基于神经网络的桑巴塔回转架参数灵敏度分析

Parameter Sensitivity Analysis of Rotary Frame of the Samba Tower Based on Aartificial Neural Network

回转架的设计在大型游乐设备--桑巴塔的研发过程中占有重要地位。其动态特性不仅关系到产品的可靠性,也影响着游客乘坐的体验感,因此,设计过程中研究回转架的动态特性十分必要。通过ANSYS谐响应分析获得了回转架与外伸臂链接铰点的横向位移幅频响应曲线。再通过幅频响应曲线得出对塔架动态特性影响最大的固有频率,以该频率及塔架von mises应力、切应力和节点位移作为指标,探索塔架设计的11个变量对这些设计指标的影响。为使运算分析效率较高,运用MATLAB软件构建了11个变量与4个设计指标的BP神经网络模型。该方法集成运用ISIGHT和MATLAB软件实现了蒙特卡洛(Monte Carlo)数值模拟的参数灵敏度分析,明确了各参数的灵敏度差异,这将为同类结构设计提供重要的借鉴方法。

The design of rotary frame plays an important role in the research process of the samba tower. The dynamic characteristic of rotary frame is not only related to the reliability of the product, but also related to the experience of the tourists, so it is very necessary to study the dynamic characteristics of the rotary frame.Through harmonic response analysis of ANSYS, it obtains lateral displacement amplitude frequency response curves of hinge points,which between the rotary frame and overhanging arms. Utilizing the amplitude frequency response curve to obtain the natural frequency,which have the greatest influence on the dynamic characteristics of the rotary frame.Using this frequency,von mises stress, shear stress and nodal displacement as design indexs to investigate the effects of 11 variables on those design indexes. In order to reduce the time cost, the BP neural network model with 11 variables and 4 design indexes is constructed by using MATLAB. This method uses the integration of ISIGHT software and MATLAB to realize the Monte Carlo numerical simulation for analyzing the parameter sensitivity. By this analysis the differences in sensitivity of parameters are clarified. The study provides a referential method for the design of similar structures.