New energy station energy storage control system and method

Abstract

The invention provides a new energy station energy storage control system and method, and the system comprises a new energy output high-precision prediction module which is used for fusing local laser radar and numerical weather forecast data, and outputting an output predicted value; the SOC curve initialization module is used for initializing an energy storage SOC reference curve based on the output predicted value and the day-ahead market electricity price information; the real-time node electricity price prediction module is used for predicting a real-time electricity price trend; and the rolling optimization control module of the energy storage system is connected with other modules, and dynamically optimizes the charging and discharging instruction and the SOC state according to the output prediction value, the energy storage SOC reference curve and the electricity price prediction value of the real-time node electricity price prediction module. According to the invention, by fusing the multi-source prediction result and the real-time market information, on the premise of satisfying various operation and market constraint conditions, a highly adaptive energy storage dynamic optimization regulation and control mechanism is constructed, and key support is provided for economical operation of a new energy station in a spot market environment.

Description

Technical Field

[0001] The present invention relates to the field of smart grid and new energy dispatching control. Specifically, it relates to a new energy power station energy storage control system and method, and in particular, a new energy power station energy storage control system and method that combines high-precision output prediction and intraday spot price rolling optimization. Background Art

[0002] With the rapid development of new energy power generation globally, the installed capacity of new energy sources such as wind power and photovoltaic power has been continuously increasing and has become an important part that cannot be ignored in the power systems of various countries. However, new energy power sources have significant volatility and uncertainty. Especially in the scenario of high proportion penetration, it poses a severe challenge to the safety and economy of power grid dispatching. Currently, new energy power generation fully participates in power market transactions. In particular, the power spot market has put forward higher requirements for the power generation plan declaration, real-time tracking execution, and deviation assessment of new energy power stations. Since it is difficult to accurately predict the output of new energy, if the prediction deviation is large, it is easy to cause deviation assessment in the spot market, directly affecting the benefits of new energy power stations. At the same time, the power spot market introduces the nodal marginal price (LMP), forming a strong price signal driving mechanism, which further promotes new energy power stations to improve their output controllability and market response ability. On the other hand, as an important resource for new energy dispatching optimization, whether the dispatching strategy of energy storage can effectively respond to price signals determines the overall market competitiveness of new energy power stations.

[0003] However, traditional numerical weather prediction (NWP) mainly relies on mesoscale numerical model simulations, with low spatial resolution (10 - 30 kilometers) and slow update frequency (updated once every 1 to 3 hours), making it difficult to capture the rapidly changing local meteorological characteristics in the area where new energy power stations are located. For example, meteorological phenomena such as local cloud mass movement and wind speed mutation often occur within minutes and have a significant impact on the output. The NWP model cannot reflect these changes in a timely manner during prediction, resulting in a large error between the ultra-short-term prediction value and the actual power generation output. This prediction error not only affects the accuracy of the new energy output plan but also causes economic losses.

[0004] On the other hand, although the energy storage systems currently supporting new energy power stations have the ability to participate in power market regulation, most still adopt static charge-discharge strategies based on the previous day's spot price, ignoring the significant fluctuations and deviations between the intraday (real-time) spot price and the previous day's price. Under this strategy, the energy storage system is difficult to respond flexibly according to the real-time nodal price and fails to fully utilize the adjustment potential of "buying low and selling high", resulting in a low rate of return. Especially in the scenario of frequent fluctuations in nodal prices, it is easier for the fixed SOC (state of charge) trajectory execution to be out of touch with the system marginal value.

[0005] In addition, new energy power stations participating in the electricity spot market must be able to predict the nodal electricity prices for future periods accurately in order to declare and dispatch energy storage resources precisely. However, most existing power stations have weak prediction capabilities for real-time electricity prices and cannot provide effective decision-making support. At the same time, traditional strategies ignore the declaration risks and economic consequences brought about by prediction errors in power output and lack an optimal dispatching mechanism for uncertainty.

[0006] Therefore, there is an urgent need for a new control system for new energy power stations that can integrate local high-precision meteorological perception and numerical weather forecast information to achieve high-frequency and accurate short-term power prediction; at the same time, combined with the dual-time-scale prediction capabilities of day-ahead and real-time electricity prices, dynamically formulate and optimize energy storage charge and discharge strategies, thereby enhancing the overall profitability and risk resistance of new energy power stations. Summary of the Invention

[0007] Aiming at the deficiencies in the prior art, the purpose of the present invention is to provide an energy storage control system and method for new energy power stations.

[0008] According to an energy storage control system for new energy power stations provided by the present invention, it includes: A high-precision new energy power output prediction module, which is used to integrate local lidar and numerical weather forecast data and output a power output prediction value; An SOC curve initialization module, which initializes the energy storage SOC reference curve based on the power output prediction value and day-ahead market electricity price information; A real-time nodal electricity price prediction module, which is used to predict the real-time electricity price trend at the node where the new energy power station is located during the operating day; A rolling optimization control module of the energy storage system, which is connected to other modules and dynamically optimizes the charge and discharge instructions and the SOC state according to the power output prediction value, the energy storage SOC reference curve, and the electricity price prediction value of the real-time nodal electricity price prediction module.

[0009] Preferably, the high-precision new energy power output prediction module periodically collects three-dimensional wind field and cloud optical characteristic data covering the prediction area through lidar equipment deployed at new energy power generation stations, and extracts key time series features through physical modeling and data processing methods.

