基于人工神经网络的参数灵敏度分析模型*

通过人工神经网络算法与参数灵敏度分析的结合，找到了一种新的工程系统功能模拟和变化分析方法。神经网络可以有效地解决复杂、非线性系统的功能模拟问题，其传递函数的可微性为参数灵敏度矩阵的求解提供了保证，从而方便寻找系统输入属性与输出属性之间的影响因子。同时，该模型具有良好的扩展性，可以更加全面地考虑系统影响因素。经实例仿真分析表明：该方法在工程分析方面，能够快速找到属性之间的关联程度，得到准确、稳定的分析结果，满足工程分析需求。

Parameter sensitivity analysis based on artificial neural network

studied new method of function simulation and analysis based on the combination of artificial neural network and parameter sensitivity analysis. Neural networks algorithm Could solve the function simulation problem of complex, nonlinear system effectively. The differentiability of the transfer function provided a guarantee to solution of sensitivity matrix, and made the searching for the impact between the input and output attributes. Meanwhile, the model had a good scalability, could be better to consider a comprehensive system factor. The conclusion shows that the method can find out correlation degree of attributes quickly and obtain a result ideally at the aspect of engineering analysis system.