基于不同优化算法的 HT250 基体再制造工艺参数优化

目的探讨不同优化算法下HT250基体再制造工艺参数的优化效果。方法利用Taguchi试验设计方法设计4因子3水平共18组试验,通过亚激光瞬间熔技术修复HT250基体的表面缺陷,利用响应曲面法(RSM)和BP神经网络-模拟退火算法(BPNN/SAA)对其修复过程的工艺参数进行优化,分析输入功率P,单次修复时间t,速度v和保护气体流量G等4个因素对修复后试样抗拉强度的影响,并对不同优化算法的优化效果、准确性和稳定性进行探讨。结果 HT250基体修复过程中最优工艺参数为:输入功率2960 W,持续时间0.62 s,速度6 mm/s,气体流量3 L/min。在此参数下所获取的修复试样最大抗拉强度为230.52 MPa。结论抗拉强度受输入功率P和单次修复时间t影响显著,对其他元素呈弱依赖性。BP神经网络-模拟退火算法较响应曲面法更适合对亚激光瞬间熔的工艺参数进行优化。

Process Parameter Optimization of Remanufactured HT250 Matrix Based on Different Optimization Algorithms

Objective To investigate the optimization effect of the remanufacturing process parameters of the HT250 matrix under different optimization algorithms. Methods Experiments were designed using a factorial design based on a Taguchi L18 orthogonal array. The surface defects of HT250 substrate were repaired by sub laser instant cladding technology, and a hybrid method that included the response surface methodology (RSM)-back propagation neural network (BPNN)-integrated simulated annealing algorithm (SAA) was proposed to search for an optimal parameter setting of the remanufactured HT250 matrix, and the effects of input power, processing time, velocity and gas flow on the tensile strength of the remanufactured sample were also analyzed in detail. In addition, the optimization results, stability and veracity were analyzed to compare the results of BPNN integrated SAA with that of the RSM approach. Results The optimal remanufactured HT250 matrix conditions were input power of 2960 W, processing time of 0. 6 s, speed of 6 mm / s, gas flow of 3 L / min. The maximum tensile strength of the remanufactured sample under these conditions was 230. 52 MPa. Conclusion The results showed that the tensile strength was significantly influenced by the input power P and single repair time t, while the influences of other factors were weak. The BPNN / SAA method was more effective than RSM for the optimization of remanufactured HT250 matrix.