Research On Complex Cloud Cluster’s Models For The Ultra-Short-Term Distributed Photovoltaic Power Prediction

Energy consumption brings many problems to the environment. Solar energy is huge. The process of photovoltaic power generation won’t produce any pollution to the environment. As a result, photovoltaic power generation will be widely used as the main energy utilization form of solar energy.However, the output power of photovoltaic system has stochastic volatility. This problem makes an influence on the safe operation of power grid. The cloud’s characteristics which make an influence on the output power of PV are taken as the research objective in this thesis, the method of the determination of target cloud is analyzed in details. This thesis focuses on the study of the cloud’s breadth characteristics. The main works as follows:1. Set up a hardware experiment platform for gathering cloud pictures with calculating the sun’s movement principle in Hangzhou, including the design of mechanical structure and control system.2. Analyses the clouds’characteristics. Tracking the target cloud according to cloud center mass point and the velocity vector, at the same time the key characteristics of those target clouds which can cause short-term output power’s fluctuation of distributed photovoltaic system are also analyzed.3. Take cloud image preprocessing with the mathematical morphology. Calculate and analysis of the dispersion characteristics of cloud and characteristics of local cloud cover. Those characteristic parameters including other cloud data will be taken as the important input parameters of prediction module of photovoltaic system, determining whether the output prediction module work or not.4. Illustrates the necessity of the convex decomposition of complex clouds concave polygon, according to the cloud breadth evolution model. First, demonstrate the cloud outline with the smallest internal polygon, judging the concave points with the positive and negative of vector product. Centroid point scattering model is set up, and put forward the rules of the convex decomposition of concave cloud contour, making the convex decomposition of complex cloud concave, making it available for cloud breadth characteristic evolution model. Finally the short-term output prediction framework of distributed photovoltaic system is built.A hardware experiment platform for gathering cloud pictures is established in this paper, which can collect cloud images, analysis and storage the characteristic automatically. Study the target cloud’s growth and extinction mechanism in its life cycle and the commonness and personality characteristics of cloud clusters. Develop the target cloud and key characteristics’ identification and extraction algorithm through MATLAB, analyzing the cloud cluster’s key characteristics, the complex cloud concave polygon contour is convex decomposed. The results of the research will provide the important input parameters and theoretical support of prediction module of photovoltaic system.