Abstract:Currently power companies have more scattered statistics channels for power generation indicators, and the processes of power generation, power sale and power distribution have produced large variety and large amount of data. Therefore, in order to solve the problem of poor data quality and low value of mined data, this paper introduces the neural network method of big data prediction and the method of visual display, taking a certain power generation enterprise as the research object, and builds a model for power generation forecast and a data platform based on it. The application service of power generation forecast supports the PC and APP, covering all the power plants and new energy projects of the certain power generation enterprise. Iterative optimization of the algorithm model in a rolling prediction mode to improve the accuracy of power generation prediction, which is advantageous to power generation arranging, fuel procurement, material planning and further improving the prediction and control of pollutant emissions. The platform focuses on the realization of multi-dimensional data management and visual display of the user interface. At the data level, the platform performs data collection, cleaning and database real-time interaction from the source, and at the user level, the platform intuitively displays data acquisition, data analysis and preprocessing, data training and data prediction, which are the four stages of big data prediction.