【摘要】 针对标准量子遗传-神经网络在进行缺陷预测时存在的收敛速度慢、易陷入局部最优等问题,提出了基于改进旋转策略的量子遗传-神经网络算法的预测模型.首先,以标准BP(back propagation)神经网络为基础,确定其拓扑结构.接着,采用动态更新旋转角度的策略对旋转角度进行更新.然后,利用改进的算法对神经网络进行优化,并构建预测模型,以提高预测的准确率.最后,在NASA数据集上进行仿真实验,结果表明改进后的预测模型的准确率更高,其平均值达到了90%.更多还原

【Abstract】 Standard quantum genetic-neural network has slow convergence speed and is easy to fall into local optimum in defect prediction. A prediction model of quantum genetic-neural network algorithm based on improved rotation strategy is proposed to tackle this problem. Firstly, the topological structure is established based on the standard back propagation(BP) neural network. Secondly, the rotation angle is updated by the strategy of dynamic updating rotation angle. Furthermore, the prediction model is constructed, and the neural network is optimized by using the improved algorithm to improve the accuracy of prediction. Finally, the simulation experiment on the NASA data sets shows that the accuracy of the improved prediction model is higher, and the average value reached 90%.更多还原