taking into account energy conversion losses). These parameters jointly define the technical boundaries of the energy storage device and ensure that the optimization strategy can be implemented within the equipment's carrying capacity.

[0083] Grid constraints refer to the grid operating boundary conditions that energy storage devices must meet for grid-connected operation. These include, but are not limited to, maximum and minimum injectable power limits at the connection point, node voltage amplitude limits, current limits, frequency deviation tolerances, and response time requirements for frequency and voltage regulation. In some special scenarios, these constraints may also include access agreements that require energy storage systems to provide ancillary services (such as peak shaving, frequency regulation, and backup). For example, this may require that a certain percentage of energy storage capacity be reserved for grid dispatch during specific periods of time.

[0084] Operational constraints are primarily used to limit the scheduling feasibility of multi-objective optimization models in the time dimension. These constraints include charging and discharging power change rate limits for energy storage systems (to avoid sudden changes that could cause system shocks), minimum charging / discharging time limits, continuous charging and discharging time window limits, and interlocking control strategies (to avoid simultaneous charging and discharging). These constraints also include requirements for the coupling of predicted output and energy storage behavior, ensuring that the energy storage system can effectively regulate renewable energy generation within the output fluctuation range, avoiding scheduling deviations amplified by prediction errors.

[0085] The multi-objective optimization model can be constructed using a mathematical modeling language (such as Pyomo, GAMS, or AMPL). Its objective function and constraints can be nonlinear, non-convex, and represent mixed optimization problems with integer decision variables. These problems can also be simplified into a solvable form through linearization or piecewise linear approximation. Of course, fuzzy logic can also be introduced to express uncertainties, or robust optimization methods can be used to handle prediction errors to achieve the multi-objective optimization model, although this embodiment is not limited thereto.

[0086] The optimization solution unit 132 can be used to call a multi-objective optimization model, perform iterative calculations on the multi-objective optimization model based on a non-dominated sorting genetic algorithm or a multi-agent optimization algorithm, and output a charging and discharging strategy for the energy storage device that meets multiple constraints.

[0087] The non-dominated sorting genetic algorithm (NSGA) can be used as a swarm intelligence algorithm for multi-objective nonlinear optimization problems. It manages the solution set hierarchically through non-dominated sorting, and combines it with a crowding distance metric to maintain solution diversity and balanced distribution. During the algorithm's iterations, selection, crossover, and mutation operations are used to generate new generations of the population, while an elite retention mechanism ensures that the optimal solution is not lost. Using the NSG GA, multiple optimal solutions under different trade-offs can be obtained in a single run, forming a Pareto frontier and providing a rich set of alternatives for strategic selection in different scenarios.

[0088] Multi-agent optimization algorithms can simulate the information exchange and strategic game-playing between multiple autonomous agents. In energy storage system scheduling, different strategies can be considered "agents," with behaviors continuously adjusted within a simulated environment based on a payoff function. For example, multi-agent optimization algorithms can be constructed using actor-critic structures within a reinforcement learning framework, Q-learning algorithms, or evolutionary strategies based on game theory. These methods are suitable for problem scenarios with complex coupling relationships and high uncertainty, and offer advantages such as strong adaptability and outstanding online optimization capabilities.

[0089] The resulting energy storage device charging and discharging strategy can be a set of sequential control instructions that clearly specify the energy storage system's charging power, discharging power, and standby state behavior over several future time periods, ensuring that all constraints are met to optimize the overall system efficiency. This strategy can take the form of a time-by-hour schedule, a policy function, or a heuristic control rule. The energy storage device charging and discharging strategy can be directly used as input to the execution unit's scheduling instructions or as a reference for optimization in higher-level energy management systems.

[0090] By using electricity price forecast results and renewable energy output forecast results as boundary conditions, and constructing a multi-objective optimization model under multiple physical, grid and operational constraints, and combining it with a non-dominated sorting genetic algorithm or a multi-agent optimization algorithm for solution, it is possible to achieve an effective trade-off between multiple optimization objectives and generate a charging and discharging strategy for energy storage devices that takes into account economic, technical and social factors, thereby significantly improving the response efficiency and operational flexibility of renewable energy sites to electricity market price fluctuations and resource output uncertainties.

