A multi-time-scale robust control algorithm optimization method for photovoltaic inverters

Abstract

The present invention relates to the technical field of photovoltaic power generation regulating electric variables. The present invention provides a photovoltaic inverter multi-time scale robust control algorithm optimization method, which aims to solve the problem that the model parameters of the multi-time scale robust control algorithm are often difficult to obtain accurately, and will change with the change of the operating state of the photovoltaic power generation system, resulting in reduced accuracy of photovoltaic power generation output power regulation. The present invention obtains photovoltaic power generation system operation data, inverter operation data, natural environment data and power grid status data in real time, and uses online learning algorithms to perform parameter identification and adaptive adjustment, constructs short-term, medium-term and long-term control models, and realizes multi-time scale refined control. The present invention significantly improves the regulation accuracy, robustness and stability of photovoltaic power generation output power, while enhancing the system's adaptive ability, and promoting the intelligent and economic development of photovoltaic power generation systems.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power generation regulating electric variables, and in particular to a multi-time scale robust control algorithm optimization method for a photovoltaic inverter. Background Art

[0002] A photovoltaic inverter is a device that converts the direct current generated by a photovoltaic solar panel into alternating current. In a photovoltaic array system, the photovoltaic inverter can be fed back to the commercial transmission system or used for off-grid power grids. The working principle of a photovoltaic inverter involves several key parts, including inverter bridge, logic control, filter circuit, etc. The main working process of a photovoltaic inverter is as follows: Inverter circuit: converts direct current into alternating current. Filter circuit: filters out unnecessary information to make the output waveform smoother. Control system: controls the operation of the entire system to ensure the stability and efficiency of the inverter process.

[0003] The multi-time scale robust control algorithm is a control strategy that combines model predictive control with robust optimization theory. The multi-time scale robust control algorithm is designed to handle dynamic systems with uncertainties and complex constraints. The multi-time scale robust control algorithm improves the stability and performance of the system by optimizing and controlling at different time scales.

[0004] Multi-time scale: refers to the process of decomposing the control problem into multiple levels or scales according to different time resolutions and prediction lengths in the control system, which can more effectively utilize system information and improve control accuracy and response speed.

[0005] Robust control: Robust control is a control method that can maintain the desired performance when there are uncertainties or disturbances in the system. Robust control methods usually assume information about the dynamic characteristics of the process and its range of variation, and ensure the stability of the system within a certain range by designing the controller.

[0006] The performance of photovoltaic inverters directly affects the stability and power generation efficiency of the entire system. The multi-time scale robust control algorithm can effectively deal with uncertain factors in the system, such as changes in light intensity, temperature, grid voltage fluctuations, etc., by optimizing the control strategy of photovoltaic inverters at different time scales, thereby improving the stability and robustness of the system. This algorithm can quickly adjust the control parameters when the system is disturbed, maintain the stability of the system output power, and reduce fluctuations and errors.

[0007] In the process of adjusting the output power of photovoltaic inverters, the multi-time-scale robust control algorithm can achieve precise control of electrical variables. Specifically, the multi-time-scale robust control algorithm can optimize and adjust the inverter's output voltage, current and other electrical variables according to the real-time status and prediction information of the system to maximize the power output and meet the requirements of the power grid. At the same time, the multi-time-scale robust control algorithm can also take into account the nonlinear characteristics and constraints of the system, and continuously improve the control accuracy and response speed through rolling optimization and feedback correction.

[0008] Although multi-time-scale robust control algorithms have made significant progress in the field of photovoltaic power generation output power regulation technology, there are still the following technical pain points that need to be solved. The accuracy of multi-time-scale robust control algorithms in photovoltaic power generation output power regulation depends largely on the accuracy of model parameters. However, in practical applications, the model parameters of multi-time-scale robust control algorithms are often difficult to obtain accurately, and they will change with the operating state of the photovoltaic power generation system. Therefore, how to accurately identify and adaptively adjust the model parameters of the multi-time-scale robust control algorithm to improve the accuracy of photovoltaic power generation output power regulation is a problem that needs to be solved. Summary of the invention

[0009] In view of the shortcomings of the prior art, the present invention provides a method for optimizing a multi-time-scale robust control algorithm for a photovoltaic inverter, so as to solve the problem that the model parameters of the multi-time-scale robust control algorithm are often difficult to obtain accurately and will change with the operating state of the photovoltaic power generation system, resulting in reduced accuracy in regulating the output power of the photovoltaic power generation.

[0010] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0011] The present invention provides a photovoltaic inverter multi-time scale robust control algorithm optimization method, comprising:

[0012] Step S101, obtaining photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid state data and photovoltaic power generation expected target data, the photovoltaic power generation system operation data includes photovoltaic power generation system voltage data, photovoltaic power generation system current data, light intensity, temperature data, photovoltaic inverter operation data includes photovoltaic inverter state and photovoltaic inverter operation parameter data, natural environment data includes temperature data, wind data and humidity data, grid state data includes grid operation state and grid load data, photovoltaic power generation expected target data includes expected power generation efficiency and expected power quality indicators;

[0013] Step S102, preprocessing the acquired photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data and photovoltaic power generation expected target data to obtain preprocessed photovoltaic power generation system data, and extracting features from the preprocessed photovoltaic power generation data to obtain a photovoltaic power generation system feature set, the photovoltaic power generation system feature set including photovoltaic power generation system voltage features, photovoltaic power generation system current features, photovoltaic power generation system light intensity features, photovoltaic power generation system temperature features, photovoltaic inverter operation state features, photovoltaic inverter operation parameter features, natural environment temperature features, natural environment wind force features, natural environment humidity features, power grid operation state features and power grid load features;

[0014] Step S103, using a line learning algorithm to analyze the operation data of the photovoltaic power generation system, identify the parameters of the photovoltaic inverter multi-time scale robust control algorithm, update the parameters of the photovoltaic inverter multi-time scale robust control algorithm in real time, receive the photovoltaic inverter multi-time scale robust control algorithm adaptive adjustment strategy, and dynamically adjust the identified photovoltaic inverter multi-time scale robust control algorithm parameters according to the real-time identification of the photovoltaic inverter multi-time scale robust control algorithm operation state and the identification of the photovoltaic inverter multi-time scale robust control algorithm parameters, and obtain the photovoltaic inverter multi-time scale robust control algorithm model after the dynamic adjustment of the parameters;

[0015] Step S104, constructing a short-term control model, a medium-term control model and a long-term control model based on the photovoltaic power generation system feature set, integrating the short-term control model, the medium-term control model and the long-term control model into the control framework, and establishing a model switching mechanism among the short-term control model, the medium-term control model and the long-term control model;

[0016] Step S105, deploy the short-term control model, the medium-term control model, the long-term control model and the model switching mechanism to the actual photovoltaic power generation system, and compare and detect the data in the actual photovoltaic power generation system according to the expected target data of photovoltaic power generation to obtain the photovoltaic power generation output electric power comparison result. If the output electric power in the photovoltaic power generation output electric power comparison result does not meet the expected electric power, the optimization algorithm is used to optimize the short-term control model, the medium-term control model, the long-term control model and the model switching mechanism until the output electric power in the photovoltaic power generation output electric power comparison result meets the expected electric power. Stop the optimization.

