论文部分内容阅读
针对过程神经元网络模型学习参数较多,正交基展开后的BP算法计算复杂、不易收敛等问题,提出了一种基于双链结构的量子粒子群学习算法.该算法用量子比特构成染色体,对于给定过程神经元网络模型,按权值参数的个数确定量子染色体的基因数并完成种群编码,通过量子旋转门和量子非门完成个体的更新与变异.算法中每条染色体携带两条基因链,提高了获得最优解的概率,扩展了对解空间的遍历,从而加速过程神经元网络的优化进程.将经过量子粒子群算法训练的过程神经元网络应用于Mackey-Glass混沌时间序列和太阳黑子预测,仿真结果表明该学习算法不仅收敛速度快,而且寻优能力强.
Aiming at the problems that the learning parameters of process neural network model are more and BP algorithm after orthogonality expansion is more complex and difficult to converge, a double-strand structure-based quantum particle swarm learning algorithm is proposed. The algorithm uses quantum bits to form chromosomes, For a given process neural network model, according to the number of weight parameters to determine the number of quantum chromosomes and complete the population coding, through the quantum rotation gate and quantum non-door to complete the individual’s update and mutation .In the algorithm, each chromosome carries two Gene chain to improve the probability of obtaining the optimal solution and extend the traversal of the solution space to accelerate the optimization process of the process neuron network.The process neural network trained by quantum particle swarm optimization algorithm is applied to the Mackey-Glass chaotic time series And sunspot prediction, the simulation results show that the learning algorithm not only converges fast, but also has excellent searching ability.