[0010] Preferably, the SOC curve initialization module initializes the energy storage SOC reference curve based on the day-ahead market electricity price information, the predicted new energy power output curve, and the declared power generation power curve before the operation of the spot market; the energy storage SOC reference curve provides an initial operation boundary for the subsequent dispatching module and serves as an auxiliary decision-making basis to guide the optimization of the declaration strategy.

[0011] Preferably, the real-time nodal price prediction module constructs a real-time nodal price prediction model based on grid operation data; by comprehensively considering the operation parameters announced by the grid, combining the known bidding space and real-time operation data, the real-time price trend of the node where the new energy power station is located within the operation day is predicted.

[0012] Preferably, the rolling optimization control module of the energy storage system integrates the following input information: - The high-precision predicted output sequence of new energy from the high-precision new energy output prediction module ; - The time series of the initialized SOC curve and the preliminary operation strategy from the SOC curve initialization module 、 ; - The nodal real-time price prediction curve of the real-time nodal price prediction module ; On this basis, with the goal of maximizing the profit, for each future rolling control window, the charge and discharge power sequence and the updated SOC state sequence are re-solved.

[0013] A new energy power station energy storage control method provided by the present invention includes: Step S1: By fusing the lidar data and numerical weather forecast data deployed in the new energy power station, the predicted value of new energy output is output; Step S2: Based on the predicted value of new energy output and the day-ahead spot price, an optimization model with the goal of maximizing profit is established, and the initial charge and discharge power sequence and the corresponding SOC state curve are solved; Step S3: According to the operation data and historical nodal price information released by the grid, the predicted price value is output; Step S4: According to the predicted output value, the SOC state curve, and the predicted price value, the charge and discharge power command and the SOC state are dynamically updated.

[0014] Preferably, the step S1 includes the following sub-steps: Step S1.1: Use a Doppler lidar to perform a stereoscopic scan, and adopt a combination of PPI and RHI modes to obtain radar volume scan data; Invert the horizontal and vertical wind speeds based on the VAD algorithm:

[0015] Wherein, is the horizontal wind speed, is the vertical wind speed, is the angle between the radar and the wind speed direction; Obtaining the optical thickness from the integrated backscattering coefficient:

[0016] where is the optical thickness, representing the scattering ability of the atmosphere; is the backscattering coefficient.

[0017] Constructing the feature matrix:

[0018] where is the time window length; is the number of vertical layers; Step S1.2: Perform time interpolation, spatial downscaling, and physical quantity standardization on the numerical weather prediction data to make it completely aligned with the lidar data in terms of time and space dimensions; For the original NWP data with a 1-hour time update frequency, use the cubic spline interpolation method to complement it to the same 5-minute interval as the radar data:

[0019] where is the interpolated meteorological variable, is the original time point, is the interpolation coefficient, obtained by fitting the hourly meteorological data values throughout the day; Enhance the spatial resolution through spatial dimension downscaling: Use bilinear interpolation or high-order interpolation to map the NWP data to the high-resolution grid within the radar field of view to ensure the alignment of feature points in each space:

[0020] where is the transformed local meteorological variable, is the interpolation weight, is the original NWP grid point; Constructing the final fused feature tensor: Align the processed NWP features with the five-dimensional optical and wind field features extracted by the radar according to time and space points to construct a unified input tensor:

[0021] where is the number of time steps, is the number of spatial sampling points, is the dimension of the fused features; Step S1.3: Construct a fusion prediction model based on the LSTM long short-term memory network to implement accurate wind power prediction; the fusion prediction model is designed with a multi-channel hybrid structure, and the lidar data and the processed numerical weather prediction data are respectively input into two parallel LSTM network channels to capture the temporal characteristics and evolution trends of different source data.

[0022] Preferably, the step S2 includes the following sub-steps: Step S2.1: Set and screen the input data and constraint conditions; the input data includes the day-ahead market electricity price sequence , the predicted new energy output curve , the electricity declaration curve already submitted in the day-ahead market and the operating constraint parameter set of the energy storage system; The sum of the output prediction value and the energy storage regulation amount does not exceed the limit:

[0023] where, is the net output power of the energy storage system at time t, with a positive value for discharging and a negative value for charging; is the over-profit recovery deviation coefficient; Step S2.2: Define the SOC recurrence expression:

[0024] where: is the state of charge of the energy storage system at time t; , are the charging and discharging powers at time t respectively; , are the charging and discharging efficiencies of the energy storage system respectively; is the rated capacity of the energy storage system; is the time interval; The SOC update satisfies the following operating boundary conditions:

[0025] ,

[0026] The above power variables also satisfy a mutually exclusive relationship: ; Step S2.3: Establish the profit maximization objective function: ; Step S2.4: Construct a time-series optimization problem with equality and inequality constraints to solve the SOC curve and the corresponding charging and discharging power sequences.

[0027] Preferably, the step S3 includes the following sub-steps: Step S3.1: Regularly obtain various grid operation data from the power trading center by building a data acquisition interface; Step S3.2: Establish an intraday nodal price prediction model for price prediction; Step S3.3: Establish a nodal price correction model to correct the price prediction results; Step S3.4: Combine the intraday nodal price prediction model and the nodal price correction model to obtain the final prediction model of the real-time nodal price.