[0017] Furthermore, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S101:

[0018] The voltage data and current data of the photovoltaic power generation system are obtained by real-time measurement and recording through sensors and monitoring equipment installed in the photovoltaic power generation system;

[0019] Use a light sensor to measure the light intensity of the photovoltaic panel to obtain light intensity data;

[0020] Acquire temperature data by monitoring the temperature of photovoltaic panels and inverters through temperature sensors;

[0021] Obtain the status of the PV inverter through the monitoring system or communication interface inside the inverter;

[0022] The operating parameter data of the photovoltaic inverter is obtained by providing an output interface for the operating parameters.

[0023] Furthermore, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S101:

[0024] Temperature data, wind data, and humidity data are obtained through the weather station data interface or environmental monitoring equipment;

[0025] The grid operation status is obtained through the communication interface with the grid operator or the public grid status information;

[0026] Grid load dataGrid operators provide real-time load data or obtain it through power monitoring systems;

[0027] The expected target data of photovoltaic power generation includes the expected output power of the photovoltaic power generation system;

[0028] The expected power quality indicators include the expected power quality indicators of the power grid.

[0029] Furthermore, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S102:

[0030] Establishing a data preprocessing sequence for the acquired photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data, and photovoltaic power generation expected target data, preprocessing the photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data, and photovoltaic power generation expected target data according to the data preprocessing sequence to obtain the preprocessed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data, and photovoltaic power generation expected target data;

[0031] Random screening is performed on the pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid status data and photovoltaic power generation expected target data to obtain pre-processed data randomly screened data;

[0032] The data in the randomly screened data after preprocessing are compared with the acquired original photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and expected target data of photovoltaic power generation. If the data in the randomly screened data after preprocessing exists in the original photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and expected target data of photovoltaic power generation, the preprocessed data is processed successfully.

[0033] Furthermore, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S102:

[0034] The pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data are stored in the blockchain, and a hash value corresponding to the pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data is generated;

[0035] Establish data retrieval and use rules. When receiving a data retrieval task command, parse the data retrieval task command to obtain data demand information. According to the data demand information, retrieve the pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data stored in the blockchain;

[0036] The retrieved pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data are parsed through hash values ​​to obtain usable pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data.

[0037] Furthermore, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S103:

[0038] Extract features related to the multi-time-scale robust control algorithm of the photovoltaic inverter from the preprocessed data. The features related to the multi-time-scale robust control algorithm of the photovoltaic inverter include voltage features of the photovoltaic power generation system, current features of the photovoltaic power generation system, light intensity features of the photovoltaic power generation system, temperature features of the photovoltaic power generation system, operating state features of the photovoltaic inverter, operating parameter features of the photovoltaic inverter, temperature features of the natural environment, wind features of the natural environment, humidity features of the natural environment, operating state features of the power grid, and load features of the power grid;

[0039] Online learning algorithms include recursive least squares and online support vector machines.

[0040] The extracted features are analyzed using an online learning algorithm to identify the parameters of the multi-time-scale robust control algorithm for photovoltaic inverters and obtain the identification results of the online learning algorithm.

[0041] According to the recognition results of the online learning algorithm, the parameters of the multi-time-scale robust control algorithm of the photovoltaic inverter are updated in real time.

[0042] Furthermore, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S103:

[0043] Analyze the adaptive adjustment strategy of the multi-time scale robust control algorithm of photovoltaic inverters, and obtain the basis for adjusting the adjustment parameters and adjustment targets;

[0044] Based on the operation data of photovoltaic power generation system, the model parameters of the multi-time scale robust control algorithm of photovoltaic inverter are identified by using parameter identification algorithm;

[0045] Determine the adjustment parameters and their adjustment direction according to the received adaptive adjustment strategy and the real-time identified operating status and model parameters;

[0046] Substituting the adjusted parameters into the multi-time-scale robust control algorithm model of the photovoltaic inverter, and updating the multi-time-scale robust control algorithm model of the photovoltaic inverter;

[0047] The updated multi-time-scale robust control algorithm model of photovoltaic inverter is verified and evaluated using real-time photovoltaic power generation system operation data. The adjusted parameters are corrected based on the verification and evaluation results.

[0048] Furthermore, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S104:

[0049] Based on the photovoltaic power generation system feature set, a neural network model is used for training to obtain a short-term control model. The short-term control model is used to respond to changes in light intensity and output a photovoltaic inverter power generation control command at the current moment.

[0050] Based on the photovoltaic power generation system feature set, a neural network model is used for training to obtain a medium-term control model. The medium-term control model is used to respond to weather trend changes and output photovoltaic inverter power generation regulation control commands in the next few hours.

[0051] Based on the feature set of the photovoltaic power generation system, a neural network model is used for training to obtain a long-term control model. The long-term control model is used to respond to seasonal changes and output photovoltaic inverter power generation regulation control commands for long-term power generation.

[0052] Furthermore, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S104:

[0053] The data output by the short-term control model, the medium-term control model and the long-term control model are compared with the real-time photovoltaic power generation system operation data to obtain the comparison results between the predicted data and the real-time data. The comparison results between the predicted data and the real-time data include the short-term prediction error, the medium-term prediction error and the long-term prediction error;

[0054] When the short-term prediction error exceeds the preset threshold, the medium-term control model is switched to process the data. If the medium-term prediction error exceeds the preset threshold, the long-term control model is switched to process the data.

[0055] Furthermore, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S105:

[0056] According to the expected power generation efficiency and expected power quality indicators in the expected target data of photovoltaic power generation, an optimization target is set, and the optimization target is the expected value of the output power of photovoltaic power generation;

[0057] Compare the actual photovoltaic power generation output electric energy power with the expected value of photovoltaic power generation output electric energy power, and calculate the error between the actual power generation and the expected value of photovoltaic power generation output electric energy power. The error between the actual power generation and the expected value of photovoltaic power generation output electric energy power includes the total error and the errors of each time scale in the short term, medium term and long term.

