Photovoltaic consumption optimization method and system

Abstract

The invention discloses a photovoltaic consumption optimization method and system, and mainly relates to the technical field of photovoltaic consumption optimization. Comprising the following steps: acquiring historical power generation data and real-time power generation data of a photovoltaic power generation system, and historical power consumption data and real-time power consumption data of a load; based on the historical power generation data and the historical power utilization data, predicting photovoltaic power generation power and load demand power in a future time period; according to the predicted photovoltaic power generation power and the load demand power, a photovoltaic consumption strategy is formulated, and the light abandoning rate is minimized; and executing the photovoltaic consumption strategy, and controlling the operation of the photovoltaic power generation system and the load. The method has the beneficial effects that efficient consumption of photovoltaic power generation is realized, the light abandoning rate is remarkably reduced through accurate prediction and an optimization strategy, the power grid interaction cost is reduced while the energy utilization rate is improved, and the operation economy and reliability of the system are enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of photovoltaic power consumption optimization, and particularly relates to a photovoltaic power consumption optimization method and system. BACKGROUND

[0002] With the rapid development and wide application of photovoltaic power generation technology, how to efficiently consume the random and volatile power generated by photovoltaic power generation has become a key problem restricting the further development of photovoltaic power generation. In the prior art, photovoltaic power consumption mainly relies on traditional grid dispatching and a simple 'generate and use' mode. Specifically, the currently widely used technologies include light abandonment control based on a fixed threshold, a dispatching scheme according to simple load prediction, and a local charging and discharging strategy of an independently running energy storage system. These methods can alleviate the fluctuation of photovoltaic power to some extent, but have obvious deficiencies. First, the light abandonment rate is high. When the light resource is good but the load demand is low, a large amount of photovoltaic power is forced to be abandoned, causing energy waste. Second, the system operation economy is poor. Photovoltaic power is not fully utilized to reduce the power purchase cost of the grid, and the fine consideration of the grid interaction cost is lacking. Third, the sub-systems (such as photovoltaic, energy storage, and load) lack coordinated optimization, and the operation strategy is rigid, making it difficult to cope with complex real-time power fluctuations.

[0003] The reasons for the above problems are as follows: First, the prediction accuracy is insufficient. The existing methods are too rough in predicting photovoltaic power generation and load demand, and cannot provide reliable data basis for optimization dispatching. Second, the optimization target is single. Most of them only focus on power balance, and do not comprehensively optimize the light abandonment rate, operation cost and other multiple targets. Finally, the control strategy lacks self-adaptive ability. The existing control model is mostly static or based on experience rules, and cannot adjust the optimal strategy in real time according to the dynamic changes of the system state.

[0004] Therefore, there is an urgent need for a photovoltaic power consumption optimization method and system to solve the above problems. SUMMARY

[0005] The purpose of the present application is to provide a photovoltaic power consumption optimization method and system, which realizes efficient consumption of photovoltaic power generation, significantly reduces the light abandonment rate through accurate prediction and optimization strategy, improves the energy utilization rate while reducing the grid interaction cost, and enhances the system operation economy and reliability.

[0006] To achieve the above purpose, the present application realizes the following technical solutions: On the one hand, the present application provides a photovoltaic power consumption optimization method, comprising the following steps: Step S1: Obtain historical and real-time power generation data of the photovoltaic power generation system, as well as historical and real-time power consumption data of the load; Step S2: Based on historical power generation data and historical power consumption data, predict the photovoltaic power generation capacity and load demand power for future periods; Step S3: Based on the predicted photovoltaic power generation and load demand, formulate a photovoltaic consumption strategy to minimize the curtailment rate; Step S4: Implement the photovoltaic consumption strategy and control the operation of the photovoltaic power generation system and load.

