Applications Of Support Vector Machine In Water Conservancy And Electric Power

Hydropower installed capacity had expended 100,000 MW since 2004, accounted a quarter of the whole national power installed capacity, provided one fifth of the countrywide electric power requirements. In recent years, China's power consumption has being grown rapidly, and many parts of the nation are short for fossil power supplying; while China is affluent in hydropower resources, it's theoretical potential of hydropower resources is 694,000MW, technical exploitable capacity is 542,000MW, economic exploitable capacity is 402,000MW and economic exploitable annual power output is 1,750 billion KW·h-1. In order to alleviate the contradiction of electricity supply and demand, the nation has increased the investment to the water & electricity investments. The general policy of Chinese power system development which is"power transmission from west to east, mutual supply between south and north, nationwide interconnection", offers new opportunity to China's hydro-energy production. While in the course of hydro-energy production, there are many problems need to be studied, and for the drawbacks of traditional algorithms. This paper focuses on the application of Support Vector Machine (SVM) in correlation area of hydroelectric power generation. The main contains of the paper are:Firstly, the paper describes the necessity, availability, and developing tendency of the application of Support Vector Machine in runner forecasting, vibration pattern recognition and characteristic fitting of hydraulic electric generating sets;Secondly, the paper focuses on the basic theories of Support Vector Machine, Cellular Automaton (CA), and so on;Thirdly, taking the average flow from 1959 to 1990 by months of Tianqiaoduan hydraulic station upstream of Zhang River Handan as original data , the paper builds the helical cellular automata model of the runoff forecast based on the basic theory of cellular automaton, and uses Support Vector Regression to do the runner forecast;Fourthly, the paper analyzes the production mechanism and formation characteristics of the pressure pulsation in turbine draft-tubes and use Support Vector Machine to analyze the pressure pulsation conditions in clusters, based on abstracting the vibration characteristics of draft tube in wavelet packet method;Fifthly, because the corresponding curve which expressed the turbine runner cavitations and pressure fluctuation characteristics are too complex to be analyzed, the paper uses support vector machine to establish the model which can express cavitations and pressure fluctuation characteristics at the same time and applies it to the establishment of the two output model of hydraulic turbine cavitations and pressure fluctuation characteristics in Zipingpu hydro-plant Sichuan province.This paper focuses on withdrawing the actual problem correlation data of water conservancy and hydroelectricity engineering, building the model based on the coherent theory, and then using support vector machine to fit or analyze the model. The main achievements are as following:Firstly, the paper builds the helical cellular automata model;Secondly, the paper uses Support vector machine to do the runoff forecast;Thirdly, the paper creates the two output model of cavitations and pressure fluctuation characteristics of hydraulic turbine.Fourthly, the paper establishes the pressure pulsation conditions cluster analysis model.