新型智能船舶舱底水油分浓度检测系统

摘要

There are some problems in the prediction model of the traditional marine oil content prediction model,such as low precision and nonlinear deviation. This paper studies an optimization method based on particle swarm optimization algorithm in com?bination with least squares support vector machine algorithm(LS-SVM)to establish the prediction model of marine oil content. Parti?cle swarm optimization(PSO)is used to optimize the parameters of the least squares support vector machine model,which avoids the blindness of selecting parameters manually. Comparisons are conducted by computing the evaluation criteria of the average rela?tive error(MRE)and root mean square error(RMSE). The experiment is designed to collect the data for modeling. Compared with the tradition least squares support vector machine model and the least squares fitting linear model,this method effectively improve the prediction precision of the oil content,and it laies a solid foundation for the new intelligent marine bilge oil content detection.%针对传统船用油分浓度计浓度预测模型存在非线性偏差影响检测精度的问题.研究采用了一种基于粒子群优化(PSO)的最小二乘支持向量机(LS-SVM)建立船用油分浓度预测模型,利用PSO算法对LS-SVM船用油分浓度的预测模型的参数进行了优化,避免了人为选择参数的盲目性.以平均相对误差(MRE)和均方根误差(RMSE)作为评价标准进行评估.设计实验采集数据进行建模,对比未经优化的最小二乘支持向量机建模、传统的最小二乘拟合线性建模,有效提高了油分浓度的预测精度,为新型智能船舶舱底水油分浓度检测奠定了基础.