基于改进粒子群优化算法的污水处理过程模拟及故障诊断新方法研究

项目名称： 基于改进粒子群优化算法的污水处理过程模拟及故障诊断新方法研究

项目编号： No.51308083

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 建筑科学

项目作者： 张晶

作者单位： 大连大学

项目金额： 25万元

中文摘要： 随着全球范围的淡水资源日益枯竭和水体污染问题的日益严峻，污水处理受到当今世界的普遍关注。污水处理过程模拟可为污水处理过程的监控和故障诊断提供科学思维和决策，对于保障污水处理过程的正常运行，提高处理效率，降低能耗具有重要的意义和实际应用价值。本项目在前期研究工作基础上，拟采用改进的粒子群优化算法(PSO)优化径向基函数神经网络(RBFNN)和支持向量机(SVM)的模型参数，并将改进后的PSO-RBFNN和PSO-SVM算法应用于稳态和非稳态污水处理过程模拟，探寻较优的污水处理过程模拟的新方法；应用灵敏度分析、连接权值法和网络图形法可进行变量重要性和网络结构的解释，考察和分析影响出水水质的重要进水参数和过程控制参数，为污水处理的过程控制和故障诊断提供科学依据和技术支持。

中文关键词： 污水处理；过程模拟；粒子群优化算法；径向基函数神经网络；

英文摘要： With global freshwater resources gradually exhausted and water pollution more serious, wastewater treatment has been paid more attention. Process simulating of wastewater treatment process is significant and valuable for ensuring the normal operation, improving the efficiency, and reducing the energy consumption as it can provide scientific idea and decision for monitoring, controlling and fault diagnosing of wastewater treatment process. Based on previous study, this project will optimize parameters of radial basis function neural network (RBFNN) and support vector machine (SVM) using improved particles swarm optimization (PSO), forming two novel algorithms, PSO-RBFNN and PSO-SVM. The performances of these two algorithms will be compared on simulating wastewater treatment process under steady state or unsteady state, in order to find out a better method for process simulating. Then variable importance and network structure of the built models will be analyzing using the Sensitivity Analysis, the Connection Weights method and the Network Interpretation Diagram method. After the important influent parameters and control parameters influencing the effluent were identified, these could provide scientific basis and technical support for the process control and fault diagnosis of wastewater treatment.

英文关键词： Wastewater treatment；Process simulating；Particles swarm optimization；Radial Basis Function neural network；