Solar photovoltaic power generation, as an inexhaustible source of clean and environmentally-friendly energy, has become the focus of energy development in the future. This paper analyzes and summarizes the prediction methods of solar photovoltaic power generation, according to the application and demand of solar photovoltaic power generation. Summarizes the advantages and disadvantages of various solar photovoltaic power generation forecasting methods, and hopes to promote and promote the development of solar photovoltaic power generation forecasting methods in China.
Under the severe situation that oil production is increasingly bottoming out and the ecological environment is rapidly deteriorating, solar energy, as a natural energy source, shows its unique advantages with abundant reserves and clean pollution-free properties. It has been recognized internationally as the most competitive in the future. One of the energy sources. Solar photovoltaic power generation has become one of the main ways to use solar energy.
Photovoltaic power generation is divided into two forms: off-grid and grid-connected. With the maturation and development of photovoltaic grid-connected technologies, grid-connected photovoltaic power generation has become the mainstream trend. Due to the rapid increase in the capacity of large-scale centralized grid-connected photovoltaic power generation systems, the inherent intermittent and uncontrollable output power of grid-connected photovoltaic power generation systems has become an important factor restricting grid-connected photovoltaic power generation. The amount of solar photovoltaic power generation is affected by factors such as the amount of solar radiation, temperature, and solar panel performance. The magnitude of solar radiation directly affects the amount of power generation. The greater the radiation intensity, the greater the power generation and the greater the power.
Solar radiation is affected by factors such as seasons and geography. It has obvious characteristics of discontinuity and uncertainty, and it has significant annual, seasonal, and daily variations, and its physical and chemical conditions such as cloudiness, humidity, and atmosphere Transparency and aerosol concentrations also affect the strength of solar radiation.
The United States, Europe, Japan and other developed countries have conducted research and experiments on solar photovoltaic power generation forecasting methods earlier. China's solar photovoltaic power generation forecasting technology started late, and a few well-known universities have successively carried out technology research focusing on modeling and simulation. This paper analyzes and summarizes the prediction methods of solar photovoltaic power generation, summarizes the advantages and disadvantages of various prediction methods, and provides an important basis for the development of domestic solar photovoltaic power generation industry.
1 Solar photovoltaic power generation forecasting principle
At present, the research on solar photovoltaic power generation prediction mainly focuses on the prediction of solar radiation intensity. The daily or hourly observation data of solar radiation constitutes a very random time series, but there are still certain deterministic laws in the solar radiation sequence. Only by fully understanding the characteristics and changing laws of solar photovoltaic power generation can we establish the conformity. The actual situation of the prediction model and method.
Solar radiation is divided into direct solar radiation and scattered solar radiation. Direct solar radiation is the radiation of sunlight through the atmosphere to the ground; scattered solar radiation is radiation that reaches the ground after being absorbed, reflected, and scattered by dust, molecules, moisture, etc. in the atmosphere. The sum of scattered solar radiation and direct solar radiation is called total radiation. The influencing factors of total solar radiation intensity include: solar elevation angle, atmospheric quality, atmospheric transparency, altitude, latitude, slope gradient, and cloud cover.
Solar photovoltaic power generation prediction is based on the principle of solar radiation, historical weather data, photovoltaic power generation data, satellite cloud data, etc., using regression models, artificial neural networks, satellite remote sensing technology, numerical simulation methods to obtain prediction information, including the solar elevation angle, Atmospheric quality, atmospheric transparency, altitude, latitude, slope aspect, clouds and other elements, based on these elements to establish a solar radiation forecast model.
2 Solar photovoltaic power generation forecasting method analysis
Solar energy trends are mainly affected by local geographical conditions and meteorological conditions. The influence of geographical conditions has obvious rules. The annual solar trajectory can be calculated based on the local latitude and longitude, and an overall change trend of solar energy can be calculated by combining the parameters of the photovoltaic array itself. However, this trend does not reflect within a few hours or even reflect the general situation of solar energy changes within a few days.
The effect of weather conditions on solar radiation is the most direct. To realize the solar trend forecast within a few hours, it is necessary to find a calculation method for deriving the solar trend based on the meteorological conditions. In recent years, with the rapid development of the solar energy industry, the demand for solar photovoltaic power generation has been increasing. The developed countries have studied solar photovoltaic power generation forecasts earlier and have developed faster. At present, China's research on solar photovoltaic power generation forecasting technology is still in its infancy, and further research and experiments are needed.
There are three main types of solar radiation prediction methods:
The first category is based on the research of historical meteorological data and photovoltaic generation data, using statistical methods for analysis and modeling;
The second category: Based on satellite cloud data and ground monitoring data, satellite and radar image processing are used to calculate the forecast method of real-time solar radiation.
The third category: prediction methods based on numerical weather prediction.
The first type of forecasting method, the establishment of its model does not consider the physical process of solar radiation changes, through the analysis and processing of historical observation data, using the historical power generation to forecast the future generation. Generally, mathematical models such as regression model prediction and neural networks are used to establish a statistical model of the correlation between photovoltaic power generation systems and meteorological elements, and to predict power generation. The model construction and operation method are relatively simple, but it is only suitable for a stable time series with little change in power generation. For large time series with large changes in power generation, the error is large.
