Dynamic Optimization Approach With Swarm Intelligent Algorithms

Dynamic optimization, also known as optimal control, is of great importance to deal with the industrial bottlenecks for energy conservation, cost reduction, potentiality exploitation and efficiency improvement. It has been widely applied in many fields, such as biotechnology engineering, aerospace engineering, petrochemical engineering, etc. Dynamic optimization has also aroused wide attention of many well-known domestic and foreign scholars. Control vector parameterization (CVP) is a commonly used numerical method for solving dynamic optimization problems, where the control vector is approximated by a group of parametric functions through time discretization, thus the original infinite dimensional problem is transformed into a nonlinear programming (NLP) problem with limited parameter. Intelligent optimization approach is not only simple and easy to implement, but also has the advantages of strong global search ability and flexibility, which makes it be win great popularity and is as a kind of important optimization approach. This thesis studies intelligent optimization approaches and their applications in dynamic optimization problem on the basis of CVP method.The main work and contributions of this thesis are as follows:(1) A general framework of intelligent dynamic optimization approach is proposed, which transforms the original dynamic optimization problem into an NLP problem on the basis of CVP method, and then the intelligent optimization algorithm is utilized to get the solution and analysis.(2) The principles of three intelligent dynamic optimization approaches (bacteria foraging optimization (BFO), invasive weed optimization (IWO) and particle swarm optimization (PSO)) are proposed on the basis of CVP method. The approaches are used to solve classic dynamic optimization problems, and the results show their effectiveness in the applications.(3) To deal with the dynamic optimization problems with strong nonlinearity, an efficient hybrid intelligent optimization algorithm is proposed. The algorithm not only possesses the strong global search ability of adaptive particle swarm optimization (APSO) algorithm, but also has the powerful local exploitation ability of differential search (DS) algorithm. The testing results of classic dynamic optimization problems demonstrate the good optimization performance, precision and convergence speed of the proposed approach.(4) To deal with the dynamic optimization problems with state constraints, an iterative multi-objective particle swarm optimization-based control vector parameterization approach is proposed. Using the method of handling state constraints and control vector parameterization, respectively, the original problem will be transformed into a multi-objective nonlinear programming problem. The results of classic state constrained dynamic optimization problems show the good optimal capability of the proposed approach. Simultaneously, the comparison of algorithms’ performance and their analysis further illustrate the good convergence performance and diversity on the Pareto front of IMOPSO.