[0058] The optimization algorithms include gradient descent, genetic algorithm, and particle swarm optimization. Configure the gradient descent method, set the initial parameters, learning rate, number of iterations, and population size of the gradient descent method, genetic algorithm, and particle swarm optimization.

[0059] Short-term control model: To address the short-term power generation forecast error, the gradient descent method is used to adjust the parameters of the short-term control model.

[0060] Medium-term control model: Based on the error of medium-term power generation forecast, the genetic algorithm is used to optimize the parameters of the medium-term control model;

[0061] Long-term control model: To address the errors in long-term power generation trend prediction, the particle swarm optimization algorithm is used to adjust the parameters of the long-term control model;

[0062] The optimization is stopped when the error between the actual power generation and the expected value of the photovoltaic power output electric energy power meets the expected electric energy power.

[0063] Beneficial effects of the present invention: By acquiring and processing photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data in real time, the present invention can more accurately understand the real-time status of the photovoltaic power generation system and external environmental changes, and then optimize the control algorithm parameters to improve the regulation accuracy of photovoltaic power generation output power.

[0064] The present invention adopts a multi-time scale robust control algorithm, which can quickly adjust control parameters when the system is disturbed by uncertain factors such as changes in light intensity, temperature, and grid voltage fluctuations, thereby maintaining the stability of the system output power, reducing fluctuations and errors, and significantly enhancing the robustness and stability of the system.

[0065] Through online learning algorithms and adaptive adjustment strategies, the present invention can update and optimize the parameters of the multi-time scale robust control algorithm of the photovoltaic inverter in real time, so that it can be dynamically adjusted as the operating state of the photovoltaic power generation system changes, thereby improving the adaptive ability of the photovoltaic power generation system.

[0066] The present invention constructs control models of three different time scales: short-term, medium-term and long-term, and automatically selects the appropriate control model under different operating conditions through the model switching mechanism, thereby realizing multi-time-scale refined control of the photovoltaic power generation system and further optimizing the power generation efficiency and power quality. By integrating the optimization algorithm and the multi-time-scale control model, the present invention makes the control process of the photovoltaic power generation system intelligent, realizes closed-loop control from data collection, preprocessing, feature extraction, model training to real-time optimization and adjustment, and greatly improves the intelligence level of the photovoltaic power generation system.

[0067] The present invention improves the regulation accuracy and stability of photovoltaic power output power, which helps to improve the overall power generation efficiency of the photovoltaic power generation system, reduce maintenance costs, increase the economy and reliability of photovoltaic power generation, and thus promote the widespread application and development of renewable energy.

[0068] In summary, the present invention shows significant beneficial effects in improving the output power regulation accuracy, robustness, adaptability, intelligence level, economy and reliability of photovoltaic power generation systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0070] Figure 1 A schematic diagram of a method flow chart provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. The technical solutions provided by the embodiments of the present invention are described in detail below in conjunction with the drawings.

[0072] In order to better understand the purpose of the present invention, the present invention is described in further detail below.

[0073] See also Figure 1 The present invention provides a photovoltaic inverter multi-time scale robust control algorithm optimization method, comprising:

[0074] Step S101, obtaining photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid state data and photovoltaic power generation expected target data, the photovoltaic power generation system operation data includes photovoltaic power generation system voltage data, photovoltaic power generation system current data, light intensity, temperature data, photovoltaic inverter operation data includes photovoltaic inverter state and photovoltaic inverter operation parameter data, natural environment data includes temperature data, wind data and humidity data, grid state data includes grid operation state and grid load data, photovoltaic power generation expected target data includes expected power generation efficiency and expected power quality indicators;

[0075] Photovoltaic power generation system voltage data: reflects the current voltage status of the photovoltaic power generation system. The photovoltaic power generation system voltage data is one of the key indicators for evaluating system stability and efficiency.

[0076] Photovoltaic power generation system current data: provides the output of system current, which is of great significance for calculating power generation and monitoring system load.

[0077] Light intensity: directly affects the power generation efficiency of photovoltaic panels and is an important reference factor for adjusting the output power of photovoltaic inverters.

[0078] Temperature data: The temperature of PV panels and inverters affects their performance and efficiency, so monitoring temperature data is necessary to optimize control algorithms.

[0079] PV inverter status: includes the operating status of the inverter, such as whether it is operating normally, whether there is a fault, etc.

[0080] Photovoltaic inverter operating parameter data: including key parameters such as output voltage, current, power factor, etc. These parameters directly reflect the operating efficiency and control effect of the inverter.

[0081] Temperature data: In addition to the temperature of the photovoltaic equipment and inverter, the temperature of the external environment also affects the power generation efficiency.

[0082] Wind data: Wind affects the light received by the photovoltaic array, which indirectly affects the power generation efficiency.

[0083] Humidity data: Humidity has a certain impact on the electrical performance and heat dissipation of photovoltaic equipment, so it is also one of the factors that need to be considered.

[0084] Grid operation status: including whether it is connected to the grid, voltage stability and other information, which is crucial for the coordinated operation of the photovoltaic inverter and the grid.

[0085] Grid load data: Real-time grid load data helps the inverter adjust output power to match the grid demand.

[0086] Expected power generation efficiency: The target power generation efficiency is set to evaluate whether the actual power generation efficiency reaches the expected level.

[0087] Expected power quality indicators: including power quality related indicators such as voltage fluctuation and harmonic content, to ensure that the power output of the inverter meets the requirements of the power grid.

[0088] By fully acquiring these data, the present invention can provide a rich information basis for subsequent control algorithm optimization, thereby improving the stability and power generation efficiency of the photovoltaic power generation system.

[0089] Step S102, preprocessing the acquired photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data and photovoltaic power generation expected target data to obtain preprocessed photovoltaic power generation system data, and extracting features from the preprocessed photovoltaic power generation data to obtain a photovoltaic power generation system feature set, the photovoltaic power generation system feature set including photovoltaic power generation system voltage features, photovoltaic power generation system current features, photovoltaic power generation system light intensity features, photovoltaic power generation system temperature features, photovoltaic inverter operation state features, photovoltaic inverter operation parameter features, natural environment temperature features, natural environment wind force features, natural environment humidity features, power grid operation state features and power grid load features;

[0090] Remove noise and outliers to ensure data accuracy and reliability, including checking for errors, missing values, outliers, etc. in the data, and filling or correcting them appropriately. Format all data into a unified format for subsequent processing and analysis. For example, sample data at different time points at a unified time interval. Standardize or normalize certain features to eliminate the impact of different dimensions on model training.