[0091] In an exemplary embodiment of the present disclosure, referring to Figure 5 As shown, the planning and layout decision module 140 may include a strategy input unit 141, a coupling modeling unit 142, and a model solving unit 143, wherein: The strategy input unit 141 can be used to obtain site selection condition information corresponding to the new energy site and receive resource attribute decision and market attribute decision inputs. The site selection condition information includes at least land availability, grid access capacity, policy incentive information and load center distance.

[0092] Among them, the site selection condition information refers to the deployment constraints and feasibility indicators of new energy sites at the spatial level. Land availability is used to evaluate the land development attributes, physical constructability and legal use rights of the proposed area. It can usually be identified and screened through the GIS geographic information system combined with terrain slope, land type, and ownership information; grid access capacity refers to the grid connection capability and remaining available capacity of the target access point, which can be obtained from the regional power grid dispatching platform and is limited by the trunk line capacity, voltage level and flow distribution; policy incentive information includes new energy subsidy policies, electricity price floating mechanisms, green certificate quota allocation plans, etc. issued at the central and local levels. This information can be obtained through the government data interface, and trends can also be extrapolated based on historical policy implementation cycles; load center distance is used to measure the length of the transmission path between the site and the main power consumption area, which has a decisive influence on the evaluation of the construction cost of the transmission channel and the power consumption path loss.

[0093] Resource attribute decision refers to a site assessment based on natural resource endowments, such as wind energy density distribution, solar radiation intensity, water availability, etc., which is usually obtained through historical measured data or satellite remote sensing data modeling; market attribute decision is an active judgment of the potential profit environment, such as the price volatility of the target area's electricity market, capacity trading activity, ancillary service pricing mechanism, energy storage market entry threshold, etc., reflecting the project's financial return potential under the future profit model.

[0094] The coupling modeling unit 142 can be used to construct a capacity configuration optimization model for new energy devices and energy storage devices based on site selection condition information, resource attribute decisions, market attribute decisions, market-based electricity price forecast results, new energy output forecast results, and energy storage device charging and discharging strategies.

[0095] The capacity allocation optimization model is a mathematical model that calculates capacity allocation based on the collaborative drive of multiple heterogeneous input factors. It is essentially a multi-objective, multi-constrained mixed-integer optimization problem. In this model, the installed capacity of new energy devices and energy storage devices is the primary decision variable. The objective functions typically include, but are not limited to, maximizing the average annual return on investment, maximizing the net present value (NPV), minimizing the curtailment rate, minimizing system carbon emissions, and maximizing peak-shaving capacity.

[0096] In the capacity allocation optimization model, the various input factors can be modeled and integrated as follows: electricity price forecasts serve as the electricity sales price boundary in the future revenue model; renewable energy output forecasts serve as the time series boundary for the amount of electricity each unit can provide; and energy storage device charging and discharging strategies constrain the energy limits that can be released or absorbed by the devices at each moment. Resource attribute decisions determine the upper limit of the deployable capacity of candidate sites within a specific spatial grid; and market attribute decisions provide supplementary constraints or incentive functions for pricing mechanisms and service models. The modeling approach can employ mixed-integer linear programming (MILP) or mixed-integer nonlinear programming (MINLP) to express the relationship between the aforementioned multi-type variables and the objective function. Alternatively, a data-driven regression model can be introduced to estimate the sensitivity and weight of each objective function. This embodiment does not specifically limit the modeling approach for the capacity allocation optimization model, as it is a common technical approach in the field and will not be further elaborated here.

[0097] The model solving unit 143 can be used to solve the capacity configuration optimization model through a competitive game analysis method or a Nash equilibrium solution method, and output an optimal capacity configuration plan for the new energy device and the energy storage device.

