Distributed photovoltaic power station optimization scheduling method

Abstract

The invention relates to the technical field of new energy power generation, in particular to a distributed photovoltaic power station optimization scheduling method. Comprising the following steps of data acquisition and preprocessing, photovoltaic power generation power prediction, load demand prediction, optimization scheduling model construction, multi-target optimization, energy storage system scheduling, real-time scheduling and feedback control. The invention provides a distributed photovoltaic power station optimization scheduling method based on data driving and multi-objective optimization, and aims to improve the operation efficiency, economical efficiency, environmental benefits and system stability of a photovoltaic power station.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy power generation, and particularly to an optimized scheduling method for a distributed photovoltaic power station. Background Art

[0002] With the growth of global energy demand and the increasingly severe environmental problems, the development and utilization of renewable energy have become an important development direction in the international energy field. Distributed photovoltaic power stations have received extensive attention and rapid development due to their environmental friendliness, flexible deployment, and effective utilization of solar energy resources. However, photovoltaic power generation is affected by multiple factors such as weather conditions and time, and has intermittency and instability, which brings great challenges to the scheduling management of the power station. At the same time, there is a mismatch between the power generation power of the distributed photovoltaic power station and the demand of the power consumption load. How to effectively solve this problem has become the key to improving the operation efficiency and economy of the distributed photovoltaic power station.

[0003] The scheduling methods of traditional distributed photovoltaic power stations mainly rely on empirical values or simple historical data statistics. These methods are difficult to adapt to complex and changeable operating environments and cannot achieve accurate power prediction. With the development of machine learning and big data technologies, it has become possible to achieve intelligent scheduling by constructing more accurate prediction models and optimization algorithms. Summary of the Invention

[0004] The present invention proposes an optimized scheduling method for a distributed photovoltaic power station based on data-driven and multi-objective optimization, aiming to improve the operation efficiency, economy, environmental benefits, and system stability of the photovoltaic power station.

[0005] The technical solution adopted by the present invention is as follows: An optimized scheduling method for a distributed photovoltaic power station, comprising the following steps:

[0006] Step 1, data collection and preprocessing: Real-time collect photovoltaic power generation data, meteorological data, and load demand data through the sensor network of the distributed photovoltaic power station, and preprocess the collected data, including data cleaning, missing value filling, and outlier processing, to ensure the accuracy and integrity of the data;

[0007] Step 2, photovoltaic power generation prediction: Based on the preprocessed data, use machine learning algorithms to construct a photovoltaic power generation prediction model to predict the photovoltaic power generation in the next period of time;

[0008] Step 3, load demand prediction: Based on historical load data and real-time load data, use time series analysis methods to predict the load demand in the next period of time;

[0009] Step 4, construction of an optimized scheduling model: According to the predicted photovoltaic power generation and load demand, construct an optimized scheduling model for the distributed photovoltaic power station;

[0010] Step 5, Multi-objective optimization: Introduce a multi-objective optimization algorithm into the optimization scheduling model, and consider economy, environmental benefits, and system stability simultaneously;

[0011] Step 6, Energy storage system scheduling: According to the results of the optimization scheduling model, formulate the charging and discharging strategies of the energy storage system to ensure that the energy storage system can effectively suppress power fluctuations and improve the stability and reliability of the system when the photovoltaic power generation fluctuates and the load demand changes;

[0012] Step 7, Real-time scheduling and feedback control: During the operation of the system, monitor the photovoltaic power generation, load demand, and the status of the energy storage system in real time, dynamically adjust the optimization scheduling strategy according to the real-time data, and correct the scheduling results through a feedback control mechanism to ensure that the system always operates in the optimal state.

[0013] As a further improvement of the present invention, the data cleaning includes removing duplicate data, correcting error data, and filling missing data; the missing value filling adopts linear interpolation method and K-nearest neighbor algorithm; the outlier processing adopts 3σ principle and box plot method.

[0014] As a further improvement of the present invention, in the photovoltaic power generation prediction step, the prediction model adopts an LSTM neural network, the input data includes historical photovoltaic power generation, meteorological data, and time characteristics, and the output data is the predicted value of the photovoltaic power generation for the next 24 hours.

[0015] As a further improvement of the present invention, in the load demand prediction step, the prediction model adopts an ARIMA model, the input data includes historical load data, time characteristics, and external factors, and the output is the predicted value of the load demand for the next 24 hours.

