短生命周期产品库存信息自动识别方法仿真

Simulation of Automatic Identification Method for Inventory Information of Short Life Cycle Products

针对当前方法在对短生命周期产品库存信息进行自动识别时存在识别准确率不高、相对误差较大,且识别耗时较长的缺点,提出一种基于BP神经网络的生命周期产品库存信息自动识别方法。基于给定的假设条件,采用主成分分析法计算短生命周期产品库存信息的协方差系数,得到短生命周期产品库存信息特征值及其正交单位特征向量之间的关系,为了减少变量的数目,选取若干个主成分,使得短生命周期产品库存信息特征值累积贡献率大于85%,计算短生命周期产品库存信息在这若干个主成分上的得分,实现短生命周期产品库存信息降维处理。在上述基础上,根据给定参数对三层BP神经网进行学习和训练,计算BP神经网络各层的输入输出特征向量以及各个神经元误差函数值,利用计算结果对训练好的BP网络进行连接权重和阈值修正,采用修正后的神经网络对短生命周期产品库存信息进行自动识别。仿真测试结果显示,提出方法能够实现短生命周期产品库存信息的高准确、低误差以及快速识别。

This article puts forward a method to automatically recognize inventory information of life cycle product based on BP neural network. Based on given assumptions, the principal component analysis method was used to calculate the covariance coefficient of inventory information of short life cycle product and obtain the relationship between the feature value of short life cycle product inventory information and its orthogonal unit feature vector. In order to reduce the number of variables, several principal components were selected, so that the cumulative contribution rate of feature value of short life cycle product inventory information was more than 85 %. Moreover, the scores of short life cycle product inventory information on these several principal components were calculated to realize dimension reduction for short-life cycle product inventory information. On this basis, the three-layer BP neural network was learned and trained according to given parameters. After th at, the input feature vector and output feature vector of each layer of BP neural network and the error function value of each neuron were calculated. Finally, connection weight and threshold correction were performed on the trained BP network, and then the modified neural network was used to automatically recognize inventory information of short life cycle product. Simulation results show that the proposed method can achieve fast recognition with high accuracy and low error for inventory information of short life cycle products.