[0098] Competitive game analysis, based on the concept of non-cooperative game modeling in game theory, is applicable to strategic deduction when multiple entities, such as renewable energy developers, power grid companies, electricity retailers, and energy storage service providers, face resource sharing or conflicting benefits. Within this modeling framework, the capacity of renewable energy devices and energy storage devices serve as the strategy spaces for each player. Each player's payoff function is reflected through electricity price forecasting models and operational simulations, and their optimal responses form a dynamic strategy evolution path. This method typically iteratively solves for a Nash equilibrium to achieve a stable capacity distribution pattern in which none of the parties are willing to unilaterally change their strategies.

[0099] Nash equilibrium methods are primarily used in scenarios where there are cross-coupling benefits between participants but overall collaboration is limited. The core principle is to derive one's own optimal strategy based on the assumption that the other party's strategy is constant, ultimately finding the optimal intersection set for all participants under multiple strategy combinations. This can be achieved using strategy search and optimal solution approximation algorithms such as evolutionary game algorithms, multi-agent reinforcement learning, or approximate dynamic programming.

[0100] The output results include the installed capacity values, time-sharing layout priorities, equipment type selection, and investment cycle arrangements of new energy devices and energy storage devices at the target site. These strategies are highly executable, economically feasible, and highly compatible with the system, forming a complete capacity configuration optimization strategy that provides a quantifiable design basis for subsequent project implementation.

[0101] By uniformly coupling site selection information, resource attribute decisions, market attribute decisions, electricity price forecast results, output forecast results, and energy storage scheduling strategies, and using competitive game analysis or Nash equilibrium methods to solve the capacity configuration optimization model, it is possible to achieve a structural balance between the rationality of resource layout and market revenue response, and output a capacity configuration plan with multi-dimensional constraint adaptability, thereby improving the scientific nature of investment decisions in the planning stage of new energy and energy storage systems and their market adaptability in later operations.

[0102] In an exemplary embodiment of the present disclosure, referring to Figure 6 As shown, the digital intelligent operation system of the new energy station may further include a data acquisition and preprocessing module 150. The data acquisition and preprocessing module 150 may include a multi-source data interface unit 151, a data cleaning unit 152, and a data normalization unit 153, wherein: The multi-source data interface unit 151 can be used to connect to the data interfaces corresponding to the meteorological monitoring system, the power market data platform, the station monitoring system and the policy information library to collect electricity price disturbance characteristic data and station output spatiotemporal data.

[0103] Meteorological monitoring systems, including meteorological bureau data interfaces, satellite remote sensing systems, ground observation station networks, boundary layer radars, and micrometeorological sensor arrays, can provide real-time and historical meteorological factor data. Electricity market data platforms can include provincial power trading centers, regional spot trading platforms, and ancillary service market settlement systems. Their output data can include historical day-ahead electricity prices, real-time electricity prices, marginal electricity prices, load forecasts, and traded electricity volume. Site monitoring systems can collect state and analog data from new energy equipment using industrial communication protocols (such as Modbus and IEC 61850), including wind speed, irradiance, temperature, current, voltage, power generation, and operating status codes. The policy information database is a database interface for government affairs, including policy announcements and incentive measures issued by various departments.

[0104] Electricity price disturbance characteristic data may include any one or a combination of historical electricity price data, seasonal climate data, geographic location information, fuel cost data, power generation technology, power line data, and supply and demand balance data. Specifically, historical electricity price data can be extracted from multiple market dimensions such as day-ahead, real-time, and spot prices at hourly granularity; seasonal climate data can be mapped to meteorological season indices or cold and warm zone classification labels through timestamps; geographic location information can identify the physical environment of the station through latitude and longitude, elevation, administrative divisions, etc.; fuel cost data can be derived from spot quotes of fuels such as coal, natural gas, and heavy oil; power generation technology can represent the station type structure (such as the proportion of wind and solar power, centralized / distributed installed capacity type); power line data can cover transmission channel topology, voltage level, and flow direction; supply and demand balance data can be calculated by the dynamic difference between load forecast and supply capacity within the region.