[0007] Preferably, in step S2, predicting the photovoltaic power generation and load power for future time periods includes: A time series analysis model was used to predict photovoltaic power generation, and a regression model was used to predict load demand power. The time series analysis model is trained based on the periodic characteristics of historical power generation data, and the regression model is fitted based on the external variables of historical electricity consumption data. The prediction process includes data preprocessing, model training, and real-time prediction sub-steps. Data preprocessing includes imputing missing values ​​and removing outliers. Model training uses cross-validation to optimize parameters. Real-time prediction updates data through a sliding window.

[0008] Preferably, the time series analysis model is an autoregressive integral moving average model, expressed as: ; in, express Photovoltaic power generation at any given time; to It is an autoregressive coefficient, used to represent the degree of influence of historical power values ​​on the current value; to It is the moving average coefficient, used to represent the degree of influence of historical noise on the current value; yes The white noise sequence at each time step represents random errors that the model cannot explain; and It is the model order, representing the lag order of the autoregressive and moving average components, respectively, and is determined by minimizing the residual variance; The model parameters are solved by the maximum likelihood estimation method and trained using historical power generation data.

[0009] Preferably, in step S3, formulating a photovoltaic grid integration strategy includes: Based on the predicted photovoltaic power generation and load demand, an optimization objective function is constructed to minimize the curtailment rate and grid interaction costs. The optimization objective function is: ; wherein, is the photovoltaic power generation power at time t; is the actual photovoltaic power generation power consumed at time t; is the cost of interaction with the power grid at time t, including power purchase and power sale fees; and are weight coefficients for balancing the importance of light rejection rate and cost, with a value range of 0 to 1, and ; is the total duration of the optimization period; The optimization process is solved using linear programming or genetic algorithm, and real-time fluctuation constraints of photovoltaic power generation and load are considered.

[0010] Preferably, the solution process of the optimization objective function includes: Multi-objective optimization is performed using a genetic algorithm, and the fitness function is defined as: ; wherein, is the fitness value for evaluating the quality of the solution, and the larger the value, the better the solution; The operations of the genetic algorithm include selection, crossover and mutation, the selection operation is based on the roulette method, the crossover operation uses single-point crossover, and the mutation operation randomly changes part of the genes of the solution with a probability ; is the mutation probability, with a value range of 0.01 to 0.1; The optimization process is iterated until the fitness value converges or the maximum number of iterations is reached, thereby generating a photovoltaic consumption strategy, including an energy storage system charging and discharging plan and a load scheduling scheme.

[0011] In another aspect, the present application also provides a photovoltaic consumption optimization system for implementing a photovoltaic consumption optimization method as described above, comprising: a data acquisition module for acquiring historical power generation data and real-time power generation data of a photovoltaic power generation system, and historical power consumption data and real-time power consumption data of a load; a prediction module for predicting photovoltaic power generation power and load demand power in a future time period based on the historical power generation data and historical power consumption data; a strategy formulation module for formulating a photovoltaic consumption strategy according to the predicted photovoltaic power generation power and load demand power to minimize light rejection rate; an execution module for executing the photovoltaic consumption strategy to control the operation of the photovoltaic power generation system and the load. ​​​

[0012] Preferably, the data acquisition module comprises: a sensor unit for collecting real-time voltage, current and irradiance data of the photovoltaic power generation system, and power data of the load; a communication unit for transmitting data to the central processing unit through wireless network or wired protocol; a storage unit for storing historical power generation data and historical power consumption data, and using a database management system for data indexing and querying; The data acquisition module also integrates data cleaning function, which filters and normalizes the collected data to ensure data quality.

[0013] Preferably, the prediction module comprises: a model training unit for training the prediction model using machine learning algorithms, including support vector machines or neural networks; a real-time prediction unit for outputting photovoltaic power generation power and load demand power in future time periods according to real-time data input; an evaluation unit for evaluating prediction accuracy by root mean square error index and dynamically adjusting model parameters; The prediction module also supports multi-source data fusion, combining meteorological data and calendar information to improve prediction accuracy.

