基于RVM的多功能自确认水质检测传感器

Multi-function self-validating water quality monitoring sensor based on relevance vector machine

提出了一种多功能自确认水质检测传感器功能模型,可以同时测量水的温度、盐度和pH值,并可对自身工作状态进行自确认。提出了一种基于相关向量机的多功能自确认传感器故障诊断和数据恢复方法,在二分类机基础上利用层次扩展方法得到基于相关向量机的多分类机,对传感器进行故障诊断;利用水质检测过程中多个参数之间的相关信息,解决了非线性、小样本条件下的传感器故障数据恢复问题。构建了多功能自确认水质检测传感器实验平台,实验结果表明故障诊断识别率达到98%,数据恢复相对误差在±4%以内,提高了传感器的可靠性。

A function model of multi-function self-validating water quality monitoring sensor is proposed, which can measure temperature, salinity and pH value at the same time and resolve the problem of self-validating working state in current water quality monitoring sensor. The fault diagnosis and data recovery methods based on relevance vector machine are studied. The multiple classifiers expanded from binary classifier based on relevance vector machine are obtained to carry out the fault diagnosis of the sensor. The relativity of multi-parameters in water quality monitoring is used sufficiently to achieve the data recovery of multi-function self-validating water quality monitoring sensor in non- linear and small sample conditions. A sensor experiment platform was constructed and experiments were carried out. Experiment results indicate that the fault diagnosis recognition rate reaches to 98% , the relative error of data recovery is less than ±4% , and the reliability of the sensor is improved.