Photovoltaic energy storage charging intelligent control method and system and storage medium

Abstract

The invention belongs to the technical field of new energy power generation and energy storage control, and discloses a photovoltaic energy storage charging intelligent control method which comprises the following specific steps: S1, dynamic data acquisition and standardization processing; s2, trend prediction and operation mode identification; and S3, generating a self-adaptive optimization strategy and performing closed-loop control. The illumination sensor, the voltage or current acquisition circuit, the SOC detection assembly, the electricity price information receiver and the meteorological data receiver form a multi-source data acquisition system, and the multi-source data acquisition system is intensively input into the local processor through the acquisition bus, so that synchronous acquisition and standardized processing of operation data are realized, and a formula is dynamically updated in combination with the charge state of the energy storage unit. After abnormal values are eliminated and data units are unified, the high-precision SOC value can be calculated in real time, the cumulative influence of measurement errors of a single sensor is remarkably reduced, and the stability and reliability of energy storage state monitoring under the dynamic working condition are ensured.

Description

Technical Field

[0001] This invention belongs to the field of new energy power generation and energy storage control technology, specifically a photovoltaic energy storage charging intelligent control method, system and storage medium. Background Technology

[0002] With the widespread deployment of distributed photovoltaic power generation, energy storage technology, and electric vehicle charging facilities, photovoltaic energy storage charging systems are playing an increasingly prominent role in energy consumption, power regulation, and energy cost optimization. By regulating the storage and release of photovoltaic power in local energy storage units, it is possible not only to improve the utilization rate of new energy sources, but also to perform peak shaving and valley filling during peak grid load periods and charge at low prices during off-peak hours, thereby achieving a dual improvement in economic benefits and energy utilization efficiency. However, photovoltaic power generation is affected by many factors such as sunlight conditions, weather changes, and load fluctuations, and its power output has obvious intermittency and volatility, which poses a challenge to the real-time adjustment of energy storage charging and discharging strategies.

[0003] In existing technologies, some photovoltaic energy storage charging systems rely on a single sensor or a single measurement path to obtain operational data, which is susceptible to noise interference, temperature drift, and sampling delay. This results in insufficient accuracy in estimating the state of charge (SOC) of the energy storage unit, thereby affecting the rationality of the scheduling strategy. At the same time, many systems only use historical statistical methods in the formulation of scheduling strategies, lacking the ability to predict changes in irradiance, electricity price fluctuations, and load demand in the short term, which leads to a lag in the system's response to sudden weather changes or rapid changes in market electricity prices.

[0004] In addition, existing charging and discharging controls are mostly single-objective optimizations, often favoring either economy or safety, failing to achieve a balance between the two. Furthermore, the execution control is mostly an open-loop structure, lacking an adaptive correction mechanism based on real-time feedback, resulting in decreased energy storage utilization efficiency, shortened equipment lifespan, and increased operating costs. Summary of the Invention

[0005] The purpose of this invention is to provide a photovoltaic energy storage charging intelligent control method, system, and storage medium to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a photovoltaic energy storage charging intelligent control method, the specific steps of which are as follows: S1. Dynamic Data Acquisition and Standardized Processing: Light intensity is acquired through a light sensor, output power of the photovoltaic array is acquired through a voltage or current acquisition circuit, state of charge and temperature of the energy storage unit are acquired through a SOC detection component, real-time electricity price is acquired through an electricity price information receiver, and weather forecast data is acquired through a meteorological data receiver. The data is transmitted to the local processor via the acquisition bus. The processor performs timestamp appending, outlier removal and unit standardization processing, and calls the dynamic update formula of the energy storage unit's state of charge to calculate the SOC in real time. S2. Trend Prediction and Operating Mode Recognition: The processor calls the prediction algorithm code in the memory, uses the light intensity trend prediction model to predict the short-term changing trends of light intensity, electricity price level and load demand, and transmits the prediction results to the pattern recognition logic circuit. Based on the SOC, electricity price, load prediction data and preset judgment conditions, the operating mode is identified as one of self-use priority, peak shaving and valley filling, electricity price arbitrage or off-grid standby. S3. Adaptive Optimization Strategy Generation and Closed-Loop Control: The processor loads the operating safety constraint parameters in the memory, including but not limited to the upper and lower limits of SOC, battery temperature control threshold, and demand control threshold, to construct a multi-objective optimization model for photovoltaic energy storage charging and discharging. Based on the prediction results and real-time feedback, it calculates the charging and discharging power and timing control strategy, and sends it to the energy storage converter and load control terminal through the communication interface. It also collects execution feedback signals. If the prediction deviation exceeds the threshold, it dynamically adjusts the weight of the optimization model to achieve closed-loop linkage between prediction and optimization.

[0007] Preferably, the specific steps of dynamic data acquisition and standardization processing in S1 are as follows: S11 acquires light intensity data through a light sensor to ensure that the sampling frequency meets the input requirements of the prediction model; the instantaneous output power of the photovoltaic array is measured through voltage acquisition circuit and current acquisition circuit, and a synchronous sampling method is used to reduce phase error; The SOC detection component is used to collect the real-time state of charge and temperature of the energy storage unit. Temperature acquisition is used for temperature control judgment and SOC correction. Real-time electricity price data from the grid side is obtained by the electricity price information receiver, and forecast information including sunlight trend, temperature, humidity and weather type is obtained by the meteorological data receiver. All the above-mentioned collected data are uniformly encoded and transmitted to the local processor through the acquisition bus to ensure data format consistency and time alignment, so as to realize the centralized input of photovoltaic power generation, energy storage status, grid price and environmental forecast information. After receiving the centralized input data, the S12 local processor first adds a collection timestamp to each data item to ensure the temporal consistency of multi-source data; then it performs outlier removal, including extreme values ​​caused by sensor acquisition anomalies, communication errors, and sudden environmental changes. The removal rules are preset in memory and can be dynamically adjusted according to the operating status; subsequently, it unifies the units of various data, such as power to watts and temperature to degrees Celsius, to ensure the compatibility of physical quantities in the calculation process. After data preprocessing, the processor calls the dynamic update formula for the state of charge of the energy storage unit, and uses the photovoltaic output power, load power, energy storage unit temperature and historical SOC value to calculate the current SOC. The calculated SOC value is used as the core input for subsequent trend prediction analysis and operation strategy optimization, ensuring that control decisions are based on accurate and verifiable energy storage state information. The formula for dynamically updating the state of charge of an energy storage unit is: In the formula: State of charge (%) at time t+Δt; State of charge at time t (%) : Charging efficiency and discharging efficiency (0~1); Charging power and discharging power (kW); Rated capacity of energy storage system (kWh); : Time interval (h); Formula origin and derivation: Based on the law of conservation of energy, the SOC value is updated after the energy increase during charging and the energy decrease during discharging are converted according to efficiency.