[0014] Preferably, the strategy formulation module comprises: an optimization algorithm unit for solving the optimization objective function according to claim 5 or 6; a strategy generation unit for generating specific photovoltaic consumption strategies according to optimization results, including control instructions for energy storage systems and adjustable loads; a simulation unit for simulating and testing the strategy before execution to evaluate the effectiveness and robustness of the strategy; The strategy formulation module also integrates human-computer interaction interface, allowing users to set weight coefficients and constraint conditions.

[0015] Preferably, the execution module comprises: a control unit for sending control signals to photovoltaic power generation systems, energy storage systems and loads to achieve power regulation; a monitoring unit for real-time monitoring of strategy execution effect and feeding back data to the data acquisition module; an alarm unit for issuing an alarm when an anomaly is detected and triggering an emergency handling program; The execution module also supports remote operation, realizing collaborative control of distributed photovoltaic consumption through cloud platform.

[0016] Compared with the prior art, the present application has the following advantages: 1. The application adopts advanced time series analysis models (such as ARIMA) and machine learning algorithms (such as SVR) to make high-precision and forward-looking predictions of photovoltaic power generation and load demand. This provides a reliable data foundation for subsequent optimization, enabling the system to predict the contradictory periods of "power generation peak" and "power consumption trough" in advance; 2. The optimization objective function of the application innovatively includes "grid interaction cost" and light rejection rate in the optimization category. This cost model accurately considers market factors such as time-of-use electricity prices and electricity purchase and sale price differences; 3. The application also builds an integrated system, which forms a closed-loop feedback between its modules (data acquisition, prediction, strategy formulation, and execution). The monitoring unit of the execution module feeds real-time data back to the system for the next round of prediction and optimization, enabling the strategy to continuously evolve and adapt. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a method flowchart of the application; Figure 2 is a system structure schematic diagram of the application. DETAILED DESCRIPTION

[0018] The application will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the application and not to limit the scope of the application. In addition, it should be understood that after reading the content taught by the application, those skilled in the art can make various modifications or modifications to the application, and these equivalent forms also fall within the scope defined by the application.

[0019] In the application, terms such as "up", "down", "left", "right", "front", "back", "vertical", "horizontal", "side", "bottom", etc. indicate the orientation or positional relationship shown in the drawings, which is only a relationship word determined for the convenience of describing the structural relationship of the components or elements of the application, and cannot be understood as a limitation of the application.

[0020] Embodiment: As shown in Figure 1 , the embodiment provides a photovoltaic consumption optimization method, comprising the following steps: Step S1: Obtain historical power generation data and real-time power generation data of a photovoltaic power generation system, and historical power consumption data and real-time power consumption data of a load; Step S2: Based on the historical power generation data and the historical power consumption data, predict the photovoltaic power generation power and the load demand power in the future time period; Step S3: According to the predicted photovoltaic power generation power and load demand power, formulate a photovoltaic consumption strategy to minimize the light rejection rate; Step S4: execute the photovoltaic accommodation strategy to control the operation of the photovoltaic power generation system and the load.