[0091] Features useful for optimizing the multi-time-scale robust control algorithm of photovoltaic inverters are extracted from the data to form a set of photovoltaic power generation system features. These features include:

[0092] Photovoltaic power generation system characteristics:

[0093] Voltage characteristics: real-time voltage value of the photovoltaic power generation system and its changing trend.

[0094] Current characteristics: real-time current value of the photovoltaic power generation system and its changing trend.

[0095] Light intensity characteristics: The light intensity received by the photovoltaic panels and its changes directly affect the output power of the photovoltaic power generation system.

[0096] Temperature characteristics:

[0097] Photovoltaic power generation system temperature characteristics: Temperature data of photovoltaic panels and inverters, used to evaluate the impact of temperature on power generation efficiency.

[0098] Natural ambient temperature characteristics: External ambient temperature affects the performance of photovoltaic equipment.

[0099] Photovoltaic inverter features:

[0100] Inverter operating status characteristics: The inverter's operating status information, such as whether it is operating normally, whether there are any faults, etc.

[0101] Inverter operating parameter characteristics: The key operating parameters of the inverter, such as output voltage, current, power factor, etc., reflect the stability and efficiency of the inverter process.

[0102] Natural environment characteristics:

[0103] Wind characteristics: Wind data that affects the light exposure and heat dissipation of the photovoltaic array.

[0104] Humidity characteristics: Humidity data that affects the electrical performance and heat dissipation of photovoltaic equipment.

[0105] Grid status characteristics:

[0106] Grid operation status characteristics: real-time operation status information of the grid, such as whether it is connected to the grid, voltage stability, etc.

[0107] Grid load characteristics: Real-time load data of the grid, used to adjust the output power of the photovoltaic power generation system to match the grid demand.

[0108] By extracting these features, the operating status and external environmental conditions of the photovoltaic power generation system can be described more accurately, providing strong support for the subsequent optimization of the control algorithm. At the same time, these features are also the basis for building short-term, medium-term and long-term control models.

[0109] Step S103, using a line learning algorithm to analyze the operation data of the photovoltaic power generation system, identify the parameters of the photovoltaic inverter multi-time scale robust control algorithm, update the parameters of the photovoltaic inverter multi-time scale robust control algorithm in real time, receive the photovoltaic inverter multi-time scale robust control algorithm adaptive adjustment strategy, and dynamically adjust the identified photovoltaic inverter multi-time scale robust control algorithm parameters according to the real-time identification of the photovoltaic inverter multi-time scale robust control algorithm operation state and the identification of the photovoltaic inverter multi-time scale robust control algorithm parameters, and obtain the photovoltaic inverter multi-time scale robust control algorithm model after the dynamic adjustment of the parameters;

[0110] In step S103, the present invention uses an online learning algorithm to analyze the operation data of the photovoltaic power generation system to optimize and dynamically adjust the parameters of the photovoltaic inverter multi-time scale robust control algorithm. The specific process is as follows:

[0111] The pre-processed photovoltaic power generation system operation data is analyzed using online learning algorithms (such as recursive least squares method, online support vector machine, etc.). The use of online learning algorithms can process newly arrived data in real time and continuously optimize model parameters according to data updates.

[0112] The current parameters of the multi-time-scale robust control algorithm for photovoltaic inverters are identified through an online learning algorithm. According to the identification results of the online learning algorithm, the parameters of the multi-time-scale robust control algorithm for photovoltaic inverters are updated in real time. The real-time update mechanism improves the algorithm's ability to adapt to the changing operating status of the photovoltaic power generation system.

[0113] The adaptive adjustment strategy of the multi-time-scale robust control algorithm of the photovoltaic inverter is received. The adaptive adjustment strategy of the multi-time-scale robust control algorithm of the photovoltaic inverter is formulated based on the system operation status, external environment changes or specific optimization goals.

[0114] Based on the real-time identification of the algorithm operation status and model parameters, the parameters in the algorithm are dynamically adjusted in combination with the adaptive adjustment strategy. The adaptive adjustment strategy can be a customized adjustment for specific operating conditions or optimization goals.

[0115] After dynamic adjustment, an updated photovoltaic inverter multi-time scale robust control algorithm model is obtained. The updated photovoltaic inverter multi-time scale robust control algorithm model can more accurately reflect the current operating status and external environmental conditions of the photovoltaic power generation system, thereby improving the regulation accuracy and stability of the photovoltaic power generation output power.

[0116] The present invention overcomes the problem that traditional control algorithms have fixed model parameters and are difficult to adapt to dynamic changes in the system. The combination of online learning algorithms and adaptive adjustment strategies enables the photovoltaic inverter multi-time scale robust control algorithm to continuously learn and optimize its own parameters, thereby maintaining high performance under various operating conditions.

[0117] Step S104, constructing a short-term control model, a medium-term control model and a long-term control model based on the photovoltaic power generation system feature set, integrating the short-term control model, the medium-term control model and the long-term control model into the control framework, and establishing a model switching mechanism among the short-term control model, the medium-term control model and the long-term control model;

[0118] In step S104, the present invention constructs control models of three different time scales, namely short-term, medium-term and long-term, based on the photovoltaic power generation system feature set, and integrates the control models of three different time scales into a unified control framework. At the same time, a switching mechanism between these models is also established to ensure that the appropriate control model can be automatically selected under different operating conditions. The specific steps are as follows:

[0119] Build a short-term control model: The short-term control model is mainly used to respond to rapidly changing environmental factors, such as instantaneous fluctuations in light intensity. It is based on real-time data in the photovoltaic power generation system feature set (such as current light intensity, temperature, voltage, current, etc.), and is trained using neural networks or other machine learning methods. The goal of the short-term control model is to quickly adjust the output of the photovoltaic inverter to respond to changes in light intensity and maintain stable power generation efficiency.

[0120] Constructing a medium-term control model: The medium-term control model is used to respond to environmental changes on a longer time scale, such as changes in weather trends (such as cloudy days turning into sunny days). It is also based on the photovoltaic power generation system feature set, but considers the data trend over a period of time. The medium-term control model predicts environmental changes in the next few hours and adjusts the output of the photovoltaic inverter in advance to optimize the overall power generation efficiency.

[0121] Building a long-term control model: The long-term control model is used to respond to seasonal changes or long-term trends. It relies on data over a longer time frame (such as historical meteorological data, seasonal change patterns, etc.), as well as the long-term operating characteristics of the photovoltaic power generation system. The goal of the long-term control model is to optimize the output of the photovoltaic inverter throughout the season or longer to maximize annual power generation and reduce maintenance costs.

