Research And Application Of Power Prediction Of Run-of-river Small Hydropower Based On Big Data Technology

With global energy problem becoming more and more serious and the trend of our country to develop clean energy, the reasonable scheduling management of power generation of the run-of-river small hydropower is of great significance. Guizhou province is one of the provinces in our country which is rich in water resource, river, water rich, and has a number of small hydropower stations. Relying on the National Science and Technology Support Project, the cluster of small hydropower power prediction system was developed by North China Electric Power University and Guizhou Power Grid Corporation which is responsible for the province’s power prediction of river small hydro.However, as the number of small hydropower station continues to grow, the power data will also be multiplied, the original system based on traditional relational database will be difficult to meet the needs of efficient storage and management, due to its inherent system has a variety of problems such as poor scalability. In addition, the main of power prediction is string computing with high performance computer in the original system, existing data processing time is long, slow speed prediction, risk control and fault tolerance rate is low in the face of the huge amount of data, which brought great inconvenience to the power computation of small hydropower.In order to solve the above problems and realize the efficient storage, access and fast computation of the power data of the run-of-river small hydropower, firstly, this paper by considering the hierarchical relationships of power grid dispatching at each levels and analyzing the characteristics of the power data of the run-of-river small hydropower, has migrated the power data to the Hadoop big data platform. For the storage scheme, this paper puts forward a placement strategy of power data based on multi-measure evaluation. Combining with the actual production speed of power data, an appending data strategy is analyzed and designed.Secondly, this paper analyses the influence factors of small hydropower power data, and combining with the correlation characteristics of small hydropower power data and meteorological information, puts forward a power prediction model of run-of-river small hydropower based on the meteorological information, and the power prediction algorithm is designed for distributed computing.Finally, using the proposed scheme set up experiment simulation platform, the prediction efficiency was compared with the actual system, the simulation results show that the prediction efficiency of distributed prediction is more efficient than the original system, and verify the feasibility and effectiveness of the scheme.