【摘要】 为了准确地对战场弹药消耗总量进行预测,首先将影响弹药用量的各种因素分为l类,然后由n位专家对每个因素评价打分,建立l个评价矩阵,在对评价矩阵中的数据进行预处理后,再用投影寻踪的方法,将评价矩阵作降维处理,得到l个综合评价指标;最后将这l个综合评价指标作为输入项,实际弹药用量作为理想输出,训练神经网络,得到网络结构参数,最终实现对新战例弹药用量的神经网络预测。更多还原

【Abstract】 In order to forecast the battlefield ammunition consumption accurately,first all factors that affect the ammunition consumption are divided into l kinds.Then n experts appraise every factor,give marks and construct l evaluation matrices.After the data of evaluation matrices being pre-handled,the orders of the evaluation matrices are reduced by using the method of project pursuit and l synthesis evaluation indexes are gotten.Last the l synthesis evaluation indexes are taken as inputs,the actual ammunition consumptions are taken as target outputs and the neural network is trained.After being trained the weight parameters of neural network are obtained and the neural network construction is determined.Thus the forecast of a new battlefield ammunition consumption can be realized by using the neural network.更多还原