[0016] As a further improvement of the present invention, in the step of constructing the optimization scheduling model, the objective function is to minimize the system operation cost, and the constraint conditions include photovoltaic power generation constraint, load demand constraint, energy storage system charging and discharging constraint, and grid power balance constraint.

[0017] As a further improvement of the present invention, in the multi-objective optimization step, the objective functions include an economic objective (minimizing the operation cost), an environmental benefit objective (minimizing carbon emissions), and a system stability objective (minimizing power fluctuations), and the NSGA-II algorithm is used to solve the multi-objective optimization problem.

[0018] As a further improvement of the present invention, in the energy storage system scheduling step, the charging and discharging strategy is based on the photovoltaic power generation prediction and the load demand prediction, and the dynamic programming algorithm is used to formulate the optimal charging and discharging plan to ensure that the energy storage system charges during the low electricity price period and discharges during the high electricity price period to maximize the economic benefits.

[0019] As a further improvement of the present invention, in the real-time scheduling and feedback control step, a model predictive control method is adopted to dynamically adjust and optimize the scheduling strategy according to real-time data, and the scheduling result is corrected through a feedback control mechanism to ensure that the system always operates in an optimal state.

[0020] Advantages of the present invention: (1) Through the data acquisition and preprocessing step, the present invention ensures the accuracy and integrity of the input data, thereby improving the accuracy of photovoltaic power generation and load demand prediction. The combined use of the LSTM neural network and the ARIMA model can better capture the long-term dependence and periodic changes of time series data, making the prediction results more reliable, and thus improving the overall operation efficiency of the system.

[0021] (2) The present invention adopts the NSGA-II algorithm for multi-objective optimization, comprehensively considering economy, environmental benefits and system stability. By minimizing the operating cost, minimizing carbon emissions and minimizing power fluctuations, it can improve the environmental friendliness and operating stability of the system on the premise of ensuring economy, so as to better meet the needs of different users and scenarios and improve the comprehensive performance of the system.

[0022] (3) The present invention can dynamically adjust and optimize the scheduling strategy according to the real-time collected data, and correct the scheduling result through a feedback control mechanism, enabling the system to quickly respond to changes in the external environment and load demand, ensuring that the system always operates in an optimal state. The charge and discharge strategy of the energy storage system is based on the dynamic programming algorithm, which can charge during the low electricity price period and discharge during the high electricity price period to maximize economic benefits, while enhancing the reliability and stability of the system. Specific implementation manners

[0023] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present application clearer, the present application will be further described in detail below in conjunction with embodiments. It should be understood that the embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0024] The present invention provides a distributed photovoltaic power station optimization scheduling method, including the following steps:

[0025] Step 1, data acquisition and preprocessing: Real-time collect photovoltaic power generation data, meteorological data, and load demand data through the sensor network of the distributed photovoltaic power station, and preprocess the collected data, including data cleaning, missing value filling, and outlier processing, to ensure the accuracy and integrity of the data. The data cleaning includes removing duplicate data, correcting error data, and filling missing data; the missing value filling adopts linear interpolation method and K-nearest neighbor algorithm; the outlier processing adopts 3σ principle and box plot method;

[0026] Step 2, Photovoltaic power prediction: Based on the preprocessed data, a machine learning algorithm is used to construct a photovoltaic power prediction model to predict the photovoltaic power in the next period of time. The prediction model uses an LSTM neural network. The input data includes historical photovoltaic power, meteorological data, and time features, and the output data is the predicted value of the photovoltaic power for the next 24 hours;

[0027] Step 3, Load demand prediction: Based on historical load data and real-time load data, a time series analysis method is used to predict the load demand in the next period of time. The prediction model uses an ARIMA model. The input data includes historical load data, time features, and external factors, and the output is the predicted value of the load demand for the next 24 hours;

[0028] Step 4, Construction of an optimal scheduling model: According to the predicted photovoltaic power and load demand, an optimal scheduling model for a distributed photovoltaic power station is constructed. The objective function is to minimize the system operation cost, and the constraint conditions include photovoltaic power constraints, load demand constraints, charge and discharge constraints of the energy storage system, and grid power balance constraints;

[0029] Step 5, Multi-objective optimization: A multi-objective optimization algorithm is introduced into the optimal scheduling model, considering economy, environmental benefits, and system stability simultaneously. The objective functions include an economic objective (minimizing the operation cost), an environmental benefit objective (minimizing carbon emissions), and a system stability objective (minimizing power fluctuations). The NSGA-II algorithm is used to solve the multi-objective optimization problem;