[0028] Preferably, the step S4 includes the following sub-steps: Step S4.1: Define the revenue maximization objective function:

[0029] where, is the charging and discharging power of the energy storage system at time t, negative for charging and positive for discharging; is the predicted nodal price at time t; is the time interval; is the length of the rolling optimization window; is the starting time of the current optimization; Step S4.2: Adopt the rolling horizon optimization method, and each time use the current time as the starting point, roll forward for a period of time , and solve the optimal charging and discharging strategy for this interval; Step S4.3: Set key constraint conditions, including: Power constraint of the energy storage system:

[0030] SOC state update constraint:

[0031] SOC boundary limit:

[0032] Excess profit recovery range constraint: The real-time charging and discharging strategy is jointly verified with the new energy predicted output together: ; After the rolling optimization is completed, output the optimal charging and discharging instructions for the current control step and update the SOC state curve of the energy storage system in real time ; Input the updated SOC and the strategy sequence as the initial conditions for the next rolling cycle to form a complete closed loop.

[0033] Compared with the prior art, the present invention has the following beneficial effects: 1. By integrating multi-source prediction results and real-time market information, the present invention constructs a highly adaptive dynamic optimization and control mechanism for energy storage on the premise of meeting various operation and market constraint conditions, providing key support for the economic operation of new energy power stations in the spot market environment.

[0034] 2. The present invention proposes a new energy power station energy storage control system and method combining high-precision output prediction and rolling optimization of electricity prices, integrating local lidar and numerical weather prediction data, and achieving high-precision short-term output prediction from 15 minutes to 1 hour.

[0035] 3. The present invention can formulate a preliminary energy storage charge and discharge curve based on the day-ahead electricity price, then roll-predict the intra-day electricity price based on real-time market data, and further dynamically optimize the charge and discharge behavior and SOC path of the energy storage system; it realizes closed-loop linkage of "perception - prediction - optimization - execution" in the overall architecture, ensuring that new energy power stations have stronger profitability and operation stability in complex spot market environments.

[0036] 4. The control system of the present invention is mainly composed of four core functional modules, namely: a high-precision new energy output prediction module, an SOC curve initialization module, a real-time nodal electricity price prediction module, and a rolling optimization scheduling module for the energy storage system; among them, the high-precision new energy output prediction module is the data perception and modeling foundation of the system, improving the output prediction accuracy through multi-source meteorological information fusion and AI algorithms, providing data support for subsequent market participation; the SOC curve initialization module formulates a preliminary charge and discharge strategy based on the prediction results and the day-ahead electricity price, providing a reasonable starting point for energy storage scheduling; the bidding space fitting module extracts the feasible strategy boundary from historical behaviors to formulate a revenue-maximizing bidding path for the power station; the real-time nodal electricity price prediction module realizes the fitting of the bidding space through the day-ahead spot electricity price and the day-ahead prediction data released by the power grid, and accurately predicts the nodal electricity price at future moments according to the real-time operation data and change trends of the power grid during the real-time operation process; finally, the rolling optimization control module integrates all the above information, aiming at revenue optimization, and outputs the optimal charge and discharge operation instructions for the energy storage system. Through the coordinated action of the four modules, the system can achieve accurate prediction, flexible response and economic operation in the spot market environment with highly uncertain new energy output and rapid electricity price fluctuations, greatly improving the regulation ability and market revenue level of new energy power stations.

[0037] Other beneficial effects of the present invention will be elaborated in the specific implementation manners through the introduction of specific technical features and technical solutions. Those skilled in the art should be able to understand the beneficial technical effects brought by the technical features and technical solutions through these introductions. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Other features, objects, and advantages of the present invention will become more apparent by reading the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 is a system block diagram of the present invention.

[0039] Figure 2 is a flowchart of the operation of the high-precision prediction module for new energy output in the present invention.

[0040] Figure 3 is a flowchart of the operation of the SOC curve initialization module in the present invention.

[0041] Figure 4 is a flowchart of the operation of the real-time nodal electricity price prediction module in the present invention.

[0042] Figure 5 is a flowchart of the operation of the rolling optimization control module of the energy storage system in the present invention.

[0043] Figure 6 is a flowchart of the method of the present invention. SPECIFIC EMBODIMENTS

[0044] The present invention will be described in detail below with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be noted that those of ordinary skill in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0045] Referring to Figure 1 as shown, a new energy power station energy storage control system includes: a high-precision prediction module for new energy output, an SOC curve initialization module, a real-time nodal electricity price prediction module, and a rolling optimization scheduling module for the energy storage system.

[0046] Among them, the new energy output high-precision prediction module is the data perception and modeling foundation of the system. It improves the output prediction accuracy through multi-source meteorological information fusion and AI algorithms, providing data support for subsequent market participation; the SOC curve initialization module formulates a preliminary charge and discharge strategy based on the prediction results and the day-ahead electricity price, providing a reasonable starting point for energy storage scheduling; the bidding space fitting module extracts the feasible strategy boundary from historical behaviors and formulates a revenue-maximizing bidding path for the power station; the real-time nodal electricity price prediction module realizes the fitting of the bidding space through the day-ahead spot electricity price and the day-ahead prediction data released by the power grid, and accurately predicts the nodal electricity price at future moments according to the real-time operation data and change trend of the power grid during the real-time operation process; finally, the rolling optimization control module integrates all the above information, aims at revenue optimization, and outputs the optimal charge and discharge operation instructions for the energy storage system. Through the coordinated action of the four major modules, this system can achieve accurate prediction, flexible response and economic operation in the spot market environment with highly uncertain new energy output and rapid electricity price fluctuations, greatly improving the regulation ability and market revenue level of new energy power stations.

