基于神经网络的弹性连杆机构动力响应分析

Dynamic analysis of the elastic linkage mechanism based on the artificial neural network

针对求解弹性连杆机构动力响应过程的复杂性和耗时性,提出了基于人工神经网络中的向后传播网络的机构动力响应分析模型.利用训练后网络所给出的近似函数关系式,推导出网络输出参数对网络输入参数的灵敏度计算公式,将训练后的网络以及求解灵敏度的方法应用于弹性连杆机构的优化设计和可靠性分析中,可减少动力响应分析时间,提高设计效率.

In accordance with the complexity and time-consuming character of the dynamic analysis of the elastic linkage mechanism, a dynamic analysis model is established based on the BP （Backpropagation） Network. By using the approximate functional relationship between the input parameters and the output parameters given by the trained network, the formula for the sensitivity of the network output parameters is derived. Application of the trained network and the formula to the optimization design and the reliability analysis of elastic linkage mechanism decreases the time for the dynamic analysis obviously and raises the design efficiency effectively.