This paper mainly studies the variable weighting combination forecast model of the weight coefficient, the weight of the recognition problem is directly related to the accuracy of prediction, the issue is the predicted difficulties and key. For a com-plicated target function such as target function is not micro, variable more resulted numerical calculation is very difficult. According to the characteristics of genetic algorithm, we use genetic algorithm for solving this complex problem is presented, and the flow chart of genetic algorithm. In addition we will this model applied to the problem of urban water supply load forecasting, made the corresponding nu-merical experiments. Numerical experiments show the effectiveness of the method of genetic algorithm. |