[0008] Preferably, the specific steps of trend prediction and operation mode recognition in S2 are as follows: The S21 local processor calls the prediction algorithm code pre-stored in the memory to execute the light intensity trend prediction model. The model takes real-time light intensity data after dynamic data acquisition and standardization processing, electricity price information and user load data as input, and uses a sliding time window method to call historical data of the past 10 to 30 minutes. The sampling period is 1 minute, and the prediction time span is the next 15 minutes to 1 hour. The calculation results include the fitted value of the light intensity change curve, the trend of electricity price fluctuation and the trend of load demand change, and also output the trend direction (rising, falling or stable). To address sensor drift or sudden environmental changes, the model adjusts its weights based on the latest feedback data after each round of calculation to maintain prediction accuracy. The above prediction results provide a forward-looking reference for energy storage scheduling and can effectively reduce the risk of strategy lag caused by rapid fluctuations in sunlight, changes in electricity prices, or sudden increases in load. The principle formula of the light intensity trend prediction model is: , In the formula: Predict the light intensity (W or m²) at time point t. Lighting prediction (W or m²) based on sliding regression; Light forecast (W or m²) based on seasonal exponential smoothing. The fusion weight coefficients (0~1) are determined by cross-validation. Formula origin and derivation: Sliding regression is suitable for capturing short-term trends, seasonal index smoothing is suitable for cyclical changes, and the weighted fusion of the two can take into account both short-term changes and seasonal patterns. Operational verification: Using historical photovoltaic monitoring data (5-minute resolution), implement dual-model prediction in MATLAB or Python, and compare the mean square error (MSE) of the single prediction method to verify the improved accuracy of the fusion prediction. The S22 processor transmits the trend prediction results output by S21 to the pattern recognition logic circuit, and inputs them into the pattern determination module along with the current energy storage unit SOC value, real-time electricity price, and load prediction data. The mode determination is based on preset determination conditions and priority rules in the memory: when multiple operating conditions are met simultaneously, the priority order is safe operation (SOC protection) > peak shaving and valley filling > electricity price arbitrage > self-consumption priority > off-grid backup. During the determination process, multiple parameters such as SOC safety range (upper and lower limits), electricity price threshold, load change rate, and prediction confidence are considered. When the input data is missing or the prediction accuracy is lower than the set threshold, it automatically switches to the fallback mode to ensure operational safety and continuity. The final mode recognition result will be used as the mode input for the photovoltaic energy storage charging and discharging multi-objective optimization model in adaptive optimization strategy generation and closed-loop control, realizing the closed-loop connection between prediction analysis and strategy optimization.

[0009] Preferably, the specific steps of adaptive optimization strategy generation and closed-loop control in S3 are as follows: The S31 processor loads runtime security constraint parameters from memory, the parameters including: SOC upper and lower limits (unit: %, range: 10%~90%, obtained from battery type and life curve calibration); Battery temperature control threshold (unit: °C, range: -10℃~50℃, determined by the cell manufacturer's recommended value and actual thermal stability). Demand control threshold (unit: kW, range: 0 to rated power limit, determined by distribution capacity and contract demand constraints). After acquiring the operating mode and prediction results output by trend prediction and operating mode recognition, as well as the real-time operating data of dynamic data acquisition and standardized processing, the processor calls the photovoltaic energy storage charging and discharging multi-objective optimization model for calculation. The model is solved by weighting the economic objectives (minimizing operating costs), safety objectives (not exceeding SOC and temperature limits), and stability objectives (minimizing load fluctuations). The weighting coefficients βi are obtained by calibrating through historical operating data and operational strategies. The model calculates and outputs the charging and discharging power (unit: kW) that meets the safety constraints and the corresponding timing control strategy, ensuring that the energy storage system balances safety and economy under the given operating mode; The charging and discharging power and timing control strategy obtained by S32 optimization calculation are sent to the energy storage converter and load control terminal through the communication interface, driving the relevant equipment to perform charging and discharging operations according to the optimization results; During execution, the processor synchronously collects the output power of the energy storage converter, the actual load power, and the status data of the energy storage unit, and compares them in real time with the prediction results in S31. When the execution deviation is detected to exceed the set threshold (e.g., power deviation > 5%, SOC deviation > 3%), the processor triggers a dynamic weight adjustment mechanism. The weight coefficient βi of the photovoltaic energy storage charging and discharging multi-objective optimization model is adaptively corrected so that the weight of the safety objective is increased or the weight of the economic objective is decreased, so as to prioritize the safety of operation. The updated weight parameters take effect immediately, the control strategy is recalculated and redistributed, achieving a closed-loop linkage between prediction and optimization strategies; To prevent frequent adjustments from causing system oscillations, a minimum time interval (e.g., 5 minutes) and a maximum single adjustment range (e.g., no more than 10%) are set for weight adjustments. This mechanism can significantly improve the system's response accuracy and operational stability under conditions such as sudden load changes, rapid weather changes, and abnormal electricity price fluctuations. The principle formula of the photovoltaic energy storage charging and discharging multi-objective optimization model is as follows: , In the formula: : Comprehensive optimization of objective function value (dimensionless weighted sum); The weight coefficients (real numbers, 0~1) of each optimization objective satisfy the following conditions: ; Electricity cost per unit time (in yuan or kWh); : The difference between the current SOC and the target SOC (%); Battery health risk factor (%) or cycle, calculated from the number of cycles and the percentage of deep discharge; System power loss (kW) is estimated from the losses of the converter and transmission lines; Formula origin and derivation: This formula originates from the multi-objective linear weighted method in operations research. It quantifies economy, energy storage stability, equipment life and efficiency as independent objectives, and after dimensional normalization, it is summed by weight to form a single optimization objective.

[0010] The present invention also provides a photovoltaic energy storage charging intelligent control system based on the above method, and the system includes: A. Data acquisition and processing unit, including light sensor acquisition component, voltage or current acquisition circuit, SOC detection component, electricity price information receiver and meteorological data receiver. Each component is connected to the local processor through acquisition bus. It is used to acquire multi-source operating data and perform timestamp appending, outlier removal and unit standardization processing. The processor is configured to call the energy storage unit's state of charge dynamic update formula to calculate SOC. B. Trend prediction and pattern recognition unit, which stores prediction algorithm code in a memory connected to the processor and is executed by the processor. The unit is configured to call the light intensity trend prediction model to generate short-term trend prediction results of light intensity, electricity price and load, and determine the operating mode based on the prediction results and SOC, electricity price and load data. C. Strategy optimization and execution control unit, including an optimization computing processor, a communication interface and an execution drive circuit. The optimization computing processor is configured to call the photovoltaic energy storage charging and discharging multi-objective optimization model to generate charging and discharging power and timing control strategies. The communication interface sends the strategies to the energy storage converter and load control terminal, and collects execution feedback signals to adjust the optimization model weights.

