基于改进粒子群置信规则库参数训练算法

Belief rule base parameter training approach based on improved particle swarm optimization

针对置信规则库中参数训练的求解问题,传统的粒子群优化算法易早熟收敛,陷入局部最优解,为更好地平衡算法的种群搜索和局部搜索能力,提出一种逐步减小惯性权重的粒子群优化算法,将其应用到置信规则库参数训练中。通过航材承修商的评估实例检验该算法的有效性,改进的粒子群算法收敛速度更快,精度更高;参数训练后的置信规则库的输出与专家评分拟合度相比未经过参数训练的置信规则库有明显提高。实验结果表明,改进粒子群算法可用于置信规则库参数训练。

For solving the problem of training parameters in the belief rule base,the premature convergence of traditional PSO causes falling into local optimal solution easily.To better balance the global search and local search capability of the algorithm,a gradual decrease inertia weight particle swarm optimization algorithm and its application were proposed in the rule base parameter training.The effectiveness of the algorithm was tested by evaluating examples of aircraft parts maintenance supplier.The im-proved particle swarm algorithm converges faster,more accurate,and output and expert ratings fitting degree of confidence of the rule base after training parameters are better than that before training.Experimental results show that the improved PSO al-gorithm proposed can be used in the belief rule base parameter training.