[0047] The specific module introduction is as follows: I. New energy output high-precision prediction module Refer to Figure 2 As shown, this module aims to achieve high-precision prediction of new energy output such as wind power and photovoltaic power at the minute-level time scale, and proposes a high-precision prediction method that fuses lidar observation data and Numerical Weather Prediction (NWP) data. This module constructs a complete spatio-temporal data processing and deep learning modeling process, covering the entire process from data collection, feature processing, model training to prediction output. Its operation steps are as follows: Step 1: Lidar three-dimensional field data collection and feature extraction This method uses lidar equipment deployed at new energy power generation stations to periodically collect three-dimensional wind field and cloud optical property data covering the prediction area, and extracts key time series features through physical modeling and data processing methods, providing input features with high spatio-temporal resolution and clear physical meaning for the subsequent prediction model. The key points of technical implementation are as follows: Data collection method: Use a Doppler lidar to perform stereoscopic scanning, and use a combination of two modes, PPI (Plan Position Indicator) and RHI (Range Height Indicator), to obtain radar volume scan data. The original echo parameters include but are not limited to: Radial wind speed , indicating the wind speed along the radar measurement direction; Backscatter coefficient , representing the reflection intensity of meteorological particles on radar waves; Signal-to-noise ratio , representing the clarity of radar signals.

[0048] Three-dimensional wind field inversion: Inverting horizontal and vertical wind speeds based on the VAD (Velocity Azimuth Display) algorithm:

[0049] Among them, is the horizontal wind speed, is the vertical wind speed, is the angle between the radar and the wind speed direction.

[0050] Optical thickness calculation: Obtaining the optical thickness by integrating the backscattering coefficient:

[0051] Among them, is the optical thickness, representing the scattering ability of the atmosphere; is the backscattering coefficient.

[0052] Characteristic matrix construction (in the form of time series):

[0053] Among them, is the time window length; is the number of vertical layers. This matrix contains information on optical thickness, wind speed, and their spatial gradients.

[0054] Step 2: Temporal and spatial alignment and dynamic fusion of multi-source data The temporal and spatial resolutions of numerical weather prediction (NWP) data are usually low and it is difficult to directly meet the requirements of short-term and nowcasting with minute-level accuracy. To achieve complete alignment with lidar data in both temporal and spatial dimensions, operations such as temporal interpolation, spatial downscaling, and physical quantity standardization need to be performed on NWP data. Through the above means, the coarse-scale meteorological data can be converted into high-quality input features with unified structure and matching dimensions, providing multi-source fusion feature data for subsequent deep learning models. The key points of the technical implementation are as follows: Temporal dimension alignment (improving temporal resolution): For the original NWP data with a 1-hour time update frequency, the cubic spline interpolation method is used to complement it to a 5-minute interval consistent with the radar data:

[0055] Among them, are the interpolated meteorological variables (such as temperature, wind speed, radiation, etc.), is the original time point, is the interpolation coefficient, which is obtained by fitting the hourly meteorological data values for the whole day.

[0056] Spatial dimension downscaling (enhancing spatial resolution): The original NWP grid is generally at a scale of 3 km, while the local area covered by the radar scan data is usually refined to the 100 m level. Bilinear interpolation or high-order interpolation is used to map the NWP data to the high-resolution grid within the radar field of view to ensure the alignment of the characteristics of each spatial point:

[0057] Among them, is the transformed local meteorological variable, is the interpolation weight, is the original NWP grid point.

[0058] Final construction of the fused feature tensor: Align the processed NWP features (such as temperature, wind speed, air pressure, radiation, etc.) with the five-dimensional optical and wind field features extracted by the radar in terms of time and spatial points to construct a unified input tensor:

[0059] Among them, is the number of time steps (such as the time window at 5-minute intervals), is the number of spatial sampling points (matching the radar scan grid), is the dimension of the fused features (including radar and NWP variables).

[0060] Through the above processing, the gap between radar data and meteorological forecast data in terms of time and space scales is bridged, and the low-frequency and low-resolution forecast data is upgraded to a format consistent with radar observations, thus providing comprehensive and synchronous multi-source meteorological inputs for the deep learning model.

[0061] Step 3: Construction and training of the hybrid machine learning model (LSTM) After completing the alignment of the time and space dimensions and the feature formatting of the multi-source meteorological data, it enters the model training stage. In this step, a fused prediction model is constructed based on LSTM (Long Short-Term Memory Network), making full use of the high-frequency local observation data of lidar and the mesoscale trend information of numerical weather prediction to achieve accurate minute-level wind power prediction. Specifically, the model design adopts a multi-channel hybrid structure: the lidar data and the processed NWP data are respectively input into two parallel LSTM network channels to capture the temporal characteristics and evolution trends of different source data.

[0062] The overall model adopts a parallel structure, mainly including two time-series input channels and one feature fusion output channel: Radar data channel: The input is the wind speed and turbulence feature sequences at different heights, such as:

[0063] Among them, represents the horizontal wind speed at time t and height z, represents the vertical wind speed at time t and height z, represents the optical thickness at time t and height z.

[0064] Use the LSTM network to extract its short-term dynamic features, and the output is:

[0065] Among them, is the time-series hidden feature model representation extracted by the LSTM of the lidar channel, is the feature matrix input by the lidar.

[0066] NWP data channel: The input is the NWP meteorological element sequence after interpolation and downscaling:

[0067] Among them, respectively represent NWP features such as wind speed, irradiance intensity, temperature, humidity, air pressure, cloud cover, etc. Similarly, input into the LSTM network to extract its time-series evolution features:

[0068] Among them, is the time-series hidden feature model representation extracted by the LSTM of the meteorological data channel, is the feature matrix input by the meteorological data.