[0105] Station output spatiotemporal data can include any one or a combination of historical output data, local meteorological data, atmospheric boundary layer data, irradiance data, and wind speed data. Historical output data can be recorded by the station monitoring system, while local meteorological data can be obtained from the station's own sensors or nearby observation points. Atmospheric boundary layer data can generally be generated through lidar and sounding system modeling. Irradiance data can be collected from ground stations or remote sensing satellite products (such as MODIS and Himawari). Wind speed data can be obtained from the unit's anemometer or a third-party wind resource assessment platform.

[0106] The data cleaning unit 152 can be used to perform data preprocessing on the electricity price disturbance characteristic data and the site output spatiotemporal data to obtain preprocessed electricity price disturbance characteristic data and site output spatiotemporal data, wherein the data preprocessing at least includes outlier removal, missing value filling and time alignment processing.

[0107] Outlier removal involves identifying and removing extreme values ​​that are inconsistent with the data distribution based on statistical distribution or rule-based models. This can be achieved using the IQR (interquartile range) rule, Z-score standardization, or dynamic interval judgment based on a sliding window. Missing value imputation reconstructs data for time slices or variable sampling interruptions. This can be achieved using strategies such as interpolation (linear or spline interpolation), forward imputation, mean imputation, regression imputation, or k-Nearest Neighbors (k-NN) imputation based on similar samples. Time alignment ensures that all data variables are aligned to a consistent time granularity (e.g., 10 minutes, 30 minutes, or 1 hour). This synchronization generates a structured time series data frame for input into the model architecture.

[0108] The data normalization unit 153 can be used to standardize the preprocessed electricity price disturbance characteristic data and site output spatiotemporal data to obtain normalized electricity price disturbance characteristic data and site output spatiotemporal data, and transmit them to the electricity price prediction module and the new energy output prediction module respectively.

[0109] The purpose of normalization is to unify the scales of variables of different physical dimensions so that they have equivalent weight in model calculations. Common normalization methods include Min-Max scaling to the interval [0, 1], Z-score normalization to a distribution with a mean of 0 and a standard deviation of 1, or logarithmic transformation for data variables with a power-law distribution. For discrete or categorical variables, label encoding, one-hot encoding, or embedding vector mapping can be used. The normalized data is encapsulated into a tensor structure according to the module interface definition format, facilitating batch inference and feature extraction as input in the electricity price forecast module and the renewable energy output forecast module.

[0110] By building a preprocessing process that integrates multi-source data collection, cleaning processing, and normalization and standardization, the integrity, consistency, and model input compatibility of electricity price disturbance characteristic data and site output spatiotemporal data can be ensured, effectively reducing the impact of data anomalies and errors on the performance of the prediction model, thereby improving the data processing efficiency of the overall system and the input quality of the prediction results, thereby improving the credibility and accuracy of the output results of subsequent prediction and optimization modules.

[0111] In an exemplary embodiment of the present disclosure, referring to Figure 7 As shown, the digital intelligent operation system of the new energy station may further include a market adaptation and dynamic update module 160, which may include a data scheduling control unit 161, a data cache and backtracking unit 162, and a data output management unit 163, wherein: The data scheduling control unit 161 can be used to dynamically determine the data collection frequency and data collection priority based on the update rate, data importance and model call requirements of different types of electricity price disturbance characteristic data and site output spatiotemporal data, and distribute them to the multi-source data interface unit and data cache and backtracking unit.

[0112] The update rate refers to the data refresh cycle of different data sources. For example, electricity spot market prices may be updated every 15 minutes, while meteorological model outputs may be updated hourly or daily. Generator operating data can be sampled at a frequency of seconds. Data importance refers to the degree to which each data variable affects the prediction accuracy of electricity price and renewable energy output forecast models. This can typically be determined through sensitivity analysis, feature importance assessment (such as SHAP value, information gain), or training error inversion. The model call requirement is determined by the type of forecast performed by the system in the current time period (such as intraday forecast, day-ahead forecast, and short-term scheduling optimization).

