Research On Applications Of Genetic Algorithm To Water Resources Engineering

Based on the knowledge of the running mechanism of genetic algorithm and the idea of crossover, which is quintessence of genetic algorithm, the basis principle, analysis technique and advance algorithm of genetic algorithm is systemically introduced into water resources engineering in this thesis. Some schemes of genetic algorithm applicable for the engineering problems of water resources are implemented, and many complex problems can be resolved by using the schemes. Solutions presented in this thesis can be summarized as follows: —Binary number coded accelerating genetic algorithm (AGA), real coded accelerating genetic algorithm (RAGA), and inter coded partheno-genetic algorithm (IPGA) have been implemented. —Large scale problems of resources optimal distribution can be resolved by using IPGA. —On the basis of AGA or RAGA, threshold auto-regressive model, threshold regressive model, bilinear time series model, logistic curve model, projection pursuit classification model, projection pursuit grade evaluation model, projection pursuit threshold auto- regressive model, and projection pursuit threshold regressive model can be established very conveniently. The above models have been successfully applied to predicting groundwater level, predicting marine temperature time series, predicting marine ice condition, predicting annual runoff, forecasting river flood, predicting drought and flood series in Jianghuai area of China, forecasting daily flow process, evaluating flood disaster loss, evaluating water quality, classifying historic flood samples, evaluating benefit of small watershed control and climatic division. —The problems of computing water surface curve in natural open channel, seepage computation of homogeneous earth dam based on the three-section method, computing critical water depth of open channel with trapezoid section, and computing the water depth of contraction section downstream overflow dam can be resolved by using AGA. The application examples show that AGA is much more convenient, high precision and general than each of the traditional methods. —The above results show that genetic algorithm can be widely applied to complex practical problems of water resources engineering, which is worth further advanced research.