论文部分内容阅读
将人工神经网络方法引入到投资项目后评价中,以内部收益率、静态投资回收期、财务净现值率、投资回报率四项指标作为效益等级评价依据,同时采集了同类项目20个样本,划分五个评价等级,以BP神经网络模型为基础,运用MATLAB验证三种BP改进算法,证实了自适应学习速率动量梯度下降反向传播算法是评价该项目的最佳改进算法,在较快的收敛速度下,达到了更小的训练误差,从而实现了对项目效益等级的评价。
The artificial neural network method is introduced into the post-investment project evaluation. The four indexes of internal rate of return, static investment recovery period, financial net present value rate and return on investment are taken as the evaluation basis of benefit level. At the same time, 20 samples of similar projects are collected, Divided five evaluation levels, based on BP neural network model, using MATLAB to verify three kinds of BP improved algorithm, confirmed that the adaptive learning rate of momentum gradient descent backpropagation algorithm is the best improved algorithm to evaluate the project, in the faster Convergence rate, to achieve a smaller training error, so as to achieve the evaluation of project efficiency level.