A Weather-Based Hybrid Method for 1-Day Ahead Hourly Forecasting of PV Power Output Abstract To improve real-time control performance and reduce possible negative impacts of photovoltaic (PV) systems, an accurate forecasting of PV output is required, which is an important function in the operation of an energy management system (EMS) for distributed energy resources. In this paper, a weather-based hybrid method for 1-day ahead hourly forecasting of PV power output is presented. The proposed approach comprises classification, training, and forecasting stages. In the classification stage, the self-organizing map(SOM) and learning vector quantization (LVQ) networks are used to classify the collected historical data of PV power output. The training stage employs the support vector regression (SVR) to train the input/output data sets for temperature, probability of precipitation, and solar irradiance of defined similar hours. In the forecasting stage, the fuzzy inference method is used to select an adequate trained model for accurate forecast, according to the weather information collected from Taiwan Central Weather Bureau (TCWB). The proposed approach is applied to a practical PV power generation system. Numerical results show that the proposed approach achieves better prediction accuracy than the simple SVR and traditional ANN methods Existing system The existing method are 1) the indirect forecasting methods 2) direct forecasting methods. For the indirect methods, solar irradiance is predicted based on historical solar irradiance and weather data, and then is converted to PV power output. The techniques used include fuzzy logic method wavelet analysis, artificial neural network (ANN), and hybrid ANN-based methods. The direct methods predict PV power output according to their historical data and associated weather information. The techniques used consist of support vector regression (SVR) ,ANN, and hybrid ANN methods. These methods as mentioned above are useful for PV power output forecasting. However, the fuzzy logic-based method cannot learn directly from historical data. The limitations of wavelet analysis are the complex model structure and the high computational cost required. Moreover, ANN-based methods can be used for all the classification and forecasting problems, but require the user to specify various parameters of the model, especially those related to the network topology. As a result, their performance obtained is still not satisfied. Proposed system This paper presents an alternative approach combining a self organizing map (SOM) , a learning vector quantization (LVQ) network, the SVR method, and the fuzzy inference approach to make 1-day ahead hourly forecasting of PV power generation. The SOM transforms input patterns to a two-dimensional map of features in a topological-ordered fashion. With a simple structure and efficient learning, the LVQ can effectively classify the input vectors based on vector quantization. The SVR has a good learning ability and accurate prediction, even when there are few training samples used. The fuzzy inference method is used to select a proper SVR sub model for more accurate predictions. The proposed method comprises classification, training, and forecasting stages. In the classification stage, the SOM and LVQ networks are used to classify the collected historical input data. A number of SVR models are then used to learn the collected historical input/output data sets in the training stage. As the SVR models are trained properly, a fuzzy inference method is used to select an appropriate model among the trained SVR models to achieve more accurate forecasting. Advantage improve system reliability, maintain power quality, Increase the penetration level of the PV power generation system. The PV power generation has a close relationship with weather conditions, such as the temperature, solar irradiance, hourly solar angle, and the geographical location. The training stage employs the support vector regression (SVR) to train the input/output data sets for temperature, probability of precipitation, and solar irradiance of defined similar hours.To more accurately capture the weather conditions for every 3 h of the next day to predict the PV power output, the fuzzy inference method is used to select a proper SVR submodel in the forecasting stage based on all the available weather predictions, like the maximum temperature and probability of precipitation, The real PVpower generation data and irradiance are sampled every 5 min. The hourly PV power output and irradiance data are then obtained by averaging the data collected within 1 h. The temperature sensor is read analog signal and is converted through ADC The sampling times of temperature and probability of precipitation are all in the scale of 1 h. Since the solar irradiance is available from 6:00 to 19:00 in summer season and from 7:00 to 17:00 in non-summer seasons, the daily PV output data are collected only during these intervals. In the data collection process, the hourly PV power output data may be lost due to the recording errors or other special events. If abnormal data are collected, the training of SVR or ANN tends to be unstable. In this paper, the missing data are filled in using the data of same hours on the latest similar day.The lcd display the 13. Hourly foreeasting of PV power output Solar Pannel array + - Solar charge ctroller + DC load Power driven ckt battery A LCD Display Micro controller Power supply Temperature sensor status of PV power output.