The Research Of The Eddy Current Sensor Based On The Multiple Regression And Support Vector Machine (SVM)

With the rapid development of information technology,the accuracy, speed access to information in various industries have become increasingly demanding, but the access to the information channel that the sensor development can not meet this requirement. This is because sensor measurements is usually affected by other parameters of the environment, the measurement results are volatile, leading to inaccurate results. Therefore, you need to add non-target sensor during measurement to monitor environmental variables, so as to eliminate cross-sensitivity. The intelligent sensor system that the data of target and non-target sensor processed by microprocessor represents the development direction of the sensor, and a variety of sensors characterize the properties of different information, which requires a processing method to handle data, which led to data fusion algorithm to produce. Data fusion is integrating information from the target and non-target sensor to reduce the impact of environmental factors on the sensor, to achieve more accurate measurement results.In this paper, under the background of the sensor development status and data fusion technology, analyses the working principle of eddy current sensor and its temperature characteristics,then introduces the multiple regression and support vector machine to eddy current sensors to reduce the impact that temperature to sensor. On the basis of two-dimensional calibration experiment,deduced multiple regression and support vector machine algorithm data fusion model and achieve a non-linear compensation of eddy current sensor, getted various technical parameters of the model, compared the two kind of fusion algorithm. The eddy current sensors, temperature sensors and microprocessors combine to achieve a displacement measurement intelligent sensor system. System is mainly dealing the output of eddy current sensor and the temperature sensor by fusion algorithm placed through the processor’s to finish temperature compensation. The system’s LCD can display the displacement distance and ambient temperature. Temperature compensation result shows that the precision of the sensor have improved two to three orders of magnitude.