基于PCA和改进的KNN算法的船舶尾气识别算法

Ship Exhaust Recognition Algorithm Based on PCA and Improved KNN Algorithm

通过无人机搭载气体传感器,可以方便地检测码头靠泊船只的尾气,通过分析尾气中的硫化物和氮化物的含量,来检测靠泊船只是否使用违规燃油,但是实践发现船舶尾气中的硫化物和氮化物的识别率不高,因此采用气体传感阵列对采集的气体进行信息处理,以此提升采样气体的分类准确率。传统的K-近邻算法采用近邻决策原则,对尾气的分类效果较好。但在应对大量气体数据时分类时间过长,效率比较低下。针对这个问题,提出一个基于主成分分析和样本聚类的K-近邻算法相结合的船舶尾气分类算法。实验结果表明,改进的K-近邻算法能在保持分类的准确率的条件下,大大减少分类的时间。

The UAV equipped with gas sensors can easily detect the berthing vessel dock exhaust gas by analyzing the exhaust of sulfide and nitride content, to detect whether the berthing vessels using non-compliant fuel, but the practice found in the ship exhaust vulcanization The recognition rate of matter and nitride is not high, therefore, the gas sensing array is used to process the collected gas to improve the classification accuracy of the sampled gas. The traditional K-nearest neighbor algorithm adopts the principle of nearest neighbor decision-making, and the classification of tail gas is better. However, when dealing with a large amount of gas data, the classification time is too long and the efficiency is relatively low. To solve this problem, proposes a marine tail gas classification algorithm based on K-nearest neighbor algorithm of principal component analysis and sample clustering. Experimental results show that the improved K-nearest neighbor algorithm can greatly reduce the classification time while maintaining the accuracy of classification.