Improvement and Application of Simulated Annealing in Optimization of Thermal Control Systems of Power Plants

Simulate annealing is a global optimum algorithm with strong locally search ability, but its efficiency is the main obstacle to put into uses practically. In this paper, a return temperature strategy is added to enhance searching efficiency and a memory is used to ensure the quantity of solution. In the later stage, adding return temperature strategy to raise temperature and to enhance the accepting rate indirectly and the probability accepting the deterioration solutions can help to jump out the local optimization. In the meantime, a new initial solution is formed again in case of returning temperature and optimization proceeds as new searching direction. On the other hand, by means of adding a memory to keep the best result at this time, when simulate annealing process is end, the last solution compares with the one in memory, the better one is as the final solution, thus will enhance the quantity of solution. The practical application in thermal processes of power plants, such as the super-heater temperature and feed-water control systems, etc, shows that it is effective of the algorithm to raise optimization efficiency and reinforce practicality