[0030] Step 6, Energy storage system scheduling: According to the results of the optimal scheduling model, a charge and discharge strategy for the energy storage system is formulated to ensure that the energy storage system can effectively suppress power fluctuations and improve the stability and reliability of the system when the photovoltaic power fluctuates and the load demand changes. The charge and discharge strategy is based on the photovoltaic power prediction and load demand prediction, and a dynamic programming algorithm is used to formulate an optimal charge and discharge plan to ensure that the energy storage system charges during the low electricity price period and discharges during the high electricity price period to maximize economic benefits;

[0031] Step 7, Real-time scheduling and feedback control: During the operation of the system, the photovoltaic power, load demand, and status of the energy storage system are monitored in real time. The optimal scheduling strategy is dynamically adjusted according to the real-time data, and the scheduling results are corrected through a feedback control mechanism to ensure that the system always operates in the optimal state. The model predictive control method is used to dynamically adjust the optimal scheduling strategy according to the real-time data, and the scheduling results are corrected through a feedback control mechanism to ensure that the system always operates in the optimal state.

[0032] Example:

[0033] This example provides an optimal scheduling method for a distributed photovoltaic power station, and the specific steps are as follows:

[0034] (1) Data collection and preprocessing

[0035] The sensor network of the distributed photovoltaic power station collects the following data in real time: (1) Photovoltaic power generation data: including the output power, voltage, current, etc. of photovoltaic modules. (2) Meteorological data: including solar radiation intensity, temperature, humidity, wind speed, etc. (3) Load demand data: including the real-time electricity load of users, historical electricity load, etc.

[0036] The collected data may have duplicates, missing values or outliers, so preprocessing is required: (1) Data cleaning: Remove duplicate data and correct obvious error data (such as outliers caused by sensor failures). (2) Missing value filling: For missing data, use linear interpolation to fill continuous missing values and use the K-nearest neighbor algorithm to fill discrete missing values. (3) Outlier handling: Use the 3σ principle (that is, data outside the range of the mean ± 3 times the standard deviation is regarded as an outlier) and the box plot method (identify outliers through the interquartile range) for processing.

[0037] (2) Photovoltaic power generation prediction

[0038] Use the LSTM (Long Short-Term Memory) neural network as the photovoltaic power generation prediction model. LSTM can capture long-term dependencies in time series data and is suitable for processing the time series characteristics of photovoltaic power generation.

[0039] Input data: (1) Historical photovoltaic power generation: Photovoltaic power generation data for the past 24 hours. (2) Meteorological data: Meteorological forecast data such as solar radiation intensity, temperature, humidity, etc. for the next 24 hours. (3) Time features: including time information such as hour, week, season, etc.

[0040] Output data: The model outputs the predicted value of the photovoltaic power generation for the next 24 hours, and the prediction result will be used for subsequent optimal scheduling.

[0041] (3) Load demand prediction

[0042] Use the ARIMA (Autoregressive Integrated Moving Average) model for load demand prediction. The ARIMA model can handle trends and seasonal variations in time series data and is suitable for load demand prediction.

[0043] Input data: (1) Historical load data: Load demand data for the past 24 hours. (2) Time features: including time information such as hour, week, season, etc. (3) External factors: factors such as holidays and special events that may affect load demand.

[0044] Output data: The model outputs the predicted value of the load demand for the next 24 hours, and the prediction result will be used for the construction of the optimal scheduling model.

[0045] (4) Optimization of the scheduling model construction

[0046] Objective function: The objective of the optimized scheduling model is to minimize the system operation cost, which specifically includes: (1) Photovoltaic power generation cost: The marginal cost of photovoltaic power generation. (2) Energy storage system cost: The charge and discharge costs of the energy storage system. (3) Grid power purchase cost: When the photovoltaic power generation is insufficient, the cost of purchasing power from the grid.

[0047] Constraint conditions: (1) Photovoltaic power generation power constraint: The photovoltaic power generation power cannot exceed its maximum output capacity. (2) Load demand constraint: The system must meet the load demand of users. (3) Energy storage system charge and discharge constraint: The charge and discharge power and capacity of the energy storage system cannot exceed its rated value. (4) Grid power balance constraint: The total power generation of the system must be equal to the total load demand.

