人工神经网络在木材损伤识别中的应用

利用人工神经网络(ANN)对含有不同缺陷程度的木材进行了定性和定量解析.通过选择合理的神经网络结构,建立有效的训练样本集,确定合理的参数及训练方法,对3种不同缺陷类型木材进行了解析,考察了网络的泛化能力.结果表明:该网络能够对不同缺陷程度的木材进行准确地识别,而且本研究为人工神经网络在木材缺陷损伤的定性和定量分析方面提供了一种有效的方法.

Damage-Type Recognition of Wood Based on Artificial Neural Network

The artifical neural network(ANN)can approximate the arbitrary nonlinear mapping relationship.So it is a powerful damage identification tool for unknown systems.The result is proved by the experiment.The different damage-types of the wood can be recognized by ANN in acoustic emission testing if the reasonable neural network iS chosen.Then,the efficient train-sample collection is constructed,and the reasonable parameter and train means are decided.Results indicate that ANN is suitable for the identification and the quantification of the wood damage analysis.