用粒子群算法求解打靶点的一种方法

针对打靶法中打靶点寻找困难的问题,提出了一种改进的粒子群算法.该方法将粒子适应度定义为终点状态的目标优化函数,并通过人工神经网络建立微分方程组的初始状态与终点状态的映射来提高运行速度.针对打靶点要求精度低,但不得陷入局部极值点的特点,在一次搜索结束后,逐步提高搜索水平,并重新调整粒子搜索能力来进行二次搜索,从而提高了得到全局最优值的命中率.最后通过一个实例验证了该方法的有效性.

Particle Swarm Optimization for Solving a Shooting Point

The particle swarm optimization is improved to overcome the difficulty in finding the shooting point of two-point boundary value problem.The key fitness function is defined as the optimization of end-point states and the running speed is improved by mapping the initial states on the end-point states in the whole process using an off-line trained neural network to form a differential equation set.A secondary search technique to upgrade searching ability is proposed to avoid the possibility of trapping search result in local extremum.After finishing the first search,all particles are renewed to enhance gradually their seeking ability in the secondary search so as to improve the hit rate of globally optimal values.An example is given to show the validity of the shooting method proposed.