[0011] Preferably, the data acquisition and processing unit includes: (1) The data acquisition and processing unit includes the following components: Light sensor acquisition component: used to measure the instantaneous light intensity received on the surface of the photovoltaic array. The unit is W or m2, and the value range is 0~1200W or m2; Voltage acquisition circuit: Acquires the terminal voltages of the photovoltaic array and energy storage units. ,unit Range 0-500 (Energy storage); Current acquisition circuit: Acquires the output current of the photovoltaic array and the charging and discharging current of the energy storage. , Unit A, range 0~500A; SOC detection component: measures the state of charge (SOC) and temperature of the energy storage unit. SOC is measured in percent and ranges from 0 to 60°C. Electricity price information receiver: Obtains real-time electricity prices via wired or wireless communication. Unit: yuan or kWh, with a value range of 0.0 to 5.0 yuan or kWh; Meteorological data receiver: Receives weather forecast data for the next hour, including indicators such as sunshine trends, temperature, humidity, and cloud cover; All acquisition components are connected to the local processor via an acquisition bus (supporting CAN or RS485 protocols), and a time synchronization mechanism (such as NTP network time synchronization) is used to ensure that multi-source data are acquired and transmitted under the same time reference. (2) After receiving the above multi-source running data, the local processor executes the following in sequence: Timestamp appended: Provides the precise collection time for each data record, with an accuracy better than 1 second, ensuring time-series consistency in subsequent analysis; Outlier removal: Use the 3σ principle or a preset threshold (such as light intensity change rate > 100W or m² or s) to remove unreasonable data points, and fill in the missing values ​​with linear interpolation of nearest neighbor values ​​after removal; Unit standardization: Voltage is standardized to V, current to A, power to kW, and energy to kWh to ensure consistent dimensions in formula inputs; SOC calculation: Call the dynamic update formula of the energy storage unit's state of charge. The formula input includes real-time current, voltage, temperature and the previous SOC result; To prevent SOC calculation distortion caused by abnormal data acquisition, this unit is equipped with an abnormal value alarm mechanism. When the temperature or current exceeds the limit continuously (e.g., temperature > 60℃ or current > 500A for more than 5 seconds), the system will suspend SOC updates and enter a safety protection mode, waiting for manual confirmation or environmental recovery.

[0012] Preferably, the trend prediction and pattern recognition unit includes: (1) The trend prediction and pattern recognition unit includes a non-volatile memory (such as NAND Flash) connected to the local processor, which stores the algorithm code of the light intensity trend prediction model. The specific mathematical expression, symbol definition, unit and value range of the model have been given in the above method, and the training and calibration method of the model coefficients is explained (fitting and optimization based on historical light intensity, electricity price and load data). The processor calls the model, and the input parameters include: Real-time light intensity Units: W or m², range: 0~1200W or m²; Real-time electricity price Units are yuan or kWh, ranging from 0.0 to 5.0 yuan or kWh; Current user load power Unit: kW; Range: 0 to the upper limit of rated power. The model output consists of short-term predictions, including: Light intensity change trend (predicted average light intensity change rate for the next 1-15 minutes, in W or m² or min). Electricity price fluctuation trend (predicted electricity price change in the next hour, in yuan or kWh). Load demand trend (forecast of average power change over the next 30 minutes, in kW). The prediction results are timestamped with an accuracy of ≤1 second and stored in a cache for subsequent pattern determination. (2) The prediction result is transmitted to the pattern recognition logic circuit via the data bus, and the logic circuit performs a comprehensive analysis on it along with the following information: Current SOC value (in %, calculated by the "Data Acquisition and Processing Unit"); Real-time electricity price (in yuan or kWh); Real-time load power (in kW); During the analysis, the system invokes preset pattern determination conditions, such as: If SOC>80%, electricity price is low and load is small → the operation mode is determined to prioritize self-use; If the electricity price is high and the SOC is greater than 50%, it is considered peak shaving and valley filling. If electricity prices fluctuate significantly and the predicted price is low, it is considered electricity price arbitrage. If the probability of a power grid outage is high, it is determined to be an off-grid standby system. To avoid excessively frequent mode switching, the mode recognition unit has a built-in anti-shake mechanism: The mode will only switch if the judgment condition is met continuously for no less than 3 prediction periods (e.g., 15 minutes). When multiple conditions are met simultaneously, the mode is selected according to priority (safety > economy > efficiency); The mode determination result will serve as a direct input to the strategy optimization and execution control unit, ensuring that the subsequent charging and discharging control strategy is highly matched with the predicted trend, and that the system can maintain stable operating logic under abnormal weather, sudden load changes, and other conditions.

[0013] Preferably, the strategy optimization and execution control unit includes: (1) Optimize the computing processor (which can be a multi-core CPU or an embedded SoC) to run the photovoltaic energy storage charging and discharging multi-objective optimization model; The communication interface (supporting RS485, CAN or Ethernet) is used for bidirectional data interaction with the energy storage converter and load control terminal. The execution drive circuit is used to convert control signals into drive instructions that can be executed by the device and to collect execution feedback; The photovoltaic energy storage charging and discharging multi-objective optimization model has been given specific mathematical expressions, symbol definitions, dimensions, and coefficient calibration methods in the above method. The inputs include: Prediction results from the trend prediction and pattern recognition unit (light intensity, electricity price, load trends); Current SOC value (%), energy storage unit temperature (°C); Runtime safety constraint parameters: SOC upper and lower limits (e.g., 90%, 20%, calibrated based on battery life curves); Battery temperature control threshold (e.g., 50℃, determined based on thermal stability tests); Demand control threshold (e.g., 50kW, set based on contract or grid requirements); The optimization objectives include economic efficiency (such as maximizing electricity price arbitrage profits), safety (avoiding overcharging and over-discharging), and operational stability (reducing frequent switching). The optimal charging and discharging power is obtained by solving a multi-objective weighted solution. (Unit: kW) and corresponding timing control strategies (1-15 minutes per cycle); (2) The generated charging and discharging strategy is sent to the energy storage converter (controlling charging and discharging power) and the load control terminal (adjusting load power distribution) through the communication interface. The execution drive circuit converts the processor's digital control signal into an analog or PWM signal to match the input requirements of the device, and collects the execution feedback signal in real time, including: Actual charge / discharge power (unit: kW); Actual SOC change curve (unit: %); Equipment operating status indicators (such as over-temperature, over-current, and fault codes); The optimized computational processor compares the feedback signal with the predicted value: If the charge / discharge power deviation exceeds ±5%, or the SOC deviation exceeds ±3%, the strategy execution is deemed inconsistent with the prediction. The revised strategy is immediately issued through the communication interface, completing the closed-loop control process of measurement-execution-feedback-optimization. This closed-loop mechanism can effectively reduce the impact of environmental changes, equipment performance degradation, prediction errors and other factors on operational stability, and ensure the efficient and reliable operation of the energy storage system under modes such as self-use priority, peak shaving and valley filling, electricity price arbitrage and off-grid backup.

[0014] This invention also provides a computer-readable storage medium for intelligent control of photovoltaic energy storage charging. When the computer program is executed by a processor, the processor performs the method described above. The computer program includes: Data acquisition and SOC calculation instructions are used to execute the dynamic update formula for the state of charge of the energy storage unit to calculate the SOC. Trend prediction and pattern recognition instructions are used to execute the light intensity trend prediction model and determine the operating mode. Strategy optimization and execution instructions are used to execute the photovoltaic energy storage charging and discharging multi-objective optimization model to calculate the control strategy and send it to the execution end.