[0048] (5) Multi-objective optimization

[0049] Objective function: Introduce multi-objective optimization into the optimized scheduling model. The objective functions include: (1) Economic objective: Minimize the system operation cost. (2) Environmental benefit objective: Minimize carbon emissions. (3) System stability objective: Minimize power fluctuations

[0050] Optimization algorithm: Use NSGA-II (Non-dominated Sorting Genetic Algorithm) for multi-objective optimization. NSGA-II can effectively handle the trade-off relationships between multiple objectives and generate a set of Pareto optimal solutions for decision-makers to choose.

[0051] (6) Energy storage system scheduling

[0052] Charge and discharge strategy: Based on the prediction of photovoltaic power generation power and load demand, use the dynamic programming algorithm to formulate the optimal charge and discharge plan for the energy storage system. The specific strategies are as follows: (1) Charge during low electricity price periods: During periods with low electricity prices (such as at night), the energy storage system charges to reduce the charging cost. (2) Discharge during high electricity price periods: During periods with high electricity prices (such as during the day), the energy storage system discharges to reduce the cost of purchasing power from the grid.

[0053] Objective: Through the charge and discharge strategy of the energy storage system, maximize the economic benefits, while suppressing the fluctuations of photovoltaic power generation power and improving the stability and reliability of the system.

[0054] (7) Real-time scheduling and feedback control

[0055] Real-time monitoring: During the operation of the system, the following data is monitored in real time: (1) Photovoltaic power generation power: The deviation between the real-time photovoltaic power generation power and the predicted value. (2) Load demand: The deviation between the real-time load demand and the predicted value. (3) Energy storage system status: The charge and discharge status, remaining capacity, etc. of the energy storage system.

[0056] Dynamic adjustment: The model predictive control (MPC) method is adopted to dynamically adjust and optimize the scheduling strategy according to real-time data. The specific steps are as follows: (1) Real-time data input: Input the real-time monitored photovoltaic power generation, load demand, and energy storage system status into the optimization scheduling model. (2) Re-optimization: Based on the real-time data, re-perform the optimization scheduling calculation to generate a new scheduling strategy. (3) Feedback control: Correct the scheduling results through a feedback control mechanism to ensure that the system always operates in an optimal state.

[0057] Feedback control mechanism: (1) Deviation correction: When there is a large deviation between the real-time data and the predicted value, the system will automatically adjust the charge and discharge strategy of the energy storage system to cope with sudden changes in photovoltaic power generation or load demand. (2) Stability guarantee: Through feedback control, the system can quickly respond to changes in the external environment and load demand, ensuring the stability and reliability of the system.

[0058] System operation effect

[0059] Through the above steps, the optimization scheduling system of the distributed photovoltaic power station can achieve the following effects: (1) Improve prediction accuracy: By combining the LSTM and ARIMA models, the prediction accuracy of photovoltaic power generation and load demand is significantly improved, reducing the prediction error. (2) Optimize operating costs: Through multi-objective optimization and reasonable scheduling of the energy storage system, the system operating costs are significantly reduced, and the economic benefits are improved. (3) Reduce carbon emissions: Through optimized scheduling, the system reduces the demand for power purchase from the grid, reduces carbon emissions, and improves the environmental benefits. (4) Enhance system stability: The charge and discharge strategy of the energy storage system effectively suppresses the fluctuations in photovoltaic power generation, improving the stability and reliability of the system.

[0060] Case analysis

[0061] A certain distributed photovoltaic power station has an installed capacity of 1 MW, an energy storage system capacity of 500 kWh, and the grid power purchase prices are 0.5 yuan / kWh (peak hours) and 0.3 yuan / kWh (valley hours). Through the optimized scheduling method of the present invention, the operation effect of the system within one day is as follows: (1) Photovoltaic power generation prediction: The LSTM model predicts the photovoltaic power generation for the next 24 hours, and the prediction error is less than 5%. (2) Load demand prediction: The ARIMA model predicts the load demand for the next 24 hours, and the prediction error is less than 3%. (3) Optimized scheduling result: Through the NSGA-II algorithm, on the premise of ensuring the load demand, the operating cost of the system is reduced by 15%, the carbon emissions are reduced by 10%, and the power fluctuation is reduced by 20%. (4) Energy storage system scheduling: The energy storage system charges during the low electricity price period and discharges during the high electricity price period, maximizing the economic benefits.

