Improved Genetic Algorithm For Power Systems Reactive Power Optimization And Its Program Development

It's the fundamental principle of Genetic algorithms (GA) that the fittest member of population has the highest probability for survival. In GA, which uses multiple-point search instead of single-point search and works with the coded structure of variables instead of the variables themselves, the only information required for searching its solution is the objective function and it makes the method simple and powerful in solving the mixed-integer nonlinear optimization problems. Moreover, GA has no special demand for its start point and is good at detecting and handling the infeasible solutions.Reactive power optimization (RPO) has received worldwide interest for its significant influence on secure and economic operation of power systems. The objective of RPO is to allocate reactive power generation so that the active power transmission loss is minimized, while satisfying numbers of operation constraints.However, as one of the most complicated mix-integer nonlinear problems with the characteristics of multi-objective, multi-uncertainty, multi-restriction, multi -extremum and discreteness property, the reactive power optimization problem is not yet solved completely. Since GA could search for a global solution using multiple path and handle discrete problems naturally, it has been developed for solving RPO problem.The main deficiency of application of GA in RPO problem is its premature convergence especially when the system becomes very large. On the basis of surveying the various algorithms for the RPO problems, this thesis has firstly proposed a mathematical model for real-time dynamic RPO, then presented an improved GA aiming at the characteristics of RPO problem, and lastly developed a set of application software for RPO. The research contents can be outlined as bellow:(1) It has analyzed the basic principle and characteristics of GA and investigated the GA coding strategy and its operators in detail.(2) It has proposed a new mathematical model for real-time dynamic RPO by analyzing kinds of the current mathematical models for RPO problems. The proposed model can decrease the computation complexity of RPO problem, and avoid over frequent operation of control variables.(3) It has presented an improved GA based on decimal integer coding for solving...