[0069] Feature fusion and prediction output: Fuse the output features of the two channels, and a very short-term high-precision new energy output prediction model can be constructed through concatenation (Concat) or attention mechanism (Attention):

[0070] Or:

[0071] Among them, is the feature model representation after fusion. The predicted value of the wind power at a future moment is output through a fully connected network regression:

[0072] Among them, the predicted future wind power value, and the prediction step is .

[0073] The model uses the conventional mean squared error (MSE) as the loss function to measure the difference between the predicted value and the actual value; the model parameters are continuously updated through backpropagation and gradient optimization algorithms (such as Adam) to improve the prediction accuracy.

[0074] By constructing an LSTM hybrid model that integrates lidar and numerical weather prediction information, the response ability of short-term wind power prediction to sudden meteorological disturbances can be significantly enhanced, providing more accurate data support for improving the new energy grid connection and consumption efficiency and power market trading decisions.

[0075] II. SOC Curve Initialization Module Referring to Figure 3 as shown, in the scenario where the new energy power station participates in the spot power market trading, the charging and discharging behavior of the energy storage system and its state of charge (SOC) regulation strategy have a decisive impact on the economic benefits. The current spot market implements a two-tier electricity price mechanism, that is, the day-ahead market electricity price and the real-time (intraday) electricity price coexist. Among them, the day-ahead electricity price is announced one day in advance, reflecting the overall power supply and demand balance of the operating day and having a strong price guiding function. And according to a large amount of historical data, the trend of the real-time electricity price is usually highly correlated with the day-ahead electricity price, showing a certain degree of synchronization. Therefore, the energy storage system can determine the charging and discharging plan for the next day and the corresponding SOC curve based on the day-ahead electricity price on the day before operation.

[0076] The purpose of this module is to initialize the energy storage SOC reference curve that meets market rules, has reasonable operation logic, and maximizes the arbitrage potential based on the day-ahead market electricity price information, the predicted new energy output curve, and the declared power generation curve before the spot market operation. This SOC curve not only provides the initial operation boundary for the subsequent scheduling module but also can be used as an auxiliary decision-making basis to guide the optimization of the declaration strategy. The working process of the module mainly consists of the following four steps: Step 1: Preparation of input data and constraint conditions: The main inputs of the SOC initialization module include: : The day-ahead market electricity price sequence (with a 15-minute cycle); : The predicted new energy output curve; : The electricity declaration curve already submitted in the day-ahead market; Set of operating constraint parameters for the energy storage system (upper limits of charge and discharge power, upper and lower bounds of SOC, energy efficiency, initial state, etc.). Under the spot market mechanism, to avoid the behavior of "excessive profit-making", the trading institution usually sets a revenue recovery rule: the positive deviation between the declared power curve and the actual output + the energy storage discharge power shall not exceed a certain proportion, otherwise the revenue corresponding to the excess part will be recovered; however, if it is a negative deviation (under-prediction), no compensation will be given. Therefore, during the SOC initialization process, it should be ensured that the predicted output plus the energy storage regulation amount does not exceed the limit:

[0077] Among them, is the net output power of the energy storage system at time t, positive for discharge and negative for charge; is the excess profit recovery deviation coefficient, usually between 10% and 30%, set according to market rules.

[0078] Step 2: Modeling the recursive relationship of the SOC curve When planning the SOC curve, the principle of energy conservation needs to be followed, and the following SOC recursive expression is defined:

[0079] Among them: is the state of charge of the energy storage system at time t; , are the charge and discharge powers at time t respectively; , are the charge and discharge efficiencies (0 - 1) of the energy storage system respectively; is the rated capacity of the energy storage system; is the time interval, which is 15 minutes in the spot market.

[0080] The SOC update also needs to meet the following operating boundary conditions:

[0081] ,

[0082] The above power variables should also satisfy the mutual exclusion relationship (only charging or discharging can occur at the same time):

[0083] Step 3: Establish the objective function for maximizing revenue The goal of SOC curve derivation is: on the premise of ensuring compliance with market constraints and the physical operation constraints of the energy storage system, maximize the net income of the energy storage for charging and discharging in different time periods. The objective function can be expressed as:

[0084] This target value represents the total income obtained by the energy storage system from charging during low electricity price periods and discharging during high electricity price periods on the spot operation day.

[0085] Step 4: Initialize the solution of the SOC curve To solve the above SOC curve and the corresponding charge and discharge power sequences, a time-series optimization problem with equality constraints and inequality constraints needs to be constructed. This problem belongs to a typical mixed-integer nonlinear programming (MINLP) or linear programming (LP) problem. In the embodiment of the present invention, to reduce the computational complexity, linearization processing and relaxation strategies are preferably used for modeling.

[0086] The solution process includes the following steps: Time domain discretization: Divide the operation day into several time periods with fixed intervals (such as 15 minutes). Let the total number of time periods be T (corresponding to 96 points), and convert all variables into time-series vector form.

[0087] Expansion of state variables: Expand the decision variables of each time period into vector form, including:

[0088]

[0089]

[0090] Optimization modeling: Based on the goal of maximizing the profit function, the SOC recurrence formula, the mutual exclusion constraint of charge and discharge power, the SOC boundary conditions, the declaration deviation constraint, whether to charge / discharge, etc., construct a mixed-integer programming (MILP) model.

[0091] Optimization solution: Call a solver (such as CPLEX, SCIP) to solve, and output the optimal initial SOC curve and the preliminary operation strategy 、

[0092] Feasibility verification and smoothing processing: For jumps, spikes or discontinuous segments in the solution results, perform necessary post-processing (such as moving average, interpolation correction) to ensure actual executability.