[0021] As shown in Figure 2 The embodiment also provides a photovoltaic accommodation optimization system, comprising: a data acquisition module configured to acquire historical power generation data and real-time power generation data of the photovoltaic power generation system, and historical power consumption data and real-time power consumption data of the load; a prediction module configured to predict photovoltaic power generation power and load demand power in a future time period based on the historical power generation data and the historical power consumption data; a strategy formulation module configured to formulate a photovoltaic accommodation strategy according to the predicted photovoltaic power generation power and the load demand power, so as to minimize the light rejection rate; an execution module configured to execute the photovoltaic accommodation strategy to control the operation of the photovoltaic power generation system and the load; The data acquisition module comprises: a sensor unit configured to collect voltage, current and irradiance data of the photovoltaic power generation system, and power data of the load in real time; a communication unit configured to transmit the data to a central processing unit through a wireless network or a wired protocol; a storage unit configured to store the historical power generation data and the historical power consumption data, and to index and query the data by using a database management system; The data acquisition module further integrates a data cleaning function to filter and normalize the collected data, so as to ensure the data quality; The prediction module comprises: a model training unit configured to train a prediction model by using a machine learning algorithm, wherein the algorithm comprises a support vector machine or a neural network; a real-time prediction unit configured to output the photovoltaic power generation power and the load demand power in the future time period according to real-time data input; an evaluation unit configured to evaluate the prediction accuracy by using a root mean square error index, and to dynamically adjust model parameters; The prediction module further supports multi-source data fusion, and combines meteorological data and calendar information to improve the prediction accuracy; The strategy formulation module comprises: an optimization algorithm unit configured to solve the optimization objective function according to claim 5 or 6; a strategy generation unit configured to generate a specific photovoltaic accommodation strategy according to the optimization result, wherein the strategy comprises control instructions for an energy storage system and an adjustable load; a simulation unit configured to perform simulation test before the strategy execution, and to evaluate the effectiveness and robustness of the strategy; The strategy formulation module further integrates a man-machine interaction interface, which allows a user to set weight coefficients and constraint conditions; The execution module comprises: a control unit for sending control signals to the photovoltaic power generation system, the energy storage system and the load to realize power regulation; a monitoring unit for monitoring the effect of strategy execution in real time and feeding back data to the data acquisition module; an alarm unit for issuing an alarm when an abnormality is detected and triggering an emergency handling program; The execution module also supports remote operation, and the collaborative control of distributed photovoltaic consumption is realized through a cloud platform.

[0022] This embodiment takes photovoltaic power station A as an example to illustrate the application of the above method and system in photovoltaic power station A.

[0023] First, the photovoltaic consumption optimization system described above is deployed in photovoltaic power station A: Among them, the sensor unit of the data acquisition module is installed in the photovoltaic array, the inverter, the grid connection point and each main power distribution line in the park. These sensors collect real-time power generation data (including DC side voltage, current, AC side power, irradiance, and environmental temperature) and real-time power consumption data (including total load power in the park, and power of key adjustable loads such as central air conditioner and electric vehicle charging pile group) in real time; The communication unit adopts a hybrid network of industrial Ethernet and 5G wireless network, and transmits the collected data to the central processing unit of the power station control center at a frequency of once per second; The storage unit adopts a time series database, which records historical power generation data and historical power consumption data for more than one year, and the data granularity is accurate to 15 minutes. The data cleaning function built-in this module will automatically execute: for missing values caused by temporary communication interruption, linear interpolation method is used for filling; for obvious abnormal values (such as irradiance not being zero at night) caused by sensor failure, they are excluded and corrected according to the trend of the previous and subsequent data.

[0024] The prediction module starts at 2:00 a.m. every day to perform power prediction for the next 24 hours (with 15 minutes as a time point, a total of 96 points): In view of the intermittency and periodicity of photovoltaic power generation, this embodiment adopts an autoregressive integrated moving average model for prediction, which is expressed as: ; Model order determined by AIC criterion is (4, 1, 2), that is , , ; Among them, represents the photovoltaic power generation power at the time to be predicted ; to These represent the autoregressive coefficients, which are derived from a year's worth of historical data using the maximum likelihood estimation method, quantifying the first four time points. The power value for the current The degree of influence of the power value at any given time, for example It may be large, indicating Power at time There is a strong correlation at all times; and This represents the moving average coefficient, which is also derived from training. It is used to represent the impact of historical prediction errors (noise) on the current value, enabling the model to better respond to sudden fluctuations. express White noise at any given time represents the completely random part that the model cannot explain using historical data and coefficients, and its mean is zero. This indicates that the original power time series has undergone a difference processing to eliminate its instability and make it more stable, thus meeting the prerequisite for the application of the ARIMA model. The model uses 96*7 historical power data points from the past 7 days as input to predict the power values ​​for the next 96 points in a rolling manner. The evaluation unit calculates the root mean square error between the predicted and actual values.

