CAJ | 학술논문

为了找到一种适用于激光修锐砂轮工艺参量预测和优化的方法，采用神经网络和粒子群算法，建立了激光修锐砂轮工艺参量优化模型。首先构建了工艺参量与工件表面粗糙度之间映射关系的神经网络模型，然后基于预测模型采用粒子群算法实现工艺参量优化，最后采用粒子群算法优化获取的5组工艺参量进行了激光修锐试验。结果表明，样本值与神经网络仿真输出值的相对误差小于3％，试验值与期望值的相对误差控制在6％以内。综合说明该优化模型具备良好的优化能力。
위료조도일충괄용우격광수예사륜공예삼량예측화우화적방법，채용신경망락화입자군산법，건립료격광수예사륜공예삼량우화모형。수선구건료공예삼량여공건표면조조도지간영사관계적신경망락모형，연후기우예측모형채용입자군산법실현공예삼량우화，최후채용입자군산법우화획취적5조공예삼량진행료격광수예시험。결과표명，양본치여신경망락방진수출치적상대오차소우3％，시험치여기망치적상대오차공제재6％이내。종합설명해우화모형구비량호적우화능력。
In order to find a method of prediction and optimization of laser dressing grinding wheel , an optimization model of process parameters for laser dressing grinding wheels was established based on the neural network and particle swarm optimization .Firstly, the neural network model mapping the relationship between the process parameters and the specimen surface roughness was constructed .Then, the process parameters were optimized by means of the particle swarm optimization algorithm based on the predication model .Finally, laser dressing experiments were carried out based on 5 groups of parameters optimized by the particle swarm algorithm .Experimental results show that the relative error between the sample value and output value from neural network is less than 3%and the relative error between the test value and the expected value is lower than 6%.In conclusion , the model has good ability of optimization .