Research In The Short-time Heavy Precipitation And Photovoltaic Power Generation Based On Deep Learning

Machine learning is a branch of artificial intelligence,and in many cases almost synonymous with artificial intelligence.Deep learning originated in the study of neural networks.In the 1960 s,inspired by neuroscience research on human brain structure,artificial neural networks were proposed to simulate human brain processing data in order to make the machine be similar to the process of human intelligence.In 2006,Hinton proposed Deep Belief Nets(DBNs)and its learning algorithm,which greatly accelerated the development of deep learning.Deep learning has been effectively applied in many fields until now and has great potential in growth.One of the key studies of meteorological practitioners is how to recognize and predict short-time heavy rainfall accurately and effectively.The short-time heavy rainfall is an important meteorological disaster that is mainly caused by strong convective weather,which is related to the physical parameters such as air humidity,moisture in the atmosphere,temperature and humidity.In this paper,a recognition model of the short-time heavy rainfall based on physical parameters and deep learning model DBNs has been constructed.Firstly,SMOTE algorithm is used to synthesize a few samples of the short-time heavy rainfall,which is much less than normal weather,to adjust the distribution of the original data set.Secondly,a deep learning model with a Gaussian Boltzmann machine,which is based on the observed data from automatic monitoring stations in a local ground and the physical quantities commonly used in weather forecast analysis,has been constructed.Thirdly,the automatic recognition model of short-term heavy rainfall has been achieved.Through the analysis of the experimental results,the model can accurately recognize the short-time heavy rainfall,and have a good performance on the POD,FAR and CSI of short-time heavy rainfall recognition.Improving the short-term power prediction of photovoltaic panels is the key issue,so that solar photovoltaic power generation can be effectively integrated into the current power grid,which is significant to improve the utilization of solar photovoltaic power generation.Restricted Boltzmann Machine(RBM)is a kind of autocoder which can be used to construct a deep learning model and can be used to reconstruct the input data.Based on the linear regression model,the Gaussian Restricted Boltzmann Machine(GBRBM)is constructed for the power prediction of photovoltaic power generation.The model firstly reconstructs the original data using the GBRBM,and then constructs a linear regression model on the reconstructed data.Finally,the model is applied to the output power data of the photovoltaic panel provided by GEFCom2014.Experiments show that the Boltzmann machine can greatly enhance the regression prediction ability of the general model for the regression of photovoltaic power.