基于灰色BP神经网络的实验材料供应预测

针对单一BP神经网络对实验器材需求量预测准确度偏低的情况，提出了一种将灰关联分析与BP神经网络相结合的实验材料需求预测模型。先用灰关联分析法计算出影响需求量的各因子之间的关联度，然后选择关联度较大的3个优势因子作为BP神经网络的训练样本，建立了3层BP网络预测模型。以某实验材料的实际需求量为实例进行算法检验，对比分析了灰色BP网络模型和单一BP网络模型的预测准确性。实验结果表明：灰色BP网络模型将原有6 10 1的BP网络结构简化为3 6 1结构，灰色BP网络模型预测的最大相对误差仅为-1.36%，而单一BP网络模型的预测最大相对误差为-4.18%，灰色BP模型比单一BP模型的预测精度更高，结构更简单。

Prediction of laboratory equipment support based on grey relation analysis and BP neural network

To deal with low accuracy of single BP neural network in the prediction of laboratory equipment demand, a prediction model based on grey relation analysis and BP neural network is proposed. The correlation degrees between influencing factors were first calculated using grey relation analysis method. Three major factors with higher correlation degrees were chosen as training samples for the BP neural network and a three layered BP network prediction model was established. The real laboratory equipment demand of an experiment was taken as an example to test the algorithm. And a comparative analysis of the prediction accuracy of the grey BP network model and the single BP network model was done. The result shows that the grey BP network model reduces the topology of BP network from 6 10 1 to 3 6 1. Its maximum relative error is -1.36%, while the one of the single BP network model is -4.18%. The prediction accuracy of the grey BP network model is higher and its structure is also simpler.