钛合金动态本构模型参数敏感度分析及识别方法

为分析钛合金动态本构模型参数的敏感度，克服了传统单因素分析法的弊端，采用拉丁超立方抽样方法在整个参数空间中抽样，利用非参数统计方法中的Spearman秩相关分析法对本构参数随机输入样本集与其对应的目标函数输出结果集作相关性分析。建立了用Spearman秩相关系数等效求解参数敏感度的表达式，进而实现了参数敏感度的整体性分析，提高了参数敏感度分析结果的可靠性。基于参数敏感度分析结果，为减少本构模型参数识别的计算量，快速、精确、可靠地搜索最优解，本文建立了基于改进小生境算法、可疑峰值点判断策略和局域精确搜索技术的改进遗传算法。该方法克服了基本遗传算法存在的无法同时满足精度、可靠性和节省计算时间的缺陷，避免了容易产生早熟现象，增强了局部寻优能力。数值计算结果表明：本文提出的方法显著提高了遗传算法的全局搜索能力且计算精度高。本文提出的参数敏感度整体性分析方法和改进遗传算法不仅能很好地实现钛合金的动态本构参数识别，而且对其它工程材料本构参数的高精度识别具有重要参考价值。

Parameter Sensitivity Analysis and Identification Method for Dynamic Constitutive Relationship of Titanium Alloy

In order to analyze parameter sensitivity of dynamic constitutive relationship, the shortcoming of the single factor analysis method was overcome by using the Latin Hypercube Sampling Method in the entire parameter space, and, by use of nonparametric statistical method of spearman rank correlation analysis on constitutive parameters of the random input sample set and objective function of the corresponding output correlation analysis on the result set, the expression to solve the parameter sensitivity equivalently was set up in terms of Spearman Rank Correlation coefficient. The overall analysis of parameter sensitivity for dynamics relationship was realized. The reliability of the parameter sensitivity analysis results was improved, obviously. Based on the parameter sensitivity analysis results, in order to further reduce the computational burden of the constitutive model parameters identification and search the optimal solution rapidly, accurately and reliably, the improved genetic algorithm was established by means of improved niche genetic algorithm, suspicious peak point judgment strategy and Local accurate searching technology. This method overcomes that the basic genetic algorithm cannot satisfy the requirement of precision, reliability and saving computation time and avoids prone to premature phenomena and strengthens local optimization ability. The numerical calculation results show that this method has markedly improved the global search ability of genetic algorithm and has high calculation precision. In this paper, the parameter sensitivity integrity analysis method and improved genetic algorithm not only can well realize the parameters identification of the dynamic constitutive of titanium alloy, and the high precision recognition for other engineering material dynamic constitutive parameters has important reference value.