One Class Of Control And Layout Problems In Evolutionary Algorithm And Application

The 3D-trajectory of the horizontal well and the packing problem of the artificial satellite module belong to the nonlinear multistage optimal control and the layout optimization with performance constraints respectively. So it is valuable to study numerical optimization algorithms of the problems in both theory and application. The main results in the paper may be summarized as follows:1. An improved evolution strategy (MOC( - )-ES) is constructed in combination with the characteristics of the nonlinear multistage optimal control. The algorithm greatly cuts down the running time. Then the MMOC( + )-ES is constructed to solve the multimodal optimal control by the definition of the similarity.2. Based on the non-overlap theory, we provide the non-overlap algorithm to solve the non-overlap constraint, which can compute the overlap value. An improved genetic algorithm (PLGA) is constructed to solve the layout problem with performance constraints. In this algorithm, the layout projects are unnecessary to be feasible while iterating. We only need give a penalty value to the unfeasible project.3. The control technology and optimal design technology are the core of drilling the horizontal well. If we apply the optimal control into the control of the trajectory of the horizontal well and construct a satisfactory algorithm , a great breakthrough will be made in drilling the horizontal well. In this paper, we discuss the controllability of optimal control system of the 3D design of the horizontal well trajectory . Combining the characteristics of the system, we apply MOC( + )-ES to solve it. The numerical results show that the system is mutimodal. Then we apply the MMOC( + )-ES to solve the dynamic system and provide the design projects.4. The layout of the satellite module is the layout optimization with performance constraints. It is a topological optimization, and NP-hard. We apply PLGA constructed into it and compute them when the number of the rectangular elements is five, six. The satisfactory layout projects are obtained. At last we compare the results with those obtained by another existed algorithm.