Design And Decoupling Analysis Of Three-axis Coal-cutting Force Sensor

In this paper. a kind of three-axis coal-cutting force sensor is presented. We can obtain force information of shearer in the cutting process through the sensor, which provides a basis for in-depth research of cut coal mechanism, quantitative assessing of cut coal features, reasonable options and the use of coal cutters and cut teeth.The sensor measures three-axis force signals through stress deformation of the strain gauge in elastic body.Firstly, sensor is comprised of elastic body, fitting and a protective covering. The elastic body is the core part of the sensor. It is a torus structure, and four slots are opened in the torus structure of the above and below. Slot is symmetrical,45°phase difference.The thin-walled area between the above slot and the below slot is named shear stress area.Different shear strain area corresponding measure different direction of force. The size of elastic body structure is optimized through orthogonal tests.Secondly, the static analysis, modal analysis, harmonic response analysis and transient response analysis of sensor had been run by finite element software Ansys. Strain changing regularity of the sensor in the static load was obtained, so that we can determine the position of the strain gauge patch was determine. Meanwhile, the natural frequency, vibration mode and frequency response characteristics of the sensor were derived.Thirdly, stain gauge was pasted on the elastic body. Through the reasonable Bridge circuit, coupling between dimensions was eliminated in theory. Sensor calibration experiment was carried out using the method of plus standard farmar. Nonlinearity, sensitivity, hysterisis error etc static performance index of the sensor were obtained by the static calibration. Lastly, the static linear decoupling was carried out by least square method, and BP neural network method and independent component analysis method was used for static nonlinear decoupling.What's more, it was confirmed that BP neural network decoupling method was much more accurate than the least square method and the independent component analysis method.Figure [63] Table [13] Reference [44]...