基于CCPSO及PP模型的洪灾评估方法及其仿真应用

Flood Disaster Evaluating Method Based on CCPSO along with Projection Pursuit Model and Its Application

以求解洪水灾情评估问题为背景,针对洪灾评估模型参数难以优化这一问题,研究了一种混沌文化粒子群算法(CCPSO)。该算法将PSO纳入文化算法的框架,并在算法中引入局部遍历搜索性能较强的混沌搜索,组成基于PSO的群体空间以及基于混沌优化的信念空间,通过两个种群的独立演化及信息交流来提升算法的全局寻优能力。典型的测试函数的测试结果表明,CCPSO可以有效克服PSO存在的早熟收敛问题,全局收敛能力较PSO有较大提高。同时,为提高洪水灾情评估的灾情分辨率,提出一种基于CCPSO及投影寻踪模型的洪灾评估方法,该方法采用一种修正Logistic曲线来建立洪灾评估的投影寻踪模型,并使用CCPSO来优化投影指标函数以及模型参数。仿真应用结果验证了该方法的合理性及有效性。

Based on the ground of solving flood disaster evaluation problem, a chaotic cultural particle swam optimization algorithm （CCPSO） was proposed to overcome the difficulties of optimal parameter settings for the flood disaster evaluation model. The algorithm proposed inserts PSO into the framework of cultural algorithm model. Meanwhile, chaotic search which behaves well in local searching was adopted to buildup the PSO based population space and the chaotic optimization based belief Space. Through the independently evolution and the information exchange of two spaces, the global optimizing ability of the proposed algorithm was improved. A typical benchmark function was tested and the experimental result shows that CCPSO can avoid premature effectively and has better global convergence ability than that of PSO. Meanwhile, a new flood disaster evaluating method based on CCPSO along with projection pursuit model was proposed to enhance the grade resolution of flood disaster evaluation. A modified logistic curve was adopted to buildup the projection pursuit model for the flood disaster evaluation by the proposed method. Meanwhile, the projection direction and the value of project function of the model were optimized by CCPSO. The result of the simulation demonstrates the validity and the feasibility of the proposed method.