[0025] Since load is strongly correlated with external variables such as weekdays / holidays and temperature, this embodiment uses a support vector machine regression model. Input features include: historical load for the same period, date type (weekday, weekend, holiday), predicted temperature, and humidity. The model training unit uses data from the past year to fit the relationship between these features and load. On the prediction day, the real-time prediction unit inputs the corresponding feature variables and outputs the load demand power for the next 96 time points. .

[0026] The strategy formulation module starts at 4:00 AM every day, based on the process described above. and Curve, formulate the optimal absorption strategy: Based on the predicted photovoltaic power generation and load demand, an optimization objective function is constructed to minimize the curtailment rate and grid interaction costs. The objective function to be optimized is: ; in, This represents the index of time points from 1 to 96; Indicates prediction Photovoltaic power generation at any given time (kW); express The actual photovoltaic power generation absorbed at any given time (kW) includes the portion directly supplied to the load. and the part charged into the energy storage ( ), This refers to the power of light discarded, and one of the optimization goals is to minimize its sum of squares; express The cost (in yuan) of constantly interacting with the power grid is calculated as follows: when When purchasing electricity from the grid, , when When (selling electricity to the grid), , in, and It is a time-of-use electricity price. The purchase price of electricity in the area where Power Station A is located is 1.2 yuan / kWh during peak hours (08:00-12:00, 18:00-22:00), 0.7 yuan / kWh during normal hours (12:00-18:00, 22:00-24:00), and 0.3 yuan / kWh during off-peak hours (00:00-08:00). The retail price of electricity is uniformly 0.4 yuan / kWh. and These are weighting coefficients, set based on the operational experience of Power Plant A. , This indicates that the current strategy focuses more on reducing energy waste (lowering the curtailment rate of solar power) while also taking into account economic operation. The constraints on the above objective function include power balance constraints, energy storage system constraints, and grid interaction power constraints: Power balance constraints: ; Constraints of energy storage systems: ( (To protect the battery, its state of charge should be limited to between 20% and 95%; the charging and discharging power should not exceed its rated power.) Power grid interaction constraints: (Limited by transformer capacity).

[0027] The optimization algorithm unit uses a genetic algorithm to solve this constrained, complex, nonlinear optimization problem: A solution (i.e., a complete 24-hour strategy) is encoded as a chromosome containing 96 time points. , , Decision variables; Define the fitness function as the reciprocal of the objective function to facilitate finding the maximum value: ; This is the fitness value, used to evaluate the quality of the solution; a larger value indicates a better solution. In the evolutionary process, a roulette wheel selection method is used. Assume the current population has 3 individuals with fitness values ​​of... , , The overall fitness is 50, and individual 3 has the highest selection probability. Individual 1 has a minimum percentage of 20%, ensuring a higher probability of passing on superior genes to the next generation; secondly, a single-point crossover is used to randomly select a pair of parent chromosomes at a random time point (e.g., The strategy involves exchanging the latter half of the genes to produce two new offspring; finally, the mutation probability is used to determine the offspring's genetic makeup. To introduce diversity and avoid getting trapped in local optima, certain genes on the chromosome are subjected to random small perturbations, such as randomly changing the charging power from 100kW to 95kW at a certain time point. The algorithm sets the maximum number of iterations to 1000 generations. When the fitness F of the optimal solution no longer increases significantly after 100 consecutive generations, it is considered to have converged. The strategy generation unit then decodes this optimal solution into a specific photovoltaic consumption strategy. This strategy specifies in detail how much power the energy storage system should charge / discharge in each 15-minute time period in the next 24 hours, as well as the power plan to be exchanged with the grid. The simulation unit will conduct simulation tests on the strategy based on historical typical daily data to verify whether it can still maintain good robustness under abnormal conditions (such as large prediction deviations).

