基于ReliefF和PSO-SVM的木质材料超声无损分类技术

Ultrasonic Non-destructiveClassification for Wood Materials Based on ReliefF and PSO-SVM

指出了木质材料是一种广泛使用的可再生原料,对其进行正确分类可提高原料的利用率并防止在交易时欺诈行为的发生。鉴于常用木质材料分类方法的不足和支持向量机参数选择困难导致分类准确率低的问题,以超声无损检测手段为基础,提出了一种基于ReliefF特征选择和PSO-SVM分类器的木质材料分类新方法。该方法根据接收到的木材超声信号,通过ReliefF算法选取出最有代表的特征,然后使用粒子群优化算法对支持向量机的参数进行优化,提高了木质材料的分类效率和准确率。实验表明,基于ReliefF和PSO-SVM的方法能有效解决木质材料的在线分类问题。

Wood material is a widely used and renewable raw to improve its utilization and avoid fraud in trading activities. materials, which needs to be classified correctly m order Given the lack of traditional wood classification methods and the low classification accuracy resulted from difficulties of choosing suitable parameters in the Support Vector Machine, a new wood classification method was proposed in this paper combining the Ultrasonic Non--destructive Testing , ReliefF feature selection algorithm and PSO--SVM classifier.By using the ReliefF algorithm to select the most representative features from the received wood ultrasonic signal and optimizing the parameters in the Support Vector Machine with Particle Swarm Optimization algorithm, the proposed method improved the efficiency and accu- lacy of the classification of wood materials. The experimental results showed that, the proposed method based onRe- liefF and PSO--SVM could effectively solve the problem of online classification of wood materials.