基于BP算法与遗传算法的气体识别模型应用

Application of Gaseous Recognization Model Based on BP Algorithms and Genetic Algorithms

本文采用遗传学习算法和误差反向传播算法（BP 网络）相结合来训练前馈人工神经网络（BPN），使网络收敛速度加快并避免局部极小。依据算法建立网络模型，用小批量训练替代单样本训练和大批量样本训练，提高网络的训练速度。通过模拟，预测结果表明，该算法收敛速度快，预测精度高，为气体模糊识别和预报提供了一种新思路和新方法。

The mixed algorithm combining genetic algorithm and BP is used to train BPN to speed up network training and avoid local minimums. The network model is established by algorithm, the network training speed will be improved by a small batch of trainings, instead of a single-training or a big batch training. The prophecy results from computer simulation shows that this algorithm can speed restraint and get high accuracy, which provides a new method for the recognition and prediction of gaseous ambiguity.