[0093] Finally, the derived SOC curve not only meets the requirements of the market trading boundary, but also can maximize the economic benefits on the basis of meeting the physical operation constraints, ensuring the efficient operation of the energy storage system in the spot market environment.

[0094] III. Real-time nodal price prediction module Refer to Figure 4 As shown, this module aims to build a refined real-time nodal price prediction model based on grid operation data, providing accurate price signals and references for new energy power stations to participate in spot market trading and the optimal dispatching of energy storage systems. Considering that the nodal price in the spot market is affected by multiple factors such as the supply and demand situation of the internal grid of the node, the grid congestion status, and the bidding behavior of thermal power units, this module predicts the real-time price trend of the node where the new energy power station is located during the operation day by comprehensively considering the operation parameters announced by the grid, combining the known bidding space before the day and the real-time operation data. The working process of the module mainly consists of the following four steps: Step 1: Market operation data collection and calculation of bidding space Under the mechanism of the electricity spot market, the grid operator needs to release a large amount of operation-related data every day. This module obtains various grid operation data from the power trading center regularly through building a data collection interface, specifically including: System-adjusted load curve ; Output of non-marketized units ; Output of new energy units ; Output of hydropower and other types of units ; Load of external transmission channels ; Channel power flow and constraint information; Day-ahead spot price and real-time spot price; Day-ahead nodal price and real-time nodal price; Nodal congestion price; Structural information such as power generation and transmission maintenance plans; The data collection module supports interface docking with the data release platform of the power trading center, automatically pulling and updating data according to a preset cycle (such as every 5 minutes or 15 minutes) to ensure the timeliness and integrity of information.

[0095] Based on the above-mentioned operation data, the bidding space of the operation day can be deduced in combination with the market mechanism. The specific calculation is as follows:

[0096] Bidding space represents the available capacity interval of thermal power units that can participate in market dispatching, that is, the dominant bidding section where the real electricity price is formed. Against the background of the bids determined before the operation day, the real-time updated grid operation data is input into this bidding space, and the price change trend at the operation moment can be deduced, providing high-quality structured input for the downstream model.

[0097] Step 2: Intra-day nodal price prediction model Real-time nodal price Essentially, it is the result of solving the marginal price for a given bidding space based on factors such as the current system supply and demand status, network power flow, and congestion constraints. Considering that the nodal price is composed of the system marginal price and the nodal congestion price superimposed, we have:

[0098] where is the real-time spot price of node n; depends on the real-time net load demand and the thermal power bidding space; is closely related to the grid topology, channel load level, and node location.

[0099] Based on the above principle, this module uses the fixed bidding space from the day before and the real-time input quantity on the operating day to establish a nodal price prediction model:

[0100] where: are the real-time operation indicators of the power grid; are the historical marginal price data of node n; is a fitting function established through deep learning or machine learning models.

[0101] Step 3: Congestion risk assessment and nodal price correction model Since the nodal price is significantly affected by channel constraints, this module takes the main channel load as an important input variable to identify potential congestion risks. By analyzing the correlation between the channel load rate and the historical nodal congestion price, a congestion sensitivity index is constructed, and the system price prediction result is corrected:

[0102] where is a congestion model function established through deep learning or machine learning models.

[0103] Step 4: Model training and real-time nodal price prediction mechanism Based on the above results, the final prediction model for real-time nodal price is:

[0104] where the input features: Previously unified bidding parameters: including the unified dispatching load forecast of the whole network, non-market output forecast, power transmission plan, new energy forecast output, etc., to form the bidding space estimation for the operating day; Actual hourly load and output data for the operating day: including real-time unified dispatching load, output of various power sources, channel load, and operating status of important units; Historical node price data: including the day-ahead electricity price, real-time node electricity price, congestion price, etc. of each node, to capture the historical evolution law of the spatial price structure.

[0105] Target variable: The target of real-time node electricity price prediction is , where n represents the node number and t represents the time point (15-minute granularity).

[0106] Model structure: The model adopts a dual-stream modeling structure to balance the fitting of the system marginal electricity price and the description of the congestion risk between nodes: Main model : Used to fit the evolution law of the system-level marginal electricity price. Taking indicators such as the unified dispatching load at the whole network or regional level and the size of the bidding space as the main inputs, it predicts the basic trend of the system price; Correction model : Used to predict the node congestion premium (or discount) caused by factors such as the power grid topology structure and channel load changes. By inputting the real-time load of each channel, the historical node congestion price sequence, the channel-node association matrix, etc., it fits the congestion offset term.

[0107] The model structure can be implemented through various technical paths, including but not limited to: traditional machine learning models based on gradient boosting decision tree (GBDT), random forest (RF), etc., or time series modeling methods based on deep learning structures such as LSTM and Transformer. In actual deployment, flexible selection and parameter tuning can be carried out according to data integrity, operation efficiency, and prediction accuracy requirements.

[0108] The finally output node electricity price prediction curve , will be used as an important input feature for the subsequent revenue evaluation of new energy power stations and the energy storage optimal scheduling module, providing a basic support for accurately participating in the real-time spot market.

[0109] IV. Rolling Optimization Control Module of Energy Storage System As shown in Figure 5 , based on the previous calculation results, this module further constructs an optimization and control mechanism for the energy storage system for the rolling time period of the operating day to achieve dynamic update of the charge and discharge strategy and revenue improvement. Specifically, this module integrates the following input information: High-precision new energy predicted output sequence from the high-precision new energy output prediction module ; The initialized SOC curve from the SOC curve initialization module and the time series of the preliminary operation strategy 、 ; The predicted curve of the real-time nodal electricity price from the real-time nodal electricity price prediction module .