[0122] Integrated control framework: Integrate short-term, medium-term and long-term control models into a unified control framework. The framework is responsible for coordinating the operation between the various models to ensure that control strategies of different time scales can be seamlessly switched during the operation of the photovoltaic power generation system.

[0123] Establish a model switching mechanism: Establish a model switching mechanism in the control framework to dynamically select the appropriate control model based on the real-time photovoltaic power generation system operation data and forecast information. When the prediction error of the short-term control model exceeds the preset threshold, the system will automatically switch to the medium-term control model; if the medium-term control model also fails to meet the requirements, it will further switch to the long-term control model. This mechanism ensures that the control strategy can be flexibly adjusted according to changes in the system operation status and external environmental conditions.

[0124] By constructing and integrating control models of different time scales and establishing an effective model switching mechanism, the present invention can significantly improve the stability and power generation efficiency of photovoltaic inverters under different operating conditions. This multi-time scale robust control strategy provides strong support for the intelligent operation and optimization of photovoltaic power generation systems.

[0125] Step S105, deploy the short-term control model, the medium-term control model, the long-term control model and the model switching mechanism to the actual photovoltaic power generation system, and compare and detect the data in the actual photovoltaic power generation system according to the expected target data of photovoltaic power generation to obtain the photovoltaic power generation output electric power comparison result. If the output electric power in the photovoltaic power generation output electric power comparison result does not meet the expected electric power, the optimization algorithm is used to optimize the short-term control model, the medium-term control model, the long-term control model and the model switching mechanism until the output electric power in the photovoltaic power generation output electric power comparison result meets the expected electric power. Stop the optimization.

[0126] In step S105, the present invention deploys the short-term control model, the medium-term control model, the long-term control model and the model switching mechanism constructed in the early stage to the actual photovoltaic power generation system, and performs real-time comparison detection and optimization to make the output power of the photovoltaic power generation system meet the expected target. The specific steps are as follows:

[0127] Integrate short-term control models, medium-term control models, long-term control models, and model switching mechanisms into the control system of the photovoltaic power generation system. This usually involves installing and configuring these models and mechanisms in the photovoltaic inverter or related control system hardware and software.

[0128] According to the expected target data of photovoltaic power generation (including expected power generation efficiency and expected power quality indicators), the real-time data in the actual photovoltaic power generation system is compared and tested. This includes monitoring the key parameters of the photovoltaic power generation system such as voltage, current, output power, etc., and comparing them with the expected targets.

[0129] Analyze the comparison results of the output power of photovoltaic power generation. If the output power does not meet the expected power (that is, there is a significant difference between the actual power generation and the expected power generation), further optimization is required.

[0130] Use optimization algorithms (such as gradient descent method, genetic algorithm, particle swarm optimization algorithm, etc.) to optimize the short-term control model, medium-term control model, long-term control model and model switching mechanism. The optimization goal is to minimize the error between the actual power generation and the expected power generation, and improve the stability and power generation efficiency of the photovoltaic power generation system.

[0131] Appropriate optimization algorithms are used for the short-term, medium-term and long-term control models. For example, for the error in short-term power generation forecast, the gradient descent method can be used to adjust the parameters of the short-term control model; for the error in medium-term power generation forecast, the genetic algorithm can be used to optimize the parameters of the medium-term control model; for the error in long-term power generation trend forecast, the particle swarm optimization algorithm can be used to adjust the parameters of the long-term control model.

[0132] The optimization process requires multiple iterations to gradually approach the optimal solution. In each iteration, the model parameters are adjusted according to the current optimization results and the performance of the photovoltaic power generation system is re-evaluated.

[0133] When the output power of the photovoltaic power generation system meets the expected power (i.e. the error is within an acceptable range), the optimization process is stopped. At this point, the control strategy and model parameters of the photovoltaic power generation system have reached the optimal or near-optimal state.

[0134] The present invention improves the effectiveness and reliability of the multi-time scale robust control algorithm of photovoltaic inverters in practical applications. Through real-time comparative detection and optimization adjustment, the photovoltaic power generation system can continuously maintain the best operating state and achieve efficient and stable power output.

[0135] Specifically, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S101:

[0136] The voltage data and current data of the photovoltaic power generation system are obtained by real-time measurement and recording through sensors and monitoring equipment installed in the photovoltaic power generation system;

[0137] Use a light sensor to measure the light intensity of the photovoltaic panel to obtain light intensity data;

[0138] Acquire temperature data by monitoring the temperature of photovoltaic panels and inverters through temperature sensors;

[0139] Obtain the status of the PV inverter through the monitoring system or communication interface inside the inverter;

[0140] The operating parameter data of the photovoltaic inverter is obtained by providing an output interface for the operating parameters.

[0141] Specifically, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S101:

[0142] Temperature data, wind data, and humidity data are obtained through the weather station data interface or environmental monitoring equipment;

[0143] The grid operation status is obtained through the communication interface with the grid operator or the public grid status information;

[0144] Grid load dataGrid operators provide real-time load data or obtain it through power monitoring systems;

[0145] The expected target data of photovoltaic power generation includes the expected output power of the photovoltaic power generation system;

[0146] The expected power quality indicators include the expected power quality indicators of the power grid.

[0147] Specifically, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S102:

[0148] Establishing a data preprocessing sequence for the acquired photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data, and photovoltaic power generation expected target data, preprocessing the photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data, and photovoltaic power generation expected target data according to the data preprocessing sequence to obtain the preprocessed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data, and photovoltaic power generation expected target data;

[0149] Random screening is performed on the pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid status data and photovoltaic power generation expected target data to obtain pre-processed data randomly screened data;

[0150] The data in the randomly screened data after preprocessing are compared with the acquired original photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and expected target data of photovoltaic power generation. If the data in the randomly screened data after preprocessing exists in the original photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and expected target data of photovoltaic power generation, the preprocessed data is processed successfully.

[0151] Specifically, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S102:

[0152] The pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data are stored in the blockchain, and a hash value corresponding to the pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data is generated;

[0153] Establish data retrieval and use rules. When receiving a data retrieval task command, parse the data retrieval task command to obtain data demand information. According to the data demand information, retrieve the pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data stored in the blockchain;

[0154] The retrieved pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data are parsed through hash values ​​to obtain usable pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data.

