基于集成RSM和NSGA-Ⅱ的SLM成形GH3625高温合金尺寸精度优化研究

Optimization study of dimensional accuracy of SLM forming GH3625 superalloy based on integrated RSM and NSGA-II

摘要:

选区激光熔化(SLM)是目前应用最广的金属材料增材制造技术之一，由于成形过程中存在复杂的热物理过程，导致其成形精度仍无法满足实际应用的要求。因此，为了提高SLM成形零件的尺寸精度，本研究提出了一种集成响应面法(RSM)和非支配排序遗传算法-Ⅱ(NSGA-Ⅱ)的方法来优化SLM成形GH3265高温合金的尺寸精度。首先采用响应面法建立了SLM工艺参数与X、Y和Z方向的尺寸误差模型，然后通过NSGA-Ⅱ对模型进行多目标优化。结果表明：响应面法构建的模型具有较高的预测精度，其相关系数R2依次为0.9456、0.9842、0.9704；优化算法在迭代1500次时，能够获得最优的加工参数区间：激光功率250.8-310 W、扫描速度1028-1400 mm/s、扫描间距0.071-0.084 mm；最终，试验验证结果表明该集成方法具有较高的可靠性，其X、Y、Z方向尺寸误差的ARE依次为5.95%、4.92%、3.97%。

Abstract:

Selective laser melting (SLM) is one of the most widely used technologies for additive manufacturing of metallic materials, and its forming accuracy still cannot meet the requirements of practical applications due to the complex thermo-physical processes in the forming process. Therefore, in order to improve the dimensional accuracy of SLM formed parts, this study proposes an integrated Response Surface Methodology (RSM) and Non-dominated Sorting Genetic Algorithm-II (NSGA-II) approach to optimize the dimensional accuracy of SLM formed GH3265 superalloy. Firstly, a model of SLM process parameters with dimensional errors in X, Y and Z directions was developed using response surface methodology, and then the model was optimized by NSGA-II for multi-objective optimization. The results show that: the model constructed by the response surface method has high prediction accuracy, and the correlation coefficients R2 are 0.9456, 0.9842, and 0.9704 in order; The optimization algorithm is able to obtain the optimal interval of the processing parameters at 1500 iterations: the laser power is 250.8-310W, the scanning speed is 1028-1400mm/s, the hatching space is 0.071-0.084 mm; The experimental validation results show the high reliability of the integrated method with ARE of 5.95%, 4.92%, and 3.97% for dimensional errors in X, Y, and Z directions, in that order.