[0113] In specific implementations, the data scheduling control unit can establish a priority-based scheduling strategy matrix, categorizing data sources into multiple levels, such as high-frequency critical categories (such as real-time electricity price fluctuations and wind speed data), medium-frequency environmental categories (such as boundary layer temperature and atmospheric pressure), and low-frequency structural categories (such as geographic information and fuel prices). The corresponding sampling periods and synchronization frequencies are then set. Control logic can be implemented through a preset rule table, empirical parameter adjustments, or an adaptive data scheduling strategy based on reinforcement learning. The data scheduling control unit can also dynamically adjust data collection tasks by throttling, merging, and delaying them based on the current task load and computing resource status to prevent system resource overload.

[0114] The data caching and backtracking unit 162 can be used to batch cache and time index the collected electricity price disturbance characteristic data and site output spatiotemporal data according to the received data collection frequency and data collection priority, so as to support the historical data backtracking requirements of different models.

[0115] The cache structure can be designed based on a multi-layer partitioning mechanism, including a raw data layer, a cleaned data layer, and a normalized feature layer. Each layer is organized by time windows (e.g., the last hour, six hours, 24 hours, seven days) and spatial partitions (e.g., by site ID or electricity price node number) to support fast retrieval and differentiated storage strategies. To improve storage efficiency, cache units can utilize circular buffers, log-structured merge storage (LSM-tree), or key-value databases (e.g., RocksDB) for efficient writing and time-range reading.

[0116] Time index management can construct a multi-level index structure through timestamps and variable names, allowing different models to directly call historical fragment data as input or verification data when performing tasks such as sliding window prediction, rolling optimization, and error assessment. For example, in the day-ahead forecast, the electricity price forecast model may require the historical disturbance factor sequence of the last 30 days, and the new energy output forecast model needs to match the power generation output data under similar historical weather conditions. The system can directly intercept the corresponding paragraph based on the time index. To support model training and strategy evaluation, a label database can also be established in the cache to store historical true values ​​and model output results, realizing training sample generation and error backtracking analysis.

[0117] The data output management unit 163 can be used to extract the latest data of target priority from the data cache and backtracking unit according to the data collection priority, and output it synchronously to the electricity price prediction module and the new energy output prediction module in the model calling order.

[0118] The data output management unit is responsible for decoupling and controlling the flow of data from the cache to the model. Its core functions include data screening, format adaptation, batch scheduling, and anomaly detection. Data screening logic can be defined based on a sampling priority table and model input templates to extract valid data entries from the cache that match the current time window, time granularity, and spatial range. Format adaptation can ensure that data conforms to the neural network input format requirements through methods such as tensor reconstruction, dimensionality expansion, and missing value supplementation. Batch scheduling can use sliding window or time wheel scheduling strategies to uniformly allocate multiple model requests, avoiding redundant computations caused by repeated reads.

[0119] In some optional implementations, the data output management unit can also implement dynamic streaming data distribution through stream processing frameworks (such as Apache Flink and Apache Kafka Streams), or deploy microservice components on edge computing nodes to implement local caching and push functionality. This ensures that the system can continuously provide timely, prioritized, and uniformly formatted input data to the model module even in dynamic data updates, significantly improving prediction accuracy, response speed, and computing resource utilization.

[0120] By establishing a dynamic data scheduling mechanism based on update rate and priority, and combining it with cache management and index control strategies, we can achieve hierarchical caching, historical backtracking and on-demand distribution control of multi-source data. This can continuously ensure the timeliness and stability of model inputs in the high-frequency fluctuation environment of the power market, significantly improve the system's adaptability to dynamic changes in data streams, and thereby enhance the response speed and real-time decision-making capabilities of the digital operation system in the prediction, optimization and execution links.

[0121] It should be noted that although several modules or units of the digital intelligent operation system of the new energy station are mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be concretized in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided to be concretized by multiple modules or units.