[0110] On this basis, this module aims to maximize the profit and, for each future rolling control window, re-solves the charge-discharge power sequence and the updated SOC state sequence .

[0111] Step 1: Definition of the objective function (profit maximization) The rolling optimization control takes maximizing the profit of the energy storage system as the core objective, and the objective function can be qualitatively described as follows:

[0112] Where: : The charge-discharge power of the energy storage system at time t, negative for charging and positive for discharging; : The predicted nodal electricity price at time t; : The time interval (usually 15 minutes); : The length of the rolling optimization window; : The starting time of the current optimization.

[0113] Step 2: Establishment of the rolling window mechanism This module adopts the rolling horizon optimization method. Each time, starting from the current time , it rolls forward for a period of time , and solves the optimal charge-discharge strategy for this interval. After each optimization is completed, the previous small segment (such as 1-2 time steps) is used as the execution control instruction, and the optimization is repeated at the next moment according to the latest prediction data to ensure that the strategy has dynamic adaptability.

[0114] Step 3: Key constraint conditions The following operating constraints need to be satisfied in the rolling optimization: Power constraint of the energy storage system:

[0115] SOC state update constraint:

[0116] SOC boundary limit:

[0117] Excess profit recovery range constraint: The real-time charging and discharging strategy needs to be jointly verified with the predicted output of new energy to ensure that the excess profit recovery range of the day-ahead declared output upper limit is not exceeded, that is:

[0118] Step 4: Control strategy output and SOC update After the rolling optimization is completed, the optimal charging and discharging instructions for the current control step are output , and the SOC state curve of the energy storage system is updated in real time . The updated SOC and strategy sequence will be used as the initial conditions for the next rolling cycle, forming a complete closed loop.

[0119] By integrating the multi-source prediction results and real-time market information, the present invention constructs a highly adaptive energy storage dynamic optimization and control mechanism on the premise of meeting various operation and market constraint conditions, providing key support for the economic operation of new energy power stations in the spot market environment.

[0120] Refer to Figure 6 As shown, a method for controlling energy storage in a new energy power station includes: Step S1: By integrating the lidar data and numerical weather forecast data deployed in the new energy power station, the predicted output value of new energy is output; Step S2: Based on the predicted output value of new energy and the day-ahead spot electricity price, an optimization model with the goal of maximizing revenue is established, and the initial charging and discharging power sequence and the corresponding SOC state curve are solved; Step S3: According to the operation data released by the power grid and the historical node electricity price information, the predicted electricity price value is output; Step S4: According to the predicted output value, SOC state curve and predicted electricity price value, the charging and discharging power instructions and SOC state are dynamically updated.

[0121] Those skilled in the art know that, in addition to implementing the system and its various devices, modules, and units provided by the present invention in the form of pure computer-readable program code, the method steps can be logically programmed to enable the system and its various devices, modules, and units provided by the present invention to be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers, etc., to achieve the same functions. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered as a kind of hardware component, and the devices, modules, and units included therein for implementing various functions can also be regarded as the structures within the hardware component; the devices, modules, and units for implementing various functions can also be regarded as either software modules for implementing the method or structures within the hardware component.

[0122] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Without conflict, the embodiments of the present application and the features in the embodiments can be combined arbitrarily with each other.

Claims

1. A new energy power station energy storage control system, characterized in that Including: A new energy output high-precision prediction module, which is used to fuse local lidar and numerical weather prediction data and output an output prediction value; An SOC curve initialization module, which initializes the energy storage SOC reference curve based on the output prediction value and the day-ahead market electricity price information; A real-time node electricity price prediction module, which is used to predict the real-time electricity price trend of the node where the new energy power station is located within the operating day; A rolling optimization control module of the energy storage system, which is connected to other modules and dynamically optimizes the charge and discharge commands and the SOC state according to the output prediction value, the energy storage SOC reference curve, and the electricity price prediction value of the real-time node electricity price prediction module.

2. The energy storage control system for a new energy power station according to claim 1, wherein The new energy output high-precision prediction module periodically collects three-dimensional wind field and cloud optical property data covering the prediction area through lidar equipment deployed in the new energy power generation station, and extracts key time series features through physical modeling and data processing methods.

3. The energy storage control system for a new energy power station according to claim 1, wherein, Before the operation of the spot market, the SOC curve initialization module initializes the energy storage SOC reference curve based on the day-ahead market electricity price information, the predicted new energy output curve, and the declared power generation power curve; the energy storage SOC reference curve provides an initial operation boundary for the subsequent dispatching module and serves as an auxiliary decision-making basis to guide the optimization of the declaration strategy.

4. The energy storage control system for a new energy power station according to claim 1, wherein The real-time node electricity price prediction module constructs a real-time node electricity price prediction model based on the grid operation data; by comprehensively considering the operation parameters announced by the grid, combining the known bidding space and real-time operation data in advance, it predicts the real-time electricity price trend of the node where the new energy power station is located within the operating day.

5. The energy storage control system for a new energy power station according to claim 1, characterized in that The rolling optimization control module of the energy storage system integrates the following input information: - High-precision prediction output sequence of new energy output by the high-precision prediction module for new energy output ; -The timing sequence of the initialized SOC curve of the SOC curve initialization module and the preliminary operation strategy and ; - Real-time nodal price prediction curve of the real-time nodal price prediction module ; On the basis above, aiming at maximizing the profit, for each future rolling control window, re-solve the charging and discharging power sequence and the updated SOC state sequence .

