Combined Prediction Of Urban Hourly Water Consumption Using LSSVM Based On Multi-dimension Embedding Phase Space

The urban water consumption prediction is the prerequisite and basis of the water supply system, playing a important role in the urban water supply system, and accurately predict the amount of water consumption can guarantee safety, energy-saving and efficient operation of urban water supply systems. However, the urban water consumption effected by the social production activities, geographical location and natural conditions and many other factors, is still difficult to find a more satisfactory method now. Existing forecasting models and methods have certain limitations and adaptability, the prediction accuracy is difficult to achieve the practical application requirements.The different predictive methods and parameters of phase space reconstruction have a great impact on the prediction accuracy of chaotic system. In order to improve the accuracy of chaotic hour consumption prediction in urban water supply system, a combined forecasting model for urban hourly water consumption using LSSVM based on multi-dimension embedding phase space was proposed. The phase space models with different embedding dimensions were established by combining mutual information method and G-P algorithm. Combined forecasting models were solved by LSSVM which can take advantage of all information in all dimension embeddings and forecast methods. The predictive bias under the single model was merged. In this way, the forecast accuracy was improved. The simulation results of hourly water consumption forecast in Xiaoshan shows that the forecast error was blow2%and better than other forecasting results in single model. This proved the effectiveness and practicability of the approach.