[0122] In addition, in this exemplary embodiment, a control method for a digital intelligent operation system of a new energy station is also provided. Figure 8 As shown, the control method for the digital intelligent operation system of the new energy station may include: Step S810: Inputting the acquired electricity price disturbance feature data into an electricity price prediction model based on the fusion of a time series model and an attention mechanism through an electricity price prediction module to predict the market-based electricity price for future periods and outputting the market-based electricity price prediction results at multiple time scales; Step S820: Inputting the acquired spatiotemporal output data of the station into a new energy output prediction module based on time-series spatial feature fusion through the new energy output prediction module to predict the generated power of the new energy device of the new energy station and obtain a new energy output prediction result; Step S830: constructing a diversified target model based on the market-based electricity price forecast result and the new energy output forecast result through the energy storage optimization scheduling module, and generating a charging and discharging strategy for the energy storage device by solving the diversified target model; Step S840: Calculate the capacity configuration scheme corresponding to the output new energy device and the energy storage device by combining the site selection conditions of the new energy site, the market-based electricity price forecast result, the new energy output forecast result, and the energy storage device charging and discharging strategy through the planning and layout decision module.

[0123] The specific details of each step of the control method for the digital and intelligent operation system of the new energy station mentioned above have been described in detail in the corresponding digital and intelligent operation system of the new energy station, so they will not be repeated here.

[0124] It should be noted that although the steps of the method disclosed herein are depicted in a particular order in the accompanying drawings, this does not require or imply that the steps must be performed in that particular order, or that all steps must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one, and / or one step may be decomposed into multiple steps.

[0125] In addition, in an exemplary embodiment of the present disclosure, an electronic device capable of implementing a control method for a digital and intelligent operation system of a new energy station is also provided.

[0126] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods, or program products. Therefore, various aspects of the present disclosure may be implemented in the following forms: a complete hardware embodiment, a complete software embodiment (including firmware, microcode, etc.), or an embodiment combining hardware and software aspects, which may be collectively referred to herein as a "circuit," "module," or "system."

[0127] Refer to the following Figure 9 hereinafter, an electronic device 900 according to an embodiment of the present disclosure is described. Figure 9 The electronic device 900 shown is merely an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0128] like Figure 9 As shown, electronic device 900 is implemented as a general-purpose computing device. Components of electronic device 900 may include, but are not limited to, the aforementioned at least one processing unit 910, the aforementioned at least one storage unit 920, a bus 930 connecting various system components (including storage unit 920 and processing unit 910), and a display unit 940.

[0129] The storage unit stores program codes, which can be executed by the processing unit 910, so that the processing unit 910 performs the steps according to various exemplary embodiments of the present disclosure described in the above “Exemplary Method” section of this specification.

[0130] The storage unit 920 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 921 and / or a cache memory unit 922 , and may further include a read-only memory unit (ROM) 923 .

[0131] The storage unit 920 may also include a program / utility 924 having a set (at least one) of program modules 925, such program modules 925 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0132] Bus 930 may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0133] The electronic device 900 can also communicate with one or more external devices 970 (e.g., a keyboard, pointing device, Bluetooth device, etc.), one or more devices that enable a user to interact with the electronic device 900, and / or any device that enables the electronic device 900 to communicate with one or more other computing devices (e.g., a router, modem, etc.). This communication can occur via an input / output (I / O) interface 950. Furthermore, the electronic device 900 can communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network such as the Internet) via a network adapter 960. As shown, the network adapter 960 communicates with other modules of the electronic device 900 via a bus 930. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device 900, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0134] Through the description of the above embodiments, it will be readily understood by those skilled in the art that the example embodiments described herein can be implemented via software or via a combination of software and necessary hardware. Therefore, the technical solutions according to the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, or mobile hard drive) or on a network and includes several instructions for causing a computing device (such as a personal computer, server, terminal device, or network device) to execute the methods according to the embodiments of the present disclosure.

[0135] In exemplary embodiments of the present disclosure, a computer-readable storage medium is also provided, on which is stored a program product capable of implementing the aforementioned methods of this specification. In some possible embodiments, various aspects of the present disclosure may also be implemented in the form of a program product comprising program code. When the program product is executed on a terminal device, the program code is configured to cause the terminal device to execute the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of the present disclosure.