[0155] Specifically, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S103:

[0156] Extract features related to the multi-time-scale robust control algorithm of the photovoltaic inverter from the preprocessed data. The features related to the multi-time-scale robust control algorithm of the photovoltaic inverter include voltage features of the photovoltaic power generation system, current features of the photovoltaic power generation system, light intensity features of the photovoltaic power generation system, temperature features of the photovoltaic power generation system, operating state features of the photovoltaic inverter, operating parameter features of the photovoltaic inverter, temperature features of the natural environment, wind features of the natural environment, humidity features of the natural environment, operating state features of the power grid, and load features of the power grid;

[0157] Online learning algorithms include recursive least squares and online support vector machines;

[0158] The extracted features are analyzed using an online learning algorithm to identify the parameters of the multi-time-scale robust control algorithm for photovoltaic inverters and obtain the identification results of the online learning algorithm.

[0159] According to the recognition results of the online learning algorithm, the parameters of the multi-time-scale robust control algorithm of the photovoltaic inverter are updated in real time.

[0160] Specifically, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S103:

[0161] Analyze the adaptive adjustment strategy of the multi-time scale robust control algorithm of photovoltaic inverters, and obtain the basis for adjusting the adjustment parameters and adjustment targets;

[0162] Based on the operation data of photovoltaic power generation system, the model parameters of the multi-time scale robust control algorithm of photovoltaic inverter are identified by using parameter identification algorithm;

[0163] Determine the adjustment parameters and their adjustment direction according to the received adaptive adjustment strategy and the real-time identified operating status and model parameters;

[0164] Substituting the adjusted parameters into the multi-time-scale robust control algorithm model of the photovoltaic inverter, and updating the multi-time-scale robust control algorithm model of the photovoltaic inverter;

[0165] The updated multi-time-scale robust control algorithm model of photovoltaic inverter is verified and evaluated using real-time photovoltaic power generation system operation data. The adjusted parameters are corrected based on the verification and evaluation results.

[0166] Specifically, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S104:

[0167] Based on the photovoltaic power generation system feature set, a neural network model is used for training to obtain a short-term control model. The short-term control model is used to respond to changes in light intensity and output a photovoltaic inverter power generation control command at the current moment.

[0168] Based on the photovoltaic power generation system feature set, a neural network model is used for training to obtain a medium-term control model. The medium-term control model is used to respond to weather trend changes and output photovoltaic inverter power generation regulation control commands in the next few hours.

[0169] Based on the feature set of the photovoltaic power generation system, a neural network model is used for training to obtain a long-term control model. The long-term control model is used to respond to seasonal changes and output photovoltaic inverter power generation regulation control commands for long-term power generation.

[0170] Specifically, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S104:

[0171] The data output by the short-term control model, the medium-term control model and the long-term control model are compared with the real-time photovoltaic power generation system operation data to obtain the comparison results between the predicted data and the real-time data. The comparison results between the predicted data and the real-time data include the short-term prediction error, the medium-term prediction error and the long-term prediction error;

[0172] When the short-term prediction error exceeds the preset threshold, the medium-term control model is switched to process the data. If the medium-term prediction error exceeds the preset threshold, the long-term control model is switched to process the data.

[0173] Specifically, the photovoltaic inverter multi-time scale robust control algorithm optimization method provided by the present invention includes, in step S105:

[0174] According to the expected power generation efficiency and expected power quality indicators in the expected target data of photovoltaic power generation, an optimization target is set, and the optimization target is the expected value of the output power of photovoltaic power generation;

[0175] Compare the actual photovoltaic power generation output electric energy power with the expected value of photovoltaic power generation output electric energy power, and calculate the error between the actual power generation and the expected value of photovoltaic power generation output electric energy power. The error between the actual power generation and the expected value of photovoltaic power generation output electric energy power includes the total error and the errors of each time scale in the short term, medium term and long term.

[0176] The optimization algorithms include gradient descent, genetic algorithm, and particle swarm optimization. Configure the gradient descent method, set the initial parameters, learning rate, number of iterations, and population size of the gradient descent method, genetic algorithm, and particle swarm optimization.

[0177] Short-term control model: To address the short-term power generation forecast error, the gradient descent method is used to adjust the parameters of the short-term control model.

[0178] Medium-term control model: Based on the error of medium-term power generation forecast, the genetic algorithm is used to optimize the parameters of the medium-term control model;

[0179] Long-term control model: To address the errors in long-term power generation trend prediction, the particle swarm optimization algorithm is used to adjust the parameters of the long-term control model;

[0180] The optimization is stopped when the error between the actual power generation and the expected value of the photovoltaic power output electric energy power meets the expected electric energy power.

[0181] In step S105, the present invention performs a detailed optimization process for the short-term control model, medium-term control model, long-term control model and model switching mechanism actually deployed in the photovoltaic power generation system, so that the output power of the photovoltaic power generation system meets the expected target. The specific optimization steps are as follows:

[0182] According to the expected power generation efficiency and expected power quality indicators in the expected target data of photovoltaic power generation, the expected value of photovoltaic power generation output power is clarified and set as the optimization target of this step.

[0183] The actual photovoltaic power output power is monitored and recorded in real time, and compared with the set photovoltaic power output power expected value.

[0184] Calculate the error between actual power generation and expected power generation. This error not only includes the total error, but also needs to be broken down into errors in the short-term, medium-term, and long-term time scales in order to perform more targeted optimization.

[0185] Select and configure applicable optimization algorithms, including gradient descent, genetic algorithm, particle swarm optimization algorithm, etc. For each algorithm, key parameters such as initial parameters (such as initial weights, biases, etc.), learning rate, number of iterations, and population size need to be set to ensure the efficiency and effectiveness of the optimization process.

[0186] Short-term control model: In view of the error in short-term power generation forecast, the parameters of the short-term control model are adjusted using the gradient descent method. The gradient descent method calculates the error gradient and updates the parameters in the opposite direction of the gradient, gradually reducing the error and improving the prediction accuracy of the short-term control model.

[0187] Medium-term control model: Based on the error of medium-term power generation forecast, the genetic algorithm is used to optimize the parameters of the medium-term control model. The genetic algorithm searches for the optimal solution by simulating the biological evolution process (selection, crossover, mutation). It is suitable for dealing with complex problems with multiple local optimal solutions and can improve the adaptability of the medium-term control model in dealing with changes in weather trends.

[0188] Long-term control model: In view of the error in the long-term power generation trend forecast, the particle swarm optimization algorithm is used to adjust the parameters of the long-term control model. The particle swarm optimization algorithm searches for the optimal solution by simulating the foraging behavior of bird flocks. It has the advantages of fast convergence speed and easy implementation, which helps to improve the accuracy of the long-term control model in responding to seasonal changes.

[0189] In each iteration, the model parameters are adjusted according to the current optimization results and the performance of the photovoltaic power generation system is re-evaluated. Through the continuous iterative optimization process, the error between the actual power generation and the expected power generation is gradually reduced.

