The Research Of Lubrication Moisture Content Prediction

The quality of lubrication is very important to engine of motor-driven vehicle, which decides the need of motoring of lubrication. At present, lubrication monitoring system is able to monitor temperature, viscosity, density and dielectric constant. But besides of them there are moisture content and prill which are also important to decide quality of lubrication. So to compliment the lubrication monitoring system, this paper is going to finish the ability of moisture content prediction of lubrication.The first solution is to chose the right lubrication samples. According to the change of moisture content demarcated by authority the samples are picked up and recorded. After the research of relations between each two parameters, the samples are decided to get according to the change of dielectric constant for the intimate relation between moisture contend and dielectric constant.The foundation of lubrication moisture content prediction is Support Vector Machine. Make use of the LIBSVM tool case in MATLAB to complete the prediction. Meanwhile, research the SVM structure deeply to plant SVM model in the slave updated. Then a sophisticated system is finished.Now, lubrication monitoring system contains master and slave. Master communicates with supervisory computer, analysis the orders from supervisory computer, and transfers the order to slave. Slave transfers data displayed in LED to master. However, because of design complexity the monitoring system shows heat and instability, a updated system is need.This subject update lubrication monitoring system chosing STM32OF28335with2CAN buses. CANA communicates with liquid sensor and CANB transfer the5parameters to master to finish the system hardware updation. Software contains SVM to found the prediction in MCU, so the data transferred to master contain5parameters.