[0136] refer to Figure 10 As shown, a program product 1000 for implementing a control method for a digital and intelligent operation system for a new energy station according to an embodiment of the present disclosure is described. The program product 1000 can be a portable compact disc read-only memory (CD-ROM) and includes program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present disclosure is not limited thereto. In this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0137] The program product may utilize any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0138] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0139] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0140] Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0141] Furthermore, the figures above are merely illustrative of the processes included in the methods according to exemplary embodiments of the present disclosure and are not intended to be limiting. It is readily understood that the processes illustrated in the figures above do not indicate or limit the temporal order of these processes. Furthermore, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0142] Through the description of the above embodiments, it will be readily understood by those skilled in the art that the example embodiments described herein can be implemented via software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, or mobile hard drive) or on a network and includes several instructions to enable a computing device (such as a personal computer, server, touch terminal, or network device) to execute the methods according to the embodiments of the present disclosure.

[0143] Other embodiments of the present disclosure will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow from the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the claims.

[0144] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A digital intelligent operation system for new energy stations, characterized by: include: The electricity price prediction module is used to input the acquired electricity price disturbance feature data into the electricity price prediction model based on the fusion of time series model and attention mechanism to predict the market electricity price in the future period and output the market electricity price prediction results at multiple time scales; A new energy output prediction module is used to input the acquired spatiotemporal output data of the station into a new energy output prediction model based on the fusion of time-series and spatial features to predict the power generation of the new energy device of the new energy station and obtain a new energy output prediction result; an energy storage optimization scheduling module, connected to the electricity price prediction module and the new energy output prediction module, for constructing a diversified target model based on the market-based electricity price prediction results and the new energy output prediction results, and generating a charging and discharging strategy for the energy storage device by solving the diversified target model; A planning and layout decision module is coupled to the electricity price forecast module, the new energy output forecast module, and the energy storage optimization scheduling module, and is used to combine the site selection conditions of the new energy station, the market-based electricity price forecast results, the new energy output forecast results, and the energy storage device charging and discharging strategy to solve and output the capacity configuration plan corresponding to the new energy device and the energy storage device.

2. The system according to claim 1, wherein: The electricity price prediction model includes an electricity price time series feature extraction network based on a time series model structure and a feature enhancement representation network based on an attention mechanism; The electricity price prediction module includes: a first data fusion unit, configured to vectorize various multi-source heterogeneous data in the electricity price disturbance characteristic data to obtain an electricity price disturbance vector; The first model scheduling unit is used to schedule the trained electricity price prediction model to perform time series modeling on the electricity price disturbance vector through the electricity price time series feature extraction network, extract the electricity price time series features, and construct a weight matrix based on the electricity price time series features through the feature enhancement characterization network to achieve focused modeling of important disturbance factors and output the market-based electricity price prediction result.

3. The system according to claim 1, wherein: The new energy output prediction module includes a spatial feature extraction network based on a convolutional neural network structure, an output time series feature extraction network based on a time series model structure, and a feature fusion network: The new energy output prediction model includes: The second data fusion unit is used to vectorize the multi-source heterogeneous data in the station output spatiotemporal data to obtain a meteorological space vector and a new energy output time series vector; The second model scheduling unit is used to schedule the trained new energy output prediction model to extract spatial features from the meteorological space vector through the spatial feature extraction network to obtain spatial features, and to perform time series modeling on the new energy output time series vector through the output time series feature extraction network to obtain time series features, and to fuse the spatial features and the time series features through the feature fusion network to output the new energy output prediction result.

4. The system according to claim 1, wherein: The energy storage optimization scheduling module includes: an optimization model construction unit, configured to input the market-based electricity price forecast result and the renewable energy output forecast result as boundary conditions, and to construct a multi-objective optimization model with economic, technical, and social objectives in combination with physical constraints, grid constraints, and operational constraints as constraints; wherein the physical constraints, grid constraints, and operational constraints include any one or a combination of energy storage device type, initial capacity, battery cycle life, charge and discharge efficiency, equipment cost, grid load, and frequency regulation and standby requirements; The optimization solution unit is used to call the multi-objective optimization model, perform iterative calculations on the multi-objective optimization model based on a non-dominated sorting genetic algorithm or a multi-agent optimization algorithm, and output a charging and discharging strategy for the energy storage device that meets multiple constraints.