[0015] Preferably, the computer-readable storage medium may be a non-volatile memory, including but not limited to NAND Flash, solid-state drive (SSD), EEPROM or other media with long-term data retention capabilities. The medium stores a set of computer programs that can be executed by a processor. The program includes instruction codes for a data acquisition and processing module, a trend prediction and pattern recognition module, and a strategy optimization and execution control module. Program deployment and runtime environment This computer program can be deployed on one of the following hardware platforms: Edge controllers: such as industrial control boards based on the ARM Cortex-A72 architecture, which have local real-time computing and peripheral interface capabilities; Embedded devices: such as low-power processing units like STM32 and TIDSP, used for distributed energy storage node control; Cloud servers: such as x86 architecture virtual machines or container environments, used for centralized policy optimization and multi-site collaborative control; Functionality and update mechanism: The program supports remote policy parameter updates via secure communication protocols (such as HTTPS with TLS 1.3 or encrypted MQTT). The updated content includes operating mode determination threshold, SOC upper and lower limits, battery temperature control threshold, and weight parameters of multi-objective optimization model. When running locally, the program has adaptive optimization capabilities: it can automatically adjust the control strategy based on real-time collected SOC, light intensity, electricity price and load changes, achieving dynamic optimization without manual intervention; Communication and control interaction The program interacts bidirectionally with the energy storage converter and load control terminal via a communication interface. The communication interface can be RS485, CAN, Ethernet, or a wireless communication module (such as 4G or 5G). The interaction content includes: Obtain the actual charging and discharging power and equipment operating status code from the energy storage converter; Obtain current load power and operating mode confirmation information from the load control terminal; Issue control commands, including charging and discharging power settings (unit: kW), charging and discharging mode switching commands, and load distribution strategies; Consistency and Verifiability The program's execution logic is consistent with the aforementioned method and the three functional units of the system, ensuring the consistency and verifiability of execution results under different deployment scenarios. All communication and parameter update operations record timestamps (accuracy ≤ 1s) and execution status flags for subsequent auditing and fault diagnosis.

[0016] The beneficial effects of this invention are as follows: 1. This invention establishes a multi-source data acquisition system by combining a light sensor, voltage or current acquisition circuit, SOC detection component, electricity price information receiver, and meteorological data receiver. The data is then centrally input into a local processor via an acquisition bus, enabling synchronous acquisition and standardized processing of operational data. Combined with the dynamic update formula for the state of charge of the energy storage unit, it can calculate high-precision SOC values ​​in real time after eliminating outliers and standardizing data units. This significantly reduces the cumulative impact of measurement errors from a single sensor, ensuring the stability and reliability of energy storage status monitoring under dynamic operating conditions.

[0017] 2. This invention utilizes a light intensity trend prediction model to calculate trends in real-time light intensity, electricity price, and load data, and combines this with the current SOC value for comprehensive analysis. This allows for the early identification of future trends in light intensity, electricity price fluctuations, and load changes. After comparing the prediction results with preset mode determination conditions, it can automatically identify modes such as self-use priority, peak shaving and valley filling, electricity price arbitrage, or off-grid standby. This achieves real-time and accurate mode selection, avoids decision-making lag under fixed rules, and thus improves the economy and adaptability of energy storage dispatch under changing external conditions.

[0018] 3. This invention incorporates operational safety constraints and economic objectives, such as SOC upper and lower limits, battery temperature control threshold, and demand control threshold, into the optimization calculation through a multi-objective optimization model for photovoltaic energy storage charging and discharging. This generates a charging and discharging strategy that satisfies both safety and profitability objectives. After the strategy is issued, the prediction results are compared with real-time feedback. When the deviation exceeds the set threshold, the weight of the optimization model is dynamically adjusted to achieve a closed-loop linkage of prediction-execution-correction. This mechanism ensures the efficient and stable operation of the energy storage system under different operating modes, while extending equipment life and reducing operating costs. Attached Figure Description

[0019] Figure 1 This is a flowchart of the photovoltaic energy storage charging intelligent control method of the present invention; Figure 2 This is a flowchart of the photovoltaic energy storage charging intelligent control system of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] like Figures 1 to 2 As shown in the figure, this embodiment of the invention provides a photovoltaic energy storage charging intelligent control method, the specific steps of which are as follows: S1. Dynamic Data Acquisition and Standardized Processing: Light intensity is collected by a light sensor, output power of the photovoltaic array is collected by a voltage or current acquisition circuit, state of charge and temperature of the energy storage unit are collected by the SOC detection component, real-time electricity price is obtained by an electricity price information receiver, and weather forecast data is obtained by a meteorological data receiver. The data is transmitted to the local processor via the acquisition bus. The processor performs timestamp appending, outlier removal and unit standardization processing, and calls the dynamic update formula of the energy storage unit's state of charge to calculate the SOC in real time. S2. Trend Prediction and Operating Mode Recognition: The processor calls the prediction algorithm code in the memory, uses the light intensity trend prediction model to predict the short-term changing trends of light intensity, electricity price level and load demand, and transmits the prediction results to the pattern recognition logic circuit. Based on the SOC, electricity price, load prediction data and preset judgment conditions, the operating mode is identified as one of self-use priority, peak shaving and valley filling, electricity price arbitrage or off-grid standby. S3. Adaptive Optimization Strategy Generation and Closed-Loop Control: The processor loads the operating safety constraint parameters from the memory, including but not limited to the upper and lower limits of SOC, battery temperature control threshold, and demand control threshold, and constructs a multi-objective optimization model for photovoltaic energy storage charging and discharging. Based on the prediction results and real-time feedback, it calculates the charging and discharging power and timing control strategy, and sends it to the energy storage converter and load control terminal through the communication interface. It also collects execution feedback signals. If the prediction deviation exceeds the threshold, it dynamically adjusts the weight of the optimization model to achieve closed-loop linkage between prediction and optimization.