6. A new energy power station energy storage control method, based on the new energy power station energy storage control system according to any one of claims 1-5, characterized in that, Including: Step S1: Output the new energy output prediction value by fusing the lidar data deployed in the new energy power station and the numerical weather prediction data; Step S2: Based on the new energy output prediction value and the day-ahead spot electricity price, establish an optimization model with the goal of maximizing revenue, and solve the initial charge and discharge power sequence and the corresponding SOC state curve; Step S3: Output the electricity price prediction value according to the operation data released by the grid and the historical node electricity price information; Step S4: Dynamically update the charge and discharge power commands and the SOC state according to the output prediction value, the SOC state curve, and the electricity price prediction value.

7. The energy storage control method for a new energy power station according to claim 6, characterized in that The said Step S1 includes the following sub-steps: Step S1.1: Use a Doppler lidar to perform a stereoscopic scan, and adopt a combination of PPI and RHI modes to obtain radar volume scan data; Invert the horizontal and vertical wind speeds based on the VAD algorithm: Among them, is the horizontal wind speed, is the vertical wind speed, is the angle between the radar and the wind speed direction; Obtain the optical thickness by integrating the backscattering coefficient: Among them, is the optical thickness, representing the scattering ability of the atmosphere; is the backscattering coefficient; Construct a feature matrix: Among them, is the time window length; is the number of vertical layers; Step S1.2: Perform time interpolation, spatial downscaling, and physical quantity standardization on the numerical weather prediction data to make it completely aligned with the lidar data in terms of time and space dimensions; For the original NWP data with a time update frequency of 1 hour, use the cubic spline interpolation method to complete it to the same 5-minute interval as the radar data: Among them, is the interpolated meteorological variable, is the original time point, is the interpolation coefficient, which is obtained by fitting the meteorological data values of the whole day and hour; Improve the spatial resolution through spatial dimension downscaling: Use bilinear interpolation or high-order interpolation to map the NWP data to the high-resolution grid within the radar field of view to ensure the alignment of the features of each spatial point: Among them, is the transformed local meteorological variable, is the interpolation weight, is the NWP original grid point; Construct the final fusion feature tensor: Align the processed NWP features with the five-dimensional optical and wind field features extracted by the radar at time and spatial points to construct a unified input tensor: wherein, is the number of time steps, is the number of spatial sampling points, is the dimension of the fused feature; Step S1.3: Construct a fusion prediction model based on the LSTM long short-term memory network to implement accurate wind power prediction; the fusion prediction model is designed with a multi-channel hybrid structure, and the lidar data and the processed numerical weather prediction data are respectively input into two parallel LSTM network channels to capture the temporal features and evolution trends of different source data.

8. The energy storage control method for a new energy power station according to claim 7, wherein The step S2 includes the following sub-steps: Step S2.1: Set and screen input data and constraint conditions; the input data includes the day-ahead market electricity price sequence , the predicted new energy output curve , the electricity quantity declaration curve already submitted in the day-ahead market and the operation constraint parameter set of the energy storage system; The sum of the output prediction value and the energy storage regulation amount does not exceed the limit: wherein, is the net output power of the energy storage system at time t, with a positive value indicating discharging and a negative value indicating charging; is the over - profit recovery deviation coefficient; Step S2.2: Define the SOC recurrence expression: Where: is the state of charge of the energy storage system at time t; , are the charging and discharging powers at time t, respectively; , are the charging and discharging efficiencies of the energy storage system, respectively; is the rated capacity of the energy storage system; is the time interval; The SOC update satisfies the following operating boundary conditions: ， The above power variables also satisfy a mutually exclusive relationship: ； Step S2.3: Establish an objective function for maximizing revenue: ； Step S2.4: Construct a temporal optimization problem with equality constraints and inequality constraints to solve the SOC curve and the corresponding charge and discharge power sequences.

9. The energy storage control method for a new energy power station according to claim 8, wherein The step S3 includes the following sub-steps: Step S3.1: Regularly obtain various grid operation data from the power trading center by building a data acquisition interface; Step S3.2: Establish an intraday nodal price prediction model to predict the electricity price; Step S3.3: Establish a nodal price correction model to correct the electricity price prediction result; Step S3.4: Combine the intraday nodal price prediction model and the nodal price correction model to obtain the final prediction model of the real-time nodal price.

10. The energy storage control method for a new energy power station according to claim 9, characterized in that The step S4 includes the following sub-steps: Step S4.1: Define an objective function for maximizing revenue: wherein, is the charging and discharging power of the energy storage system at time t, negative for charging and positive for discharging; is the predicted nodal electricity price at time t; is the time interval; is the rolling optimization window length; is the current optimization starting time; Step S4.2: Adopt a rolling horizon optimization method. Each time, starting from the current time as the starting point, roll forward for a period of time , and solve the optimal charging and discharging strategy for this interval; Step S4.3: Set key constraint conditions, including: Energy storage system power constraint: SOC state update constraint: SOC boundary limit: Excess profit recovery range constraint: Real-time charge and discharge strategy combined with new energy predicted output Check together: ； Step S4.4: After the rolling optimization is completed, output the optimal charge and discharge command for the current control step size , and update the SOC state curve of the energy storage system in real time ; Input the updated SOC and the policy sequence as the initial conditions for the next rolling cycle to form a complete closed loop.

Images