2.1.1 Regression model prediction:
Regression model prediction Based on historical data, find out the relationship between weather changes and solar radiation and its changing rules, establish mathematical models that can be used for mathematical analysis, and predict future solar radiation. The method is characterized by using the factors of the predicted target as variables and the predicted targets as constants. Using a given set of variables and constants, study the relationships among variables. Using the regression equation obtained to represent the relative relationship between variables and constants, so as to achieve the purpose of predicting solar radiation. In a large number of experiments and practices, the variable error is large, especially at noon.
The regression model predicts that solar radiation data predictions for non-linear time series are not ideal. The artificial neural network method has less prediction error than the regression model.
2.1.2 Artificial Neural Network
The artificial neural network method uses neural network technology to establish a function model for generating electricity, total solar radiation, and plate temperature. The combination of historical data is better. At present, the most studied is the application of error back propagation algorithm (BP algorithm) for short-term expectations. The main idea of this algorithm is to input historical data and several types of factors that affect the solar radiation as input into the artificial neural network, and then generate various kinds of data in the input layer, hidden layer, and output layer to generate output; The error is an objective function that repeatedly corrects and refines the weight of the artificial neural network until it reaches the set error value.
When traditional statistics cannot meet the requirements, artificial neural networks can be used for forecasting methods. However, this method is also based on historical meteorological data. Forecasting of power generation depends heavily on the accuracy of the total solar radiation forecast: Failure to find the key to affecting photovoltaic power generation Hourly weather elements make it difficult to control sudden and random weather changes.
2.2 The second type of prediction method
The second type of forecasting method mainly uses satellite remote sensing technology to complete the prediction of solar radiation. Satellite remote sensing refers to an observational activity using an artificial satellite as a sensor platform. It is achieved by surveying the electromagnetic radiation emitted or reflected by the Earth's atmospheric system. It includes earth observation and observation activities facing the space environment. Earth observation is the main content of satellite remote sensing. High spatial resolution image data and geographic information systems are closely integrated, providing a high basis for solar radiation prediction.
In 1960, the first Terros satellite transmitted the first visible light cloud image to the earth, enabling people to see the great potential of using satellite remote sensing. From then on, with the gradual improvement of meteorological satellite technology, various satellites, such as the remote sensing of the earth’s atmosphere, the terrestrial and oceanographic features of the Earth’s surface, and the monitoring of the Earth’s environment, have gradually emerged.
The satellite remote sensing technology of the United States has always been a world leader and represents the development level of satellite remote sensing technology. Countries such as Europe, Canada, and Japan are all developing and researching remote sensing technology. China’s first geosynchronous weather satellite “Fengyun 2” was launched on June 10, 1997, marking China’s satellite remote sensing technology stepping to a new level.
After a great deal of research and practice, it shows that the hourly ground radiation data obtained by satellite remote sensing technology and the ground observation radiation data have large deviations, and the maximum error can reach 20%-25% of the root mean square error. Therefore, how to make smaller errors, accurate statistics, and predictions become the development direction of remote sensing technology.
2.3 The third forecast method
The third type of forecasting method mainly uses numerical simulation methods to predict, that is, using the mathematical physics model to analyze the atmospheric conditions and use high-speed computer to solve the forecasting method. The method establishes a system of equations based on the principles of dynamics and thermodynamics that describe the laws of atmospheric motion. After determining the initial state of the atmosphere at a certain moment, it can be solved by mathematical methods to calculate the state of the atmosphere at a certain time, which is commonly known as the weather situation. And related meteorological elements such as temperature, wind, precipitation, irradiance and so on. The numerical simulation prediction method predicts a long time. At present, it can predict data of 40 hours or even longer.
The meteorological and environmental factors in the numerical simulation method are the most complex and difficult to determine precisely. Therefore, the error of the forecast does not only exist, but the accuracy is greatly reduced for the short-term and particularly complex changes. Therefore, the improvement of accuracy has always been the focus and difficulty of current research.
3 Conclusions and Prospects
This article has read a large number of domestic and foreign solar photovoltaic power generation forecasting method literature, based on extensive research, based on a more comprehensive discussion of the research status and development direction of solar photovoltaic power generation forecasting technology, a more detailed summary and analysis of the three types of forecasting methods. Explain the advantages and disadvantages of various methods. How to continue to improve, continuously improve, and explore on the basis of existing scientific research achievements, identify key factors that affect solar radiation, accurately predict and form a multi-level, multi-information integrated forecast system, is the main research of China's solar photovoltaic power generation forecast direction.
Related Industry Knowledge
- CSP is a viable option to make up f...
- Thermal storage CSP systems are cru...
- Introduction of portable refrigerator
- Portable refrigerator principle
- Precautions for use in portable ref...
- Fluorine-free refrigerator need to ...
- What is the use of outdoor mini ref...
- Folding solar charger
- Types of photovoltaic panels
- Household photovoltaic how to make ...
- Research on solar thermal power gen...
- Principle and Application of Invert...
- Solar photovoltaic cell array insta...
- Summary analysis of common major qu...
- Solar Panel Solar Panel Inspection ...
- Basic knowledge of photovoltaic
- What is MPPT (Solar Charge Controller)
- How To Ensure Solar Inverter Effici...
- Interpretation of the principle and...
- Explore the application prospects o...