[0022] This embodiment provides a photovoltaic energy storage charging intelligent control method, system and storage medium, which is applicable to distributed photovoltaic power generation and energy storage integrated application scenarios. It can realize real-time acquisition of multi-source operation data, high-precision SOC dynamic calculation, short-term trend prediction, intelligent identification of operation mode and multi-objective optimization closed-loop control, thereby improving the utilization rate of photovoltaic power generation and the economy and safety of energy storage system. In this embodiment, the system includes a data acquisition and processing unit, a trend prediction and pattern recognition unit, and a strategy optimization and execution control unit. Each unit is connected through an internal data bus or a high-speed communication interface and forms an overall control architecture with the processor and memory. First, the data acquisition and processing unit includes a light sensor acquisition component, a voltage acquisition circuit, a current acquisition circuit, a SOC detection component, an electricity price information receiver, and a meteorological data receiver. Each component is used to acquire light intensity, photovoltaic array output power, energy storage unit state of charge and temperature, real-time electricity price, and weather forecast information, respectively. The above multi-source operating data is centrally transmitted to the local processor via the acquisition bus. The processor sequentially performs timestamp appending, outlier removal, and unit standardization to ensure data validity and computational compatibility. Then, it calls the energy storage unit state of charge dynamic update formula and calculates the current SOC value based on the processed real-time data, providing accurate energy storage state parameters for subsequent prediction and control strategy generation. Secondly, the trend prediction and pattern recognition unit includes a memory connected to the processor, which stores the algorithm code for performing prediction calculations. The processor calls the light intensity trend prediction model, and based on the real-time collected light intensity, electricity price and load data, generates short-term trend prediction results of light intensity changes, electricity price fluctuations and load demand. It then analyzes the results together with the current SOC value, real-time electricity price and load prediction data, and compares them with preset mode determination conditions to identify the current operating mode as one of self-use priority, peak shaving and valley filling, electricity price arbitrage or off-grid standby. The mode determination result is directly used as the input condition for subsequent strategy optimization and charge and discharge control. Finally, the strategy optimization and execution control unit includes an optimization computing processor, a communication interface, and an execution drive circuit. The optimization computing processor loads operational safety constraint parameters from memory, including SOC upper and lower limits, battery temperature control thresholds, and demand control thresholds. It then calls the photovoltaic energy storage charging and discharging multi-objective optimization model and, combining the prediction results with real-time feedback data, calculates a charging and discharging power and timing control strategy that meets both safety and economic requirements. The communication interface distributes the strategy to the energy storage converter and load control terminal for execution. The execution drive circuit collects operational feedback signals from the equipment and sends them back to the optimization computing processor. When the deviation between the actual execution result and the prediction exceeds a set threshold, the model weights are dynamically adjusted to achieve closed-loop linkage between prediction and optimization strategies, improving the accuracy and stability of system scheduling. The above method steps can be implemented by a computer program deployed on an edge controller, embedded device, or cloud server. The computer program is stored in a computer-readable storage medium and, when executed, performs all the functions of the photovoltaic energy storage charging intelligent control method described in this embodiment. It supports remote strategy parameter updates and local adaptive optimization, and performs bidirectional data and control command interaction with the energy storage converter and load control terminal.

[0023] The specific steps for dynamic data acquisition and standardization processing in S1 are as follows: S11 acquires light intensity data through a light sensor to ensure that the sampling frequency meets the input requirements of the prediction model; the instantaneous output power of the photovoltaic array is measured through voltage acquisition circuit and current acquisition circuit, and a synchronous sampling method is used to reduce phase error; The SOC detection component is used to collect the real-time state of charge and temperature of the energy storage unit. Temperature acquisition is used for temperature control judgment and SOC correction. Real-time electricity price data from the grid side is obtained by the electricity price information receiver, and forecast information including sunlight trend, temperature, humidity and weather type is obtained by the meteorological data receiver. All the above-mentioned collected data are uniformly encoded and transmitted to the local processor through the acquisition bus to ensure data format consistency and time alignment, so as to realize the centralized input of photovoltaic power generation, energy storage status, grid price and environmental forecast information. After receiving the centralized input data, the S12 local processor first adds a collection timestamp to each data item to ensure the temporal consistency of multi-source data; then it performs outlier removal, including extreme values ​​caused by sensor acquisition anomalies, communication errors, and sudden environmental changes. The removal rules are preset in memory and can be dynamically adjusted according to the operating status; subsequently, it unifies the units of various data, such as power to watts and temperature to degrees Celsius, to ensure the compatibility of physical quantities in the calculation process. After data preprocessing, the processor calls the dynamic update formula for the state of charge of the energy storage unit, and uses the photovoltaic output power, load power, energy storage unit temperature and historical SOC value to calculate the current SOC. The calculated SOC value is used as the core input for subsequent trend prediction analysis and operation strategy optimization, ensuring that control decisions are based on accurate and verifiable energy storage state information. The formula for dynamically updating the state of charge of an energy storage unit is: , In the formula: State of charge (%) at time t+Δt; State of charge at time t (%) : Charging efficiency and discharging efficiency (0~1); Charging power and discharging power (kW); Rated capacity of energy storage system (kWh); : Time interval (h); Formula origin and derivation: Based on the law of conservation of energy, the SOC value is updated after the energy increase during charging and the energy decrease during discharging are converted according to efficiency.

[0024] The specific steps for trend prediction and operation pattern recognition in S2 are as follows: The S21 local processor calls the prediction algorithm code pre-stored in the memory to execute the light intensity trend prediction model. The model takes real-time light intensity data after dynamic data acquisition and standardization processing, electricity price information and user load data as input, and uses a sliding time window method to call historical data of the past 10 to 30 minutes. The sampling period is 1 minute, and the prediction time span is 15 minutes to 1 hour in the future. The calculation results include the fitted value of the light intensity change curve, the trend of electricity price fluctuation and the trend of load demand change, and also output the trend direction (rising, falling or stable). To address sensor drift or sudden environmental changes, the model adjusts its weights based on the latest feedback data after each round of calculation to maintain prediction accuracy. The above prediction results provide a forward-looking reference for energy storage scheduling and can effectively reduce the risk of strategy lag caused by rapid fluctuations in sunlight, changes in electricity prices, or sudden increases in load. The principle formula of the light intensity trend prediction model is: , In the formula: Predict the light intensity (W or m²) at time point t. Lighting prediction (W or m²) based on sliding regression; Light forecast (W or m²) based on seasonal exponential smoothing. The fusion weight coefficients (0~1) are determined by cross-validation. Formula origin and derivation: Sliding regression is suitable for capturing short-term trends, seasonal index smoothing is suitable for cyclical changes, and the weighted fusion of the two can take into account both short-term changes and seasonal patterns. Operational verification: Using historical photovoltaic monitoring data (5-minute resolution), implement dual-model prediction in MATLAB or Python, and compare the mean square error (MSE) of the single prediction method to verify the improved accuracy of the fusion prediction. The S22 processor transmits the trend prediction results output by S21 to the pattern recognition logic circuit, and inputs them into the pattern determination module along with the current energy storage unit SOC value, real-time electricity price, and load prediction data. The mode determination is based on the preset determination conditions and priority rules in the memory: when multiple operating conditions are met at the same time, the priority order is safe operation (SOC protection) > peak shaving and valley filling > electricity price arbitrage > self-consumption priority > off-grid backup. During the determination process, multiple parameters such as SOC safety range (upper and lower limits), electricity price threshold, load change rate, and prediction confidence are considered. When the input data is missing or the prediction accuracy is lower than the set threshold, it automatically switches to the fallback mode to ensure safe and continuous operation. The final mode recognition result will be used as the mode input for the photovoltaic energy storage charging and discharging multi-objective optimization model in adaptive optimization strategy generation and closed-loop control, realizing the closed-loop connection between predictive analysis and strategy optimization.

