基于BP神经网络的福特精益生产衡量指标BTS预测

Forecast of an Important Indicator of FORD Lean Manufacturing Based on BP Neural Network

福特精益生产衡量指标BTS(Build To Schedule)是精益生产重要衡量指标之一,由于BTS在一定程度上可以衡量生产能力与排产计划相适应的水平,所以对BTS的预测有重要意义。文中预测将利用某汽车底盘厂后桥生产线几个月的生产数据,运用BP神经网络建模技术,对其后桥生产线的BTS指标进行预测,但由于标准BP神经网络存在难以收敛及易陷入局部最小值等缺陷,所以将利用动态陡度因子来改变激励函数的陡峭程度。试验表明,通过构建的预测模型所获得的预测值与真实值之间的误差相对较小,在根据预测结果调节生产线的排产计划后,其计划可执行率有所提升。

Nowaday Ford lean production index BTS （ Build To Schedule） is one of the important indicators to measure the lean production, because the BTS can measure the production capacity and production plan to adapt the level to a certain extent, so the prediction of BTS has important significance. Few months forecast production data will use an automobile chassis factory rear axle production line in this paper,using BP neural network modeling after the BTS on the subsequent bridgethe production line are predicted, but due to the standard BP neural network is difficult to convergence and easy to fall into local minimum and other defects, sothis paper will use dynamic steepnessfactor to change excitation the steep degree function. Test shows that the prediction gain prediction model of the construction of the value and the error between the true value is relatively small,the predicted results to measure production line scheduling executable rate also has an impor-tant guiding significance.