[0190] After each iteration, the error is recalculated and the model performance is evaluated until the optimization is stopped when the error between the actual power generation and the expected power generation meets the requirements of the expected electrical power.

[0191] When the error is reduced to an acceptable range (that is, the error between the actual power generation and the expected power generation meets the expected power), the optimization process ends. At this point, the short-term control model, the medium-term control model, the long-term control model, and the model switching mechanism have all reached the optimal or near-optimal state, and the photovoltaic power generation system can output power stably and efficiently.

[0192] The technical solution of the present invention aims to solve the problem that the model parameters of the multi-time scale robust control algorithm are often difficult to obtain accurately, and will change with the change of the operating state of the photovoltaic power generation system, resulting in reduced accuracy of photovoltaic power generation output power regulation, and is solved by the following steps:

[0193] The present invention firstly obtains various data required for the operation of the photovoltaic power generation system through multiple channels, including photovoltaic power generation system operation data (voltage, current, light intensity, temperature, etc.), photovoltaic inverter operation data (inverter status and operation parameters), natural environment data (temperature, wind force, humidity, etc.) and power grid status data (grid operation status, grid load data), as well as expected target data of photovoltaic power generation (expected power generation efficiency and expected power quality indicators).

[0194] The acquired data is preprocessed to remove noise and outliers, and feature extraction is performed to form a photovoltaic power generation system feature set. The feature set covers key information on photovoltaic power generation systems, photovoltaic inverters, natural environment, grid status, etc., providing important input for subsequent model training and parameter identification.

[0195] Use online learning algorithms (such as recursive least squares method, online support vector machine, etc.) to analyze the operation data of photovoltaic power generation system, identify and update the parameters of multi-time scale robust control algorithm of photovoltaic inverter in real time. This method can continuously adjust the parameters according to new data during the operation of the system, so that the model parameters can change dynamically with the change of system status, thereby maintaining a high parameter accuracy.

[0196] The adaptive adjustment strategy of the multi-time-scale robust control algorithm of the photovoltaic inverter is received, and the parameters in the algorithm are dynamically adjusted according to the real-time identified operating status and model parameters. This adaptive adjustment mechanism can ensure that the algorithm maintains optimal performance under different operating conditions and improve the regulation accuracy of photovoltaic power output power.

[0197] Based on the feature set of the photovoltaic power generation system, short-term, medium-term and long-term control models are constructed and integrated into the control framework. By establishing a model switching mechanism, control models of different time scales are automatically switched according to the prediction error to adapt to different changes in light intensity, weather trends and seasonal changes. At the same time, after actual deployment, the data in the system is compared and tested according to the expected target data of photovoltaic power generation, and the control model and switching mechanism are optimized using an optimization algorithm to ensure that the output power of photovoltaic power generation meets the expected target.

[0198] In summary, the present invention effectively solves the problem that the parameters of the multi-time-scale robust control algorithm are difficult to accurately obtain and change with the operating state of the system through the steps of comprehensive data acquisition, data preprocessing and feature extraction, online learning algorithm parameter identification and dynamic adjustment, adaptive adjustment strategy, and multi-time-scale control model construction and optimization, thereby improving the regulation accuracy of the photovoltaic power generation output power.

[0199] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations, and the above-described embodiments of the present invention do not constitute a limitation on the scope of protection of the present invention.

Claims

1. A photovoltaic inverter multi-time scale robust control algorithm optimization method, characterized in that: include: Step S101, obtaining photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid state data and photovoltaic power generation expected target data, the photovoltaic power generation system operation data includes photovoltaic power generation system voltage data, photovoltaic power generation system current data, light intensity, temperature data, photovoltaic inverter operation data includes photovoltaic inverter state and photovoltaic inverter operation parameter data, natural environment data includes temperature data, wind data and humidity data, grid state data includes grid operation state and grid load data, photovoltaic power generation expected target data includes expected power generation efficiency and expected power quality indicators; Step S102, preprocessing the acquired photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data and photovoltaic power generation expected target data to obtain preprocessed photovoltaic power generation system data, and extracting features from the preprocessed photovoltaic power generation data to obtain a photovoltaic power generation system feature set, the photovoltaic power generation system feature set including photovoltaic power generation system voltage features, photovoltaic power generation system current features, photovoltaic power generation system light intensity features, photovoltaic power generation system temperature features, photovoltaic inverter operation state features, photovoltaic inverter operation parameter features, natural environment temperature features, natural environment wind force features, natural environment humidity features, power grid operation state features and power grid load features; Step S103, using an online learning algorithm to analyze the operation data of the photovoltaic power generation system, identify the parameters of the multi-time-scale robust control algorithm of the photovoltaic inverter, update the parameters of the multi-time-scale robust control algorithm of the photovoltaic inverter in real time, receive the adaptive adjustment strategy of the multi-time-scale robust control algorithm of the photovoltaic inverter, and dynamically adjust the parameters of the multi-time-scale robust control algorithm of the photovoltaic inverter according to the real-time identification of the operation state of the multi-time-scale robust control algorithm of the photovoltaic inverter and the identification of the multi-time-scale robust control algorithm parameters of the photovoltaic inverter, so as to obtain the multi-time-scale robust control algorithm model of the photovoltaic inverter after the dynamic adjustment of the parameters; Step S104, constructing a short-term control model, a medium-term control model and a long-term control model based on the photovoltaic power generation system feature set, integrating the short-term control model, the medium-term control model and the long-term control model into the control framework, and establishing a model switching mechanism among the short-term control model, the medium-term control model and the long-term control model; Step S105, deploy the short-term control model, the medium-term control model, the long-term control model and the model switching mechanism to the actual photovoltaic power generation system, and compare and detect the data in the actual photovoltaic power generation system according to the expected target data of photovoltaic power generation to obtain the photovoltaic power generation output electric power comparison result. If the output electric power in the photovoltaic power generation output electric power comparison result does not meet the expected electric power, the optimization algorithm is used to optimize the short-term control model, the medium-term control model, the long-term control model and the model switching mechanism until the output electric power in the photovoltaic power generation output electric power comparison result meets the expected electric power. Stop the optimization.

2. The photovoltaic inverter multi-time scale robust control algorithm optimization method according to claim 1, characterized in that: In step S101, it includes: The voltage data and current data of the photovoltaic power generation system are obtained by real-time measurement and recording through sensors and monitoring equipment installed in the photovoltaic power generation system; Use a light sensor to measure the light intensity of the photovoltaic panel to obtain light intensity data; Acquire temperature data by monitoring the temperature of photovoltaic panels and inverters through temperature sensors; Obtain the status of the PV inverter through the monitoring system or communication interface inside the inverter; The operating parameter data of the photovoltaic inverter is obtained by providing an output interface for the operating parameters.