[0028] Starting at 00:00 on that day, the execution module begins to work: According to the strategy document, the control unit sends control signals to the relevant devices at the beginning of each 15-minute period: Midday (11:00-14:00): High power generation is predicted, while load demand is moderate. Strategy: Prioritize meeting the full load of the industrial park. Remaining power Firstly, it is used to charge energy storage systems. If the energy storage is full or the charging power has reached its limit, the control command will adjust the output power of the photovoltaic inverter. Limited to This minimizes curtailment of solar power; only if there is still a very small surplus will the power be sold to the grid. ; Evening peak hours (18:00-20:00): Photovoltaic power generation is zero, load remains at its peak, and grid electricity prices are at their highest. Strategy instruction: Energy storage systems discharge at their maximum permissible power. Purchase electricity from the power grid This approach, which supports load demand together, significantly reduces the high costs associated with grid interaction. ; Real-time tracking by monitoring unit 、 、 Key indicators such as the above, and feed these data back to the data acquisition module as historical data for the next round of prediction and optimization, forming a closed-loop optimization; If the sensor detects that the actual energy storage system If the deviation between the actual value and the expected value of the strategy exceeds 10%, or the grid frequency is abnormal, it will immediately issue a warning and trigger an emergency handling program, such as suspending the optimization strategy and switching to a pre-set conservative safe operation mode to ensure system safety.

[0029] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the described embodiments, and those skilled in the art can make various equivalent modifications or replacements without deviating from the spirit of the present application. These equivalent modifications or replacements are all included within the scope defined by the claims of the present application.

Claims

1. A method for optimizing photovoltaic power consumption, characterized in that, Includes the following steps: Step S1: Obtain historical and real-time power generation data of the photovoltaic power generation system, as well as historical and real-time power consumption data of the load; Step S2: Based on historical power generation data and historical power consumption data, predict the photovoltaic power generation capacity and load demand power for future periods; Step S3: Based on the predicted photovoltaic power generation and load demand, formulate a photovoltaic consumption strategy to minimize the curtailment rate; Step S4: Implement the photovoltaic consumption strategy and control the operation of the photovoltaic power generation system and load.

2. The photovoltaic power consumption optimization method according to claim 1, characterized in that, In step S2, predicting the photovoltaic power generation and load power for future time periods includes: A time series analysis model was used to predict photovoltaic power generation, and a regression model was used to predict load demand power. The time series analysis model is trained based on the periodic characteristics of historical power generation data, and the regression model is fitted based on the external variables of historical electricity consumption data. The prediction process includes data preprocessing, model training, and real-time prediction sub-steps. Data preprocessing includes imputing missing values ​​and removing outliers. Model training uses cross-validation to optimize parameters. Real-time prediction updates data through a sliding window.

3. The photovoltaic power consumption optimization method according to claim 2, characterized in that, The time series analysis model is an autoregressive integral moving average model, expressed as: ； in, express Photovoltaic power generation at any given time; to It is an autoregressive coefficient, used to represent the degree of influence of historical power values ​​on the current value; to It is the moving average coefficient, used to represent the degree of influence of historical noise on the current value; yes The white noise sequence at each time step represents random errors that the model cannot explain; and It is the model order, representing the lag order of the autoregressive and moving average components, respectively, and is determined by minimizing the residual variance; The model parameters are solved by the maximum likelihood estimation method and trained using historical power generation data.

