【摘要】 为获取减振器以某种概率产生某种程度异响时底层设计变量的取值区间,提出了基于广义逆推算法的减振器抗异响设计。以某型轻微异响(普通乘员可察)减振器为例,建立了对异响贡献度较高的7个设计参数与减振器杆端加速度有效值间的映射关系:条件受限玻尔兹曼机(Conditional Restricted Boltsmann Machines,CRBM)与深度信念网络(Deep Belief Network,DBN)相结合的CRBM-DBN近似模型,进而通过智能优化算法和区间搜索技术求解了设计参数的最优区间。对3支参数组合比较典型的的优化样件进行台架试验及整车主观感受,结果表明:优化样件的异响程度降低明显,进一步验证了减振器抗异响设计思路的有效性。更多还原

【Abstract】 In order to obtain a value range of design variables when a shock absorber produces a certain degree of abnormal noise with a certain probability,an anti-abnormal noise design of shock absorber based on generalized inverse algorithm is proposed. Taking a shock absorber with slight abnormal noise( visible to ordinary passenger)as an example,we establish the mapping relationship between seven parameters with high contribution to abnormal noise and an effective value of acceleration at rod end of shock absorber: a CRBM-DBN approximate model combined with conditional restricted Boltzmann machine and deep belief network. Then,the optimal interval of design parameters is solved by intelligent optimization algorithm and interval search technique. A bench test and subjective feeling of the whole vehicle are carried out on three optimized samples with three different typical parameter combinations. The results show that the abnormal noise degrees of the optimized samples decrease obviously,which further verifies effectiveness of anti-abnormal noise design idea of shock absorber.更多还原