Referring to the question that it is hard to determine the index weight of comprehensive evaluation for engineering materials selection, the model of weight determination for evaluation index based on BP neural network was established. Taking the weight determination of evaluation index of coating materials used on automobile body as an example, regarding 15 kinds of coating materials evaluated by weighted average as training samplefthe data of 8 kinds of evaluation indices and evaluation results obtained from weighted average were regarded as sample data), weights of 8 kinds of evaluation indices were calculated according to the model of weight determination mentioned above. 5 kinds of candidate coating materials used on automobile body were evaluated according to improved technique for order preference by similarity to solution(ITOPSIS) with the weights obtained The evaluation result shows that acrylic resin is the best in candidate materials, which accords with industry practice. All above shows that the weights that used in comprehensive evaluation are objective and reasonable, in other words, the weights of evaluation indices for engineering materials determined by BP neural network are reasonable and feasible.%针对工程选材综合评价中评价指标的权重难以合理确定这一问题,构建了BP神经网络法赋权模型.在此基础上,以汽车车身用有机涂层材料评价指标权重的确定为例,以经加权平均法合理评价的15种有机涂层材料作为网络训练样本(8个评价指标和加权平均法所得评价结果作为具体的样本数据),根据上述赋权模型,确定汽车车身用涂层材料8个评价指标的权重.以所得权重,根据改进理想解法对5种候选汽车车身用涂层材料进行综合评价.评价结果表明,最佳候选材料为丙烯酸树脂,与实际应用一致.说明评价所采用的权重客观合理,即采用BP神经网络法确定工程材料评价指标的权重合理可行.
展开▼