[0025] The specific steps for adaptive optimization strategy generation and closed-loop control in S3 are as follows: The S31 processor loads the runtime security constraint parameters from memory. These parameters include: SOC upper and lower limits (unit: %, range: 10%~90%, obtained from battery type and life curve calibration); Battery temperature control threshold (unit: °C, range: -10℃~50℃, determined by the cell manufacturer's recommended value and actual thermal stability). Demand control threshold (unit: kW, range: 0 to rated power limit, determined by distribution capacity and contract demand constraints). After acquiring the operating mode and prediction results output by trend prediction and operating mode recognition, as well as the real-time operating data of dynamic data acquisition and standardized processing, the processor calls the photovoltaic energy storage charging and discharging multi-objective optimization model for calculation. The model is solved by weighting the economic objectives (minimizing operating costs), safety objectives (not exceeding SOC and temperature limits), and stability objectives (minimizing load fluctuations). The weighting coefficients βi are obtained by calibrating through historical operating data and operational strategies. The model calculates and outputs the charging and discharging power (unit: kW) that meets the safety constraints and the corresponding timing control strategy, ensuring that the energy storage system balances safety and economy under the given operating mode; The charging and discharging power and timing control strategy obtained by S32 optimization calculation are sent to the energy storage converter and load control terminal through the communication interface, driving the relevant equipment to perform charging and discharging operations according to the optimization results; During execution, the processor synchronously collects the output power of the energy storage converter, the actual load power, and the status data of the energy storage unit, and compares them in real time with the prediction results in S31. When the execution deviation is detected to exceed the set threshold (e.g., power deviation > 5%, SOC deviation > 3%), the processor triggers a dynamic weight adjustment mechanism. The weight coefficient βi of the photovoltaic energy storage charging and discharging multi-objective optimization model is adaptively corrected so that the weight of the safety objective is increased or the weight of the economic objective is decreased, so as to prioritize the safety of operation. The updated weight parameters take effect immediately, the control strategy is recalculated and redistributed, achieving a closed-loop linkage between prediction and optimization strategies; To prevent frequent adjustments from causing system oscillations, a minimum time interval (e.g., 5 minutes) and a maximum single adjustment range (e.g., no more than 10%) are set for weight adjustments. This mechanism can significantly improve the system's response accuracy and operational stability under conditions such as sudden load changes, rapid weather changes, and abnormal electricity price fluctuations. The principle formula of the photovoltaic energy storage charging and discharging multi-objective optimization model is as follows: , In the formula: : Comprehensive optimization of objective function value (dimensionless weighted sum); The weight coefficients (real numbers, 0~1) of each optimization objective satisfy the following conditions: ; Electricity cost per unit time (in yuan or kWh); : The difference between the current SOC and the target SOC (%); Battery health risk factor (%) or cycle, calculated from the number of cycles and the percentage of deep discharge; System power loss (kW) is estimated from the losses of the converter and transmission lines; Formula origin and derivation: This formula originates from the multi-objective linear weighted method in operations research. It quantifies economy, energy storage stability, equipment life and efficiency as independent objectives, and after dimensional normalization, it is summed by weight to form a single optimization objective.

[0026] The present invention also provides a photovoltaic energy storage charging intelligent control system based on the above method, and the system includes: A. Data acquisition and processing unit, including light sensor acquisition component, voltage or current acquisition circuit, SOC detection component, electricity price information receiver and meteorological data receiver. Each component is connected to the local processor through the acquisition bus. It is used to acquire multi-source operating data and perform timestamp appending, outlier removal and unit standardization processing. The processor is configured to call the energy storage unit's state of charge dynamic update formula to calculate SOC. B. Trend prediction and pattern recognition unit: The prediction algorithm code is stored in the memory connected to the processor and executed by the processor. The unit is configured to call the light intensity trend prediction model to generate short-term trend prediction results of light intensity, electricity price and load, and determine the operating mode based on the prediction results and SOC, electricity price and load data. C. Strategy optimization and execution control unit, including optimization computing processor, communication interface and execution drive circuit. The optimization computing processor is configured to call the photovoltaic energy storage charging and discharging multi-objective optimization model to generate charging and discharging power and timing control strategy. The communication interface sends the strategy to the energy storage converter and load control terminal, and collects execution feedback signals to adjust the optimization model weights.

[0027] The data acquisition and processing unit includes: (1) The data acquisition and processing unit includes the following components: Light sensor acquisition component: used to measure the instantaneous light intensity received on the surface of the photovoltaic array. The unit is W or m2, and the value range is 0~1200W or m2; Voltage acquisition circuit: Acquires the terminal voltages of the photovoltaic array and energy storage units. ,unit Range 0-500 (Energy storage); Current acquisition circuit: Acquires the output current of the photovoltaic array and the charging and discharging current of the energy storage. , Unit A, range 0~500A; SOC detection component: measures the state of charge (SOC) and temperature of the energy storage unit. SOC is measured in percent and ranges from 0 to 60°C. Electricity price information receiver: Obtains real-time electricity prices via wired or wireless communication. Unit: yuan or kWh, with a value range of 0.0 to 5.0 yuan or kWh; Meteorological data receiver: Receives weather forecast data for the next hour, including indicators such as sunshine trends, temperature, humidity, and cloud cover; All acquisition components are connected to the local processor via an acquisition bus (supporting CAN or RS485 protocols), and a time synchronization mechanism (such as NTP network time synchronization) is used to ensure that multi-source data are acquired and transmitted under the same time reference. (2) After receiving the above multi-source running data, the local processor executes the following in sequence: Timestamp appended: Provides the precise collection time for each data record, with an accuracy better than 1 second, ensuring time-series consistency in subsequent analysis; Outlier removal: Use the 3σ principle or a preset threshold (such as light intensity change rate > 100W or m² or s) to remove unreasonable data points, and fill in the missing values ​​with linear interpolation of nearest neighbor values ​​after removal; Unit standardization: Voltage is standardized to V, current to A, power to kW, and energy to kWh to ensure consistent dimensions in formula inputs; SOC calculation: Call the dynamic update formula of the energy storage unit's state of charge. The formula input includes real-time current, voltage, temperature and the previous SOC result; To prevent SOC calculation distortion caused by abnormal data acquisition, this unit is equipped with an abnormal value alarm mechanism. When the temperature or current exceeds the limit continuously (e.g., temperature > 60℃ or current > 500A for more than 5 seconds), the system will suspend SOC updates and enter a safety protection mode, waiting for manual confirmation or environmental recovery.

