基于AHP和线性神经网络的模具标准评价

Evaluation of mould standard based on the AHP and linear neural network

在模具标准制修订工作中,是否采用国际标准或国外先进标准、预测标准对国情的适用性是重要环节。通过对我国现行模具标准的大量分析,确定标准评价因素,构建标准评价体系,运用层次分析法(AHP)计算各因素的重要程度值,并做量化分析和评分,以获得线性神经网络模型样本,进而对样本进行训练、验证,最终获得模具标准评价模型。该模型充分吸收了专家的知识和经验,降低了评价的人为因素。结果表明,基于AHP和线性神经网络的模具标准评价方法计算的最大相对误差为1.6%,该评价方法正确可行。

Whether to adopt international standards or advanced foreign standards and predicting the applicability of standard to the national conditions are important links in the mould standard system revision work.Through a large number of analyses of our current mould standard,the standard evaluation factors and the standard evaluation system were determined,and the importance of the various factors were calculated by using the analytic hierarchy process.And then by doing quantitative analysis and evaluation,the samples of the linear neural network model were obtained,and after training and validation samples,mould standard evaluation model was gained ultimately.By this,not only the knowledge and experience of the experts can be absorbed completely by the model,but also the evaluation of human factors has gone down.As the experiment shows,the evaluation result of maximum relative error is 1.6%,the method of the mould standard evaluation based on the AHP and linear neural network is correct and feasible.