Study On The Ultra Short Term Power Prediction System Based On The Cloud Image In The Tower Thermal Power Generation

As one of the most important factors that affect the amount of radiation in the power thermal generation, the generation and dissipation and motion of the cloud will affect the stability of the power output. Therefore, it is very important to predict the movement trend of the cloud in the mirror field. For the medium and long term changes in the amount of radiation,the energy storage device can be adjusted accordingly. However, the mutation of the amount of radiation will bring great interference to the system, in order to avoid this effect, the temperature loss of the receiver is compensated by adjusting the inlet flow rate of the cryogenic fluid through the temperature control system. However, there is a pure lag in the control system,and the reaction time is needed when the cloud is blocking the sun, and the temperature stability of the high temperature fluid can not be guaranteed. Therefore, the study of the prediction system can provide a feed-forward signal for the receiver and temperature control system.Through the ultra short term prediction results to overcome the control system of the existence of pure lag (minute level), when the occlusion occurs only after a small fluctuations, it will be able to restore the temperature of the receiver outlet fluid temperature. Different from the medium and long term prediction based on historical meteorological data and satellite cloud images, this study is based on the ground cloud picture. Capture the target image by tracking the camera system, on the basis of the combination of computer vision technology to the observation and analysis of the clouds, the method has good real-time and high accuracy, can satisfy the minute level of ultra short term irradiation power demand forecasting.In this paper,the system initialization, lens distortion correction, cloud detection, cloud matching, cloud prediction five aspects of the system research and verification.The specific research work of this paper includes:(1)The tracking system is composed of a tracking bracket, a CCD camera and a wide-angle lens, and the sun centered image is obtained, blocking the sun with the shading plate. By the RS485 communication protocol, the data interaction between the tracking support and the PC terminal is realized.Automatic exposure control is achieved by adjusting the camera exposure time and signal gain;(2)A calibration model of barrel distortion is determined, the correction of the barrel distortion of wide-angle lens is completed by the correction of the polynomial address and the concentric circle template;(3)In this paper, we propose a clustering and classification based cloud detection idea,and propose a cloud detection algorithm based on color feature and K-Means clustering,and the results were evaluated.Compared with the gray threshold segmentation method, the detection effect has been greatly improved;(4)The exact matching results are obtained by using the SIFT algorithm and the error matching elimination method,On the basis of this, a new method is proposed to match the time span of more than 1 minutes it can be obtained more matching points.The matching points and contour points of all the effective cloud are calculated by the method of cloud slice tracking;(5)According to the results of cloud slice tracking, a hierarchical prediction model combined with particle filter is proposed to predict the cloud movement within 4 minutes.According to the information of cloud amount, the variation of solar radiation is observed.