[0028] The trend prediction and pattern recognition unit includes: (1) The trend prediction and pattern recognition unit includes a non-volatile memory (such as NAND Flash) connected to the local processor, which stores the algorithm code of the light intensity trend prediction model. The specific mathematical expression, symbol definition, unit and value range of the model have been given in the above method, and the training and calibration method of the model coefficients is explained (fitting and optimization based on historical light intensity, electricity price and load data). The processor calls the model, and the input parameters include: Real-time light intensity Units: W or m², range: 0~1200W or m²; Real-time electricity price Units are yuan or kWh, ranging from 0.0 to 5.0 yuan or kWh; Current user load power Unit: kW; Range: 0 to the upper limit of rated power. The model output is short-term prediction results, including: Light intensity change trend (predicted average light intensity change rate for the next 1-15 minutes, in W or m² or min). Electricity price fluctuation trend (predicted electricity price change in the next hour, in yuan or kWh). Load demand trend (forecast of average power change over the next 30 minutes, in kW). The prediction results are timestamped with an accuracy of ≤1 second and stored in a cache for subsequent pattern determination. (2) The prediction result is transmitted to the pattern recognition logic circuit via the data bus, and the logic circuit performs a comprehensive analysis on it along with the following information: Current SOC value (in %, calculated by the "Data Acquisition and Processing Unit"); Real-time electricity price (in yuan or kWh); Real-time load power (in kW); During the analysis, the system invokes preset pattern determination conditions, such as: If SOC>80%, electricity price is low and load is small → the operation mode is determined to prioritize self-use; If the electricity price is high and the SOC is greater than 50%, it is considered peak shaving and valley filling. If electricity prices fluctuate significantly and the predicted price is low, it is considered electricity price arbitrage. If the probability of a power grid outage is high, it is determined to be an off-grid standby system. To avoid excessively frequent mode switching, the mode recognition unit has a built-in anti-shake mechanism: The mode will only switch if the judgment condition is met continuously for no less than 3 prediction periods (e.g., 15 minutes). When multiple conditions are met simultaneously, the mode is selected according to priority (safety > economy > efficiency); The mode determination result will serve as a direct input to the strategy optimization and execution control unit, ensuring that the subsequent charging and discharging control strategy is highly matched with the predicted trend, and that the system can maintain stable operating logic under abnormal weather, sudden load changes, and other conditions.

[0029] The strategy optimization and execution control unit includes: (1) Optimize the computing processor (which can be a multi-core CPU or an embedded SoC) to run the photovoltaic energy storage charging and discharging multi-objective optimization model; The communication interface (supporting RS485, CAN or Ethernet) is used for bidirectional data interaction with the energy storage converter and load control terminal. The execution drive circuit is used to convert control signals into drive instructions that can be executed by the device and to collect execution feedback; The multi-objective optimization model for photovoltaic energy storage charging and discharging has been given specific mathematical expressions, symbol definitions, dimensions, and coefficient calibration methods in the above methods. The inputs include: Prediction results from the trend prediction and pattern recognition unit (light intensity, electricity price, load trends); Current SOC value (%), energy storage unit temperature (°C); Runtime safety constraint parameters: SOC upper and lower limits (e.g., 90%, 20%, calibrated based on battery life curves); Battery temperature control threshold (e.g., 50℃, determined based on thermal stability tests); Demand control threshold (e.g., 50kW, set based on contract or grid requirements); The optimization objectives include economic efficiency (such as maximizing electricity price arbitrage profits), safety (avoiding overcharging and over-discharging), and operational stability (reducing frequent switching). The optimal charging and discharging power is obtained by solving a multi-objective weighted solution. (Unit: kW) and corresponding timing control strategies (1-15 minutes per cycle); (2) The generated charging and discharging strategy is sent to the energy storage converter (controlling charging and discharging power) and the load control terminal (adjusting load power distribution) through the communication interface. The execution drive circuit converts the processor's digital control signal into an analog or PWM signal to match the input requirements of the device, and collects the execution feedback signal in real time, including: Actual charge / discharge power (unit: kW); Actual SOC change curve (unit: %); Equipment operating status indicators (such as over-temperature, over-current, and fault codes); The optimized computational processor compares the feedback signal with the predicted value: If the charge / discharge power deviation exceeds ±5%, or the SOC deviation exceeds ±3%, the strategy execution is deemed inconsistent with the prediction. The revised strategy is immediately issued through the communication interface, completing the closed-loop control process of measurement-execution-feedback-optimization. This closed-loop mechanism can effectively reduce the impact of environmental changes, equipment performance degradation, prediction errors and other factors on operational stability, and ensure the efficient and reliable operation of the energy storage system under modes such as self-use priority, peak shaving and valley filling, electricity price arbitrage and off-grid backup.

[0030] This invention also provides a computer-readable storage medium for intelligent control of photovoltaic energy storage charging. When the computer program is executed by a processor, the processor performs the method described above. The computer program includes: Data acquisition and SOC calculation instructions are used to execute the dynamic update formula for the state of charge of the energy storage unit to calculate the SOC. Trend prediction and pattern recognition instructions are used to execute the light intensity trend prediction model and determine the operating mode. Strategy optimization and execution instructions are used to execute the photovoltaic energy storage charging and discharging multi-objective optimization model to calculate the control strategy and send it to the execution end.

[0031] The computer-readable storage medium may be a non-volatile memory, including but not limited to NAND Flash, solid-state drive (SSD), EEPROM or other media with long-term data retention capabilities. The medium stores a set of computer programs that can be executed by a processor. The program includes instruction codes for a data acquisition and processing module, a trend prediction and pattern recognition module, and a strategy optimization and execution control module. Program deployment and runtime environment This computer program can be deployed on one of the following hardware platforms: Edge controllers: such as industrial control boards based on the ARM Cortex-A72 architecture, which have local real-time computing and peripheral interface capabilities; Embedded devices: such as low-power processing units like STM32 and TIDSP, used for distributed energy storage node control; Cloud servers: such as x86 architecture virtual machines or container environments, used for centralized policy optimization and multi-site collaborative control; Functionality and update mechanism: The program supports remote policy parameter updates via secure communication protocols (such as HTTPS with TLS 1.3 or encrypted MQTT). The updated content includes operating mode determination threshold, SOC upper and lower limits, battery temperature control threshold, and weight parameters of multi-objective optimization model. When running locally, the program has adaptive optimization capabilities: it can automatically adjust the control strategy based on real-time collected SOC, light intensity, electricity price and load changes, achieving dynamic optimization without manual intervention; Communication and control interaction The program interacts bidirectionally with the energy storage converter and load control terminal via a communication interface. The communication interface can be RS485, CAN, Ethernet, or a wireless communication module (such as 4G or 5G). The interaction content includes: Obtain the actual charging and discharging power and equipment operating status code from the energy storage converter; Obtain current load power and operating mode confirmation information from the load control terminal; Issue control commands, including charging and discharging power settings (unit: kW), charging and discharging mode switching commands, and load distribution strategies; Consistency and Verifiability The program's execution logic is consistent with the aforementioned method and the three functional units of the system, ensuring the consistency and verifiability of execution results under different deployment scenarios. All communication and parameter update operations record timestamps (accuracy ≤ 1s) and execution status flags for subsequent auditing and fault diagnosis.

[0032] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0033] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent control of photovoltaic energy storage charging, characterized in that: The specific steps of this intelligent control method for photovoltaic energy storage charging are as follows: S1. Dynamic data acquisition and standardization processing: Acquire data on light intensity, photovoltaic output power, SOC, temperature, electricity price and weather forecast, transmit the data to the processor via the acquisition bus, perform timestamp appending, outlier removal and unit standardization, and call the dynamic update formula of energy storage unit state of charge to calculate SOC in real time, providing accurate state parameters for subsequent forecasts. S2. Trend Prediction and Operation Mode Recognition: The processor calls the light intensity trend prediction model to calculate the short-term trends of light intensity, electricity price and load. It compares the prediction results with SOC, electricity price and load data, and identifies the operation mode as self-use priority, peak shaving and valley filling, electricity price arbitrage or off-grid standby according to the judgment conditions, providing mode input for strategy optimization. S3. Adaptive Optimization Strategy Generation and Closed-Loop Control: Load the operating safety constraints of SOC upper and lower limits, battery temperature control threshold, and demand control threshold, call the photovoltaic energy storage charging and discharging multi-objective optimization model to generate charging and discharging power and timing control strategies, send them to the execution end and collect feedback, and dynamically adjust the model weights when the prediction deviation exceeds the threshold to achieve closed-loop optimization control.