[0062] Conclusion

[0063] Through the method of data-driven and multi-objective optimization, the present invention realizes the intelligent scheduling of distributed photovoltaic power stations. By combining the LSTM and ARIMA models, the prediction accuracy of photovoltaic power generation and load demand is improved; through the NSGA-II algorithm and dynamic programming algorithm, the operating cost, environmental benefits and stability of the system are optimized; through real-time scheduling and feedback control, the system can quickly respond to external changes and ensure that it always operates in the optimal state. This method has broad application prospects and can effectively improve the operating efficiency and economy of distributed photovoltaic power stations.

[0064] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the scheduling of a distributed photovoltaic power station, characterized in that, It includes the following steps: Step 1, data collection and preprocessing: Real-time collect photovoltaic power generation data, meteorological data, and load demand data through the sensor network of the distributed photovoltaic power station, and preprocess the collected data, including data cleaning, missing value filling, and outlier handling, to ensure the accuracy and integrity of the data; Step 2, photovoltaic power generation prediction: Based on the preprocessed data, use machine learning algorithms to construct a photovoltaic power generation prediction model to predict the photovoltaic power generation in the future for a period of time; Step 3, load demand prediction: Based on historical load data and real-time load data, use time series analysis methods to predict the load demand in the future for a period of time; Step 4, construction of the optimal scheduling model: According to the predicted photovoltaic power generation and load demand, construct the optimal scheduling model of the distributed photovoltaic power station; Step 5, multi-objective optimization: Introduce multi-objective optimization algorithms into the optimal scheduling model, and consider economy, environmental benefits, and system stability at the same time; Step 6, energy storage system scheduling: According to the results of the optimal scheduling model, formulate the charge and discharge strategies of the energy storage system to ensure that the energy storage system can effectively suppress power fluctuations and improve the stability and reliability of the system when the photovoltaic power generation fluctuates and the load demand changes; Step 7, real-time scheduling and feedback control: During the operation of the system, real-time monitor the state of photovoltaic power generation, load demand, and energy storage system, dynamically adjust the optimal scheduling strategy according to real-time data, and correct the scheduling results through the feedback control mechanism to ensure that the system always operates in the optimal state.

2. The optimized scheduling method for a distributed photovoltaic power station according to claim 1, wherein The data cleaning includes removing duplicate data, correcting incorrect data, and filling missing data; the missing value filling adopts linear interpolation method and K-nearest neighbor algorithm; the outlier handling adopts 3σ principle and box plot method.

3. A distributed photovoltaic power station optimal scheduling method according to claim 1, characterized in that, In the step of photovoltaic power generation prediction, the prediction model adopts an LSTM neural network, the input data includes historical photovoltaic power generation, meteorological data, and time features, and the output data is the predicted value of photovoltaic power generation in the next 24 hours.

4. A method for optimizing the scheduling of a distributed photovoltaic power station according to claim 1, characterized in that, In the step of load demand prediction, the prediction model adopts an ARIMA model, the input data includes historical load data, time features, and external factors, and the output is the predicted value of load demand in the next 24 hours.

5. A method for optimizing the scheduling of a distributed photovoltaic power station according to claim 1, characterized in that, In the step of constructing the optimal scheduling model, the objective function is to minimize the system operation cost, and the constraint conditions include photovoltaic power generation constraint, load demand constraint, energy storage system charge and discharge constraint, and grid power balance constraint.

6. The optimization scheduling method for a distributed photovoltaic power station according to claim 1, characterized in that In the step of multi-objective optimization, the objective functions include an economic objective (minimizing operation cost), an environmental benefit objective (minimizing carbon emissions), and a system stability objective (minimizing power fluctuations), and the NSGA-II algorithm is used to solve the multi-objective optimization problem.

7. A method for optimizing the scheduling of a distributed photovoltaic power station according to claim 1, characterized in that, In the step of energy storage system scheduling, the charge and discharge strategy is based on the prediction of photovoltaic power generation and load demand, and the dynamic programming algorithm is used to formulate the optimal charge and discharge plan to ensure that the energy storage system charges during the low electricity price period and discharges during the high electricity price period to maximize economic benefits.

8. A method for optimizing the scheduling of a distributed photovoltaic power station according to claim 1, characterized in that In the real-time scheduling and feedback control steps, the model predictive control method is adopted to dynamically adjust and optimize the scheduling strategy according to real-time data, and the feedback control mechanism is used to correct the scheduling results to ensure that the system always operates in the optimal state.