| Swarm intelligence optimization is an emerging optimization method. It simulates the behaviors of social animals and uses the information exchange and cooperation between the individuals of swarm to achieve the optimization objective. For the past few years, it has been paid the most of attention in optimization research field. Compared with traditional method, it can't be only implemented simply but also get better optimization result.Moreover,an effective method is provided to solve the NP problem which is hard for traditional method. This paper mainly studies the calculation model of swarm intelligence optimization. How to solve continuous optimization problem, system identification problem of thermodynamic system, controller design and parameter optimization problem, the economic load allocation problem is studied,around the theory and application of two typical swarm intelligent optimization algorithms——the particle swarm optimization and the ant colony optimization.In this paper, two formalization models of optimization algorithm are described based on the aggregate pattern and framework model of swarm intelligent optimization, combining the basic principle and the algorithm process of particle swarm algorithm and ant colony algorithm.It includes the definition needed by stereotyped description, stereotyped algorithm structure , stereotyped algorithm false equation and algorithm diagram.Meanwhile,an improved scheme of particle swarm algorithm is put forward, joined with the dynamic variable interval and restart strategy. The improved algorithm is used to optimize the typical function. During the identification of complex object of thermal system,based on neural network,the precision and identification efficiency are improved as the result of optimizing its best delay time with the improved algorithm. In the control system of main steam temperature, the dual mode control with the fuzzy control and PID combined is applied. The improved particle swarm algorithm is used to optimize the controller parameters to ensure that there isn't static error with the regulating time shortened, overshoots reduced and robustness and anti-jamming of system improved.Two successive ant colony algorithm models are presented and applied respectively to optimize the parameters of PID controller in thermal system and distribute the economic load in electric system. Two kinds of models correspond with two distribution schemes of successive ant colony optimization space. While for the imperfections of basic ant colony algorithm, including the slow convergence speed and low convergence rate,which are frequently encountered,models are joined by the local search and mutative scale chaos search mechanisms.Based on the characteristics of swarm intelligent optimization, optimizing in whole situation and converging fast, the iterative process of FCM algorithm based on gradient descent, is insteaded by the swarm intelligent optimization,fusing the improved particle swarm algorithm,successive ant colony algorithm and fuzzy c-means clustering (FCM) algorithm.It enhances the search ability of the algorithm in whole situation and avoids falling into its local extremum. The simulation results show that the clustering effect of the two fusion algorithms are better than the basic FCM algorithm.The major innovation points of the paper are as follows:1. One kind of improved particle swarm optimization algorithm is put forward. Optimization space is compressed by subswarm composed of excellent particles and restart strategy is introduced.2. Two successive ant colony algorithms are put forward, corresponding with two distribution schemes of optimization space. Models are joined by the local search and mutative scale chaos search mechanisms.3. Put forward the integration scheme of particle swarm intelligent algorithm, clustering ant colony intelligent algorithm and fuzzy c-means algorithm.The iterative process of FCM algorithm based on gradient descent is insteaded by the swarm intelligent optimization.
|
PDF Full Text Request