3. The photovoltaic inverter multi-time scale robust control algorithm optimization method according to claim 2, characterized in that: In step S101, it includes: Temperature data, wind data, and humidity data are obtained through the weather station data interface or environmental monitoring equipment; The grid operation status is obtained through the communication interface with the grid operator or the public grid status information; The expected target data of photovoltaic power generation includes the expected output power of the photovoltaic power generation system; The expected power quality indicators include the expected power quality indicators of the power grid.

4. The photovoltaic inverter multi-time scale robust control algorithm optimization method according to claim 1, characterized in that: In step S102, it includes: Establishing a data preprocessing sequence for the acquired photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data, and photovoltaic power generation expected target data, preprocessing the photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data, and photovoltaic power generation expected target data according to the data preprocessing sequence to obtain the preprocessed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid state data, and photovoltaic power generation expected target data; Random screening is performed on the pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, power grid status data and photovoltaic power generation expected target data to obtain pre-processed data randomly screened data; The randomly screened data after preprocessing are compared with the acquired original photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and expected target data of photovoltaic power generation. If the randomly screened data after preprocessing exist in the original photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and expected target data of photovoltaic power generation, the preprocessed data are processed successfully.

5. The photovoltaic inverter multi-time scale robust control algorithm optimization method according to claim 4, characterized in that: In step S102, it includes: The pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data are stored in the blockchain, and a hash value corresponding to the pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data is generated; Establish data retrieval and use rules. When receiving a data retrieval task command, parse the data retrieval task command to obtain data demand information. According to the data demand information, retrieve the pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data stored in the blockchain; The retrieved pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data are parsed through hash values ​​to obtain usable pre-processed photovoltaic power generation system operation data, photovoltaic inverter operation data, natural environment data, grid status data and photovoltaic power generation expected target data.

6. The photovoltaic inverter multi-time scale robust control algorithm optimization method according to claim 1, characterized in that: In step S103, it includes: Extracting the multi-time-scale robust control algorithm features of the photovoltaic inverter from the preprocessed data, the multi-time-scale robust control algorithm features of the photovoltaic inverter include the voltage features of the photovoltaic power generation system, the current features of the photovoltaic power generation system, the light intensity features of the photovoltaic power generation system, the temperature features of the photovoltaic power generation system, the operating state features of the photovoltaic inverter, the operating parameter features of the photovoltaic inverter, the temperature features of the natural environment, the wind features of the natural environment, the humidity features of the natural environment, the operating state features of the power grid, and the load features of the power grid; The extracted features are analyzed using an online learning algorithm to identify the parameters of the multi-time scale robust control algorithm of the photovoltaic inverter and obtain the identification results of the online learning algorithm. The online learning algorithm includes a recursive least squares method and an online support vector machine. According to the recognition results of the online learning algorithm, the parameters of the multi-time-scale robust control algorithm of the photovoltaic inverter are updated in real time.

7. The photovoltaic inverter multi-time scale robust control algorithm optimization method according to claim 6, characterized in that: In step S103, it includes: Based on the operation data of photovoltaic power generation system, the model parameters of the multi-time scale robust control algorithm of photovoltaic inverter are identified by using parameter identification algorithm; Determine the adjustment parameters and their adjustment direction according to the received adaptive adjustment strategy and the real-time identified operating status and model parameters; Substituting the adjusted parameters into the photovoltaic inverter multi-time scale robust control algorithm model, and updating the photovoltaic inverter multi-time scale robust control algorithm model; The updated multi-time-scale robust control algorithm model of photovoltaic inverter is verified and evaluated using real-time photovoltaic power generation system operation data. The adjusted parameters are corrected based on the verification and evaluation results.

8. The photovoltaic inverter multi-time scale robust control algorithm optimization method according to claim 1, characterized in that: In step S104, it includes: Based on the photovoltaic power generation system feature set, a neural network model is used for training to obtain a short-term control model. The short-term control model is used to respond to changes in light intensity and output a photovoltaic inverter power generation control command at the current moment. Based on the photovoltaic power generation system feature set, a neural network model is used for training to obtain a medium-term control model. The medium-term control model is used to respond to weather trend changes and output photovoltaic inverter power generation regulation control commands in the next few hours. Based on the feature set of the photovoltaic power generation system, a neural network model is used for training to obtain a long-term control model. The long-term control model is used to respond to seasonal changes and output photovoltaic inverter power generation regulation control commands for long-term power generation.

9. The photovoltaic inverter multi-time scale robust control algorithm optimization method according to claim 8, characterized in that: In step S104, it includes: The data output by the short-term control model, the medium-term control model and the long-term control model are compared with the real-time photovoltaic power generation system operation data to obtain the comparison results between the predicted data and the real-time data. The comparison results between the predicted data and the real-time data include the short-term prediction error, the medium-term prediction error and the long-term prediction error; When the short-term prediction error exceeds the preset threshold, the medium-term control model is switched to process the data. If the medium-term prediction error exceeds the preset threshold, the long-term control model is switched to process the data.

10. The photovoltaic inverter multi-time scale robust control algorithm optimization method according to claim 9, characterized in that: In step S105, it includes: According to the expected power generation efficiency and expected power quality indicators in the expected target data of photovoltaic power generation, an optimization target is set, and the optimization target is the expected value of the output power of photovoltaic power generation; Compare the actual photovoltaic power generation output electric energy power with the expected value of photovoltaic power generation output electric energy power, and calculate the error between the actual power generation and the expected value of photovoltaic power generation output electric energy power. The error between the actual power generation and the expected value of photovoltaic power generation output electric energy power includes the total error and the errors of each time scale in the short term, medium term and long term. The optimization algorithms include gradient descent, genetic algorithm, and particle swarm optimization algorithm. Configure the gradient descent method, set the initial parameters, learning rate, number of iterations, and population size of the gradient descent method, genetic algorithm, and particle swarm optimization algorithm. The short-term control model is used to adjust the parameters of the short-term control model using the gradient descent method for the error in short-term power generation prediction. The medium-term control model is used to correct the error in medium-term power generation forecast, and the genetic algorithm is used to optimize the parameters of the medium-term control model; The long-term control model is used to adjust the parameters of the long-term control model using the particle swarm optimization algorithm for the error in the long-term power generation trend prediction; The optimization is stopped when the error between the actual power generation and the expected value of the photovoltaic power output electric energy power meets the expected electric energy power.

Images