2. The intelligent control method for photovoltaic energy storage charging according to claim 1, characterized in that: The specific steps for dynamic data acquisition and standardization processing in S1 are as follows: S11 acquires light intensity data through a light sensor, measures the output power of the photovoltaic array through a voltage acquisition circuit and a current acquisition circuit; it uses a SOC detection component to acquire the state of charge and temperature information of the energy storage unit; it obtains real-time electricity price data through an electricity price information receiver, and obtains weather forecast information through a meteorological data receiver. After receiving the collected data, the S12 local processor timestamps each data item to record the time of data acquisition; it performs outlier removal to ensure data validity and performs unit standardization for subsequent calculations. Then, it calls the dynamic update formula for the energy storage unit's state of charge and calculates the current state of charge of the energy storage unit based on the processed real-time collected data.

3. The intelligent control method for photovoltaic energy storage charging according to claim 2, characterized in that: The specific steps for trend prediction and operation pattern recognition in S2 are as follows: The S21 processor calls the prediction algorithm code pre-stored in memory, executes the light intensity trend prediction model, and calculates the short-term trend results of light intensity changes, electricity price fluctuations and load demand based on real-time collected light intensity, electricity price information and user load data. The S22 processor transmits the above prediction results to the pattern recognition logic circuit, and combines the current SOC value, real-time electricity price and load prediction data with preset judgment conditions to determine the current operating mode as one of self-use priority, peak shaving and valley filling, electricity price arbitrage or off-grid reserve.

4. The intelligent control method for photovoltaic energy storage charging according to claim 3, characterized in that: The specific steps for generating the adaptive optimization strategy and implementing closed-loop control in S3 are as follows: The S31 processor loads and runs safety constraint parameters from memory, including but not limited to SOC upper and lower limits, battery temperature control threshold, and demand control threshold. Based on the prediction results and real-time operating data, it calls the photovoltaic energy storage charging and discharging multi-objective optimization model. The S32 control strategy is sent to the energy storage converter and load control terminal via the communication interface, driving the relevant equipment to perform charging and discharging operations according to the optimization results. The processor synchronously collects the execution feedback signal and compares it with the predicted data. When the deviation exceeds the set threshold, the weight parameters of the multi-objective optimization model are dynamically adjusted.

5. A photovoltaic energy storage charging intelligent control system, characterized in that: The photovoltaic energy storage charging intelligent control system is based on the method of claim 4, and the system includes: A. Data acquisition and processing unit, including light sensor acquisition component, voltage or current acquisition circuit, SOC detection component, electricity price information receiver and meteorological data receiver. Each component is connected to the local processor through acquisition bus. It is used to acquire multi-source operating data and perform timestamp appending, outlier removal and unit standardization processing. The processor is configured to call the energy storage unit's state of charge dynamic update formula to calculate SOC. B. Trend prediction and pattern recognition unit, which stores prediction algorithm code in a memory connected to the processor and is executed by the processor. The unit is configured to call the light intensity trend prediction model to generate short-term trend prediction results of light intensity, electricity price and load, and determine the operating mode based on the prediction results and SOC, electricity price and load data. C. Strategy optimization and execution control unit, including an optimization computing processor, a communication interface and an execution drive circuit. The optimization computing processor is configured to call the photovoltaic energy storage charging and discharging multi-objective optimization model to generate charging and discharging power and timing control strategies. The communication interface sends the strategies to the energy storage converter and load control terminal, and collects execution feedback signals to adjust the optimization model weights.

6. The photovoltaic energy storage charging intelligent control system according to claim 5, characterized in that: The data acquisition and processing unit includes: (1) The data acquisition and processing unit includes a light sensor acquisition component, a voltage acquisition circuit, a current acquisition circuit, a SOC detection component, an electricity price information receiver and a meteorological data receiver. Each acquisition component is connected to the local processor through an acquisition bus to acquire multi-source operation data of the photovoltaic array, energy storage unit, external power grid and environment, and to centrally transmit the data to the local processor for unified processing. (2) After receiving multi-source operating data, the local processor sequentially performs timestamp appending, outlier removal and unit standardization processing, and then calls the dynamic update formula of the energy storage unit's state of charge, and calculates the current SOC value in combination with the processed real-time data.

7. The photovoltaic energy storage charging intelligent control system according to claim 6, characterized in that: The trend prediction and pattern recognition unit includes: (1) The trend prediction and pattern recognition unit includes a memory connected to the processor, which stores algorithm code for performing prediction operations. The processor calls the program in the memory, runs the light intensity trend prediction model, and generates short-term trend prediction results of light intensity changes, electricity price fluctuations and load demand based on real-time collected light, electricity price and load data. (2) The trend prediction and pattern recognition unit will comprehensively analyze the generated prediction results with the current SOC value, real-time electricity price and load data, and compare them with the preset mode judgment conditions to determine the current operating mode as one of self-use priority, peak shaving and valley filling, electricity price arbitrage or off-grid reserve.

8. The photovoltaic energy storage charging intelligent control system according to claim 7, characterized in that: The strategy optimization and execution control unit includes: (1) The strategy optimization and execution control unit includes an optimization computing processor, a communication interface and an execution drive circuit. The optimization computing processor calls the photovoltaic energy storage charging and discharging multi-objective optimization model, and calculates the charging and discharging power and corresponding timing control strategy that meet the requirements of economy and safety by combining the prediction results and the operation safety constraint parameters. (2) The communication interface transmits the control strategy generated by the optimization computing processor to the energy storage converter and load control terminal, drives the relevant equipment to perform corresponding charging and discharging operations, and the execution drive circuit collects the real-time feedback signal of the equipment operation and provides it to the optimization computing processor. When it detects that the actual execution result deviates from the prediction beyond the threshold, it adjusts the weight of the multi-objective optimization model.

9. A computer-readable storage medium for intelligent control of photovoltaic energy storage charging, characterized in that: When the computer program is executed by a processor, the processor performs the method as described in claim 4, wherein the computer program comprises: Data acquisition and SOC calculation instructions are used to execute the dynamic update formula for the state of charge of the energy storage unit to calculate the SOC. Trend prediction and pattern recognition instructions are used to execute the light intensity trend prediction model and determine the operating mode. Strategy optimization and execution instructions are used to execute the photovoltaic energy storage charging and discharging multi-objective optimization model to calculate the control strategy and send it to the execution end.

10. A computer-readable storage medium for intelligent control of photovoltaic energy storage charging according to claim 9, characterized in that: The computer program is deployed on an edge controller, embedded device, or cloud server, supports remote policy parameter updates and local adaptive optimization, and interacts with energy storage converters and load control terminals through communication interfaces to exchange data and control commands.

Images