5. The system according to claim 1, wherein: The planning and layout decision module includes: a strategy input unit, configured to obtain site selection condition information corresponding to the new energy site and receive resource attribute decision and market attribute decision inputs, wherein the site selection condition information includes at least land availability, grid access capacity, policy incentive information, and load center distance; A coupling modeling unit, configured to construct a capacity configuration optimization model for a new energy device and an energy storage device based on the site selection condition information, the resource attribute decision, the market attribute decision, the market-based electricity price forecast result, the new energy output forecast result, and the energy storage device charging and discharging strategy; The model solving unit is used to solve the capacity configuration optimization model through a competitive game analysis method or a Nash equilibrium solving method, and output an optimal capacity configuration plan for the new energy device and the energy storage device.

6. The system according to claim 1, wherein: The digital intelligent operation system of the new energy station also includes a data acquisition and preprocessing module, which includes: a multi-source data interface unit, configured to connect to data interfaces corresponding to a meteorological monitoring system, an electricity market data platform, a station monitoring system, and a policy information repository, and collect the electricity price disturbance characteristic data and the station output spatiotemporal data; wherein the electricity price disturbance characteristic data includes any one or a combination of historical electricity price data, seasonal climate data, geographic location information, fuel cost data, power generation technology, power line data, and supply and demand balance data; and the station output spatiotemporal data includes any one or a combination of historical output data, local meteorological data, atmospheric boundary layer data, irradiance data, and wind speed data; a data cleaning unit, configured to perform data preprocessing on the electricity price disturbance characteristic data and the station output spatiotemporal data to obtain preprocessed electricity price disturbance characteristic data and station output spatiotemporal data, wherein the data preprocessing at least includes outlier removal, missing value filling, and time alignment processing; The data normalization unit is used to standardize the pre-processed electricity price disturbance characteristic data and the site output spatiotemporal data to obtain normalized electricity price disturbance characteristic data and site output spatiotemporal data, and transmit them to the electricity price prediction module and the new energy output prediction module respectively.

7. The system according to claim 6, characterized in that The digital and intelligent operation system of the new energy station also includes a market adaptation and dynamic update module, which includes: A data scheduling control unit is used to dynamically determine the data collection frequency and data collection priority according to the update rate, data importance and model call requirements of different types of electricity price disturbance characteristic data and station output spatiotemporal data, and distribute them to the multi-source data interface unit and data cache and backtracking unit; A data caching and backtracking unit, configured to cache and time-index the collected electricity price disturbance characteristic data and the site output spatiotemporal data in batches according to the received data collection frequency and the received data collection priority, so as to support the historical data backtracking requirements of different models; The data output management unit is used to extract the latest data of the target priority from the data cache and backtracking unit according to the data collection priority, and output it synchronously to the electricity price prediction module and the new energy output prediction module in the order of model calling.

8. A control method for the digital intelligent operation system of a new energy station according to any one of claims 1 to 7, characterized in that: include: The electricity price prediction module inputs the acquired electricity price disturbance feature data into an electricity price prediction model based on the fusion of a time series model and an attention mechanism to predict the market-based electricity price for future periods and output market-based electricity price prediction results at multiple time scales. The obtained spatiotemporal data of the station output is input into the new energy output prediction module based on the fusion of time series and space features through the new energy output prediction module to predict the power generation of the new energy device of the new energy station and obtain the new energy output prediction result; Constructing a diversified target model based on the market-based electricity price forecast result and the new energy output forecast result through an energy storage optimization scheduling module, and generating a charging and discharging strategy for an energy storage device by solving the diversified target model; The planning and layout decision module combines the site selection conditions of the new energy station, the market-based electricity price forecast results, the new energy output forecast results and the energy storage device charging and discharging strategy to solve the capacity configuration plan corresponding to the output new energy device and the energy storage device.

9. An electronic device, characterized in that: include: processor; and A memory having computer-readable instructions stored thereon, wherein the computer-readable instructions are executed by the processor to implement the method according to claim 8.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed by a processor, the method according to claim 8 is implemented.

Images