4. The photovoltaic power consumption optimization method according to claim 1, characterized in that, Step S3, formulating a photovoltaic grid integration strategy includes: Based on the predicted photovoltaic power generation and load demand, an optimization objective function is constructed to minimize the curtailment rate and grid interaction costs. The optimization objective function is: ； in, yes Photovoltaic power generation at any given time; yes The actual photovoltaic power generation absorbed at any given time; yes The cost of constantly interacting with the power grid, including the cost of purchasing and selling electricity; and This is a weighting coefficient used to balance the importance of curtailment rate and cost; its value ranges from 0 to 1. ; It is the total duration of the optimized time period; The optimization process employs linear programming or genetic algorithms, taking into account real-time fluctuation constraints of photovoltaic power generation and load.

5. The photovoltaic power consumption optimization method according to claim 4, characterized in that, The process of solving the optimization objective function includes: When using a genetic algorithm for multi-objective optimization, the fitness function is defined as: ； in, This is the fitness value, used to evaluate the quality of the solution; a larger value indicates a better solution. Genetic algorithms involve selection, crossover, and mutation. Selection is based on the roulette wheel method, crossover uses single-point crossover, and mutation uses probability. Randomly change some genes in the solution; It represents the mutation probability, with a value ranging from 0.01 to 0.

1. The optimization process is iterated until the fitness value converges or the maximum number of iterations is reached, thereby generating a photovoltaic consumption strategy, including energy storage system charging and discharging plans and load scheduling schemes.

6. A photovoltaic power consumption optimization system, used to implement the photovoltaic power consumption optimization method as described in any one of claims 1-5, characterized in that, include: The data acquisition module is used to acquire historical and real-time power generation data of the photovoltaic power generation system, as well as historical and real-time power consumption data of the load. The prediction module is used to predict the photovoltaic power generation and load demand power for future time periods based on the historical power generation data and historical power consumption data. The strategy formulation module is used to formulate photovoltaic consumption strategies based on the predicted photovoltaic power generation and load demand to minimize the curtailment rate. The execution module is used to execute the photovoltaic consumption strategy and control the operation of the photovoltaic power generation system and the load.

7. A photovoltaic power consumption optimization system according to claim 2, characterized in that, The data acquisition module includes: The sensor unit is used to collect real-time data on voltage, current, and irradiance of the photovoltaic power generation system, as well as the power data of the load. A communication unit is used to transmit data to a central processing unit via a wireless network or wired protocol. The storage unit is used to store historical power generation data and historical power consumption data, and a database management system is used for data indexing and querying. The data acquisition module also integrates a data cleaning function, which filters and normalizes the collected data to ensure data quality.

8. A photovoltaic power consumption optimization system according to claim 2, characterized in that, The prediction module includes: A model training unit is used to train a predictive model using machine learning algorithms, including support vector machines or neural networks. The real-time forecasting unit is used to output the photovoltaic power generation and load demand power for future time periods based on real-time data input. The evaluation unit is used to assess the prediction accuracy using the root mean square error index and to dynamically adjust the model parameters. The forecasting module also supports multi-source data fusion, combining meteorological data and calendar information to improve forecast accuracy.

9. A photovoltaic power consumption optimization system according to claim 2, characterized in that, The strategy formulation module includes: An optimization algorithm unit is used to perform the optimization objective function solution as described in claim 5 or 6; The strategy generation unit is used to generate specific photovoltaic consumption strategies based on the optimization results, including control commands, for energy storage systems and adjustable loads; The simulation unit is used to perform simulation tests before the policy is executed, and to evaluate the effectiveness and robustness of the policy. The strategy formulation module also integrates a human-computer interaction interface, allowing users to set weight coefficients and constraints.

10. A photovoltaic power consumption optimization system according to claim 2, characterized in that, The execution module includes: The control unit is used to send control signals to the photovoltaic power generation system, energy storage system, and load to achieve power regulation; The monitoring unit is used to monitor the effect of strategy execution in real time and feed back the data to the data acquisition module; An alarm unit is used to issue an alarm and trigger emergency response procedures when an anomaly is detected. The execution module also supports remote operation, enabling coordinated control of distributed photovoltaic power consumption through a cloud platform.

Images