Research On Error Compensation Of Inclinometer In Pose Measurement Of Shield Machine

To ensure the shield machine can tunnel along the designed line in engineering, it is necessary to measure its position and pose accurately in real-time. The inclinometer is an important component in pose measurement of shield machine and it is utilized to measure the pitching angle and the rolling angle directly. However, the measurement error of the inclinometer is often large due to the impact of nonlinearity and temperature drift. Therefore, the compensation methods of measurement error for the inclinometer were mainly discussed in this paper based on the analysis of pose measurement principle for shield machine. The main contents are as follows:(1)The impact of the inclinometer's error on measuring the attitude angles of shield machine was analyzed using Monte Carlo method. The results show that the measurement accuracy of the inclinometer not only determines the measurement accuracy of the pitching angle and the rolling angle directly, but also influences the measurement of the yawing angle largely.(2)To research the measurement characteristic of the inclinometer, a rotation experiment platform was designed for the inclination angle measurement experiment. The results demonstrate the measurement output of the inclinometer really has the nonlinear error and the temperature drift error. The larger the inclination angle is, the larger the nonlinear error and the temperature drift error will be.(3)The compensation method of measurement error for the inclinometer on the basis of polynomial fitting was researched. In order to overcome the shortages of the common polynomial fitting method, such as the needed fitting times are more,the calculated amount is larger and so on, the improved polynomial fitting method was put forward which was to build the linear model of the temperature drift error first and then consider the impact of the nonlinearity on the model. As a result, the measurement error of the inclinometer was controlled less than±0.3 milli-radian only 2 times polynomial fitting, but the common one needed 4 times.(4)The calibration method of Neural Network was proposed in order to decrease the measurement error of the inclinometer further. An inverse model of the inclinometer's measurement system was set up using Neural Network. The mapping relationship from the measurement output of the inclinometer and the environmental temperature to the true inclination angle was realized through Neural Network's training. Thereby, the temperature compensation and the nonlinearity correction were accomplished. The proposed method not only can take the effect of environmental temperature and the nonlinearity into account simultaneously, but also obtain a higher compensation accuracy than the method of polynomial fitting. After compensation, the maximum of the inclinometer's measurement error decreases from 1.6 milli-radian to 0.2 milli-radian.Finally, a summary of the paper and the expectation of the future work was given.