The Application And Research Of Bayesian Algorithm In A Trajectory Measuring Meter

A trajectory measuring instrument is a conventional measurement apparatus(a kind of photoelectric theodolite),which is widely used in military, construction and other industries,and main objective of the space used to measure the level and azimuth angle.As the rapid development of science and technology of our countries' aviation,aerospace and defense,a lot of experimental data need to be measured by measuring instruments.Therefore,measuring instrument for measuring the accuracy of the data put forward higher requirements.In order to further improve the system of measurement precision measurements,it's necessary to improved design for measuring instrument's hardware and software to discuss these issues.In this paper,the trajectory angle measurement system for rotary encoders is improved,and on the measurement system,the data-processing software of the system is re-engineered through the combination of hard and soft approaches to improve the system of measuring instrument measurement's data accuracy.First of all,the measurement system structure and its working principle are analyzed. Basing on that,the breakdown device of the rotary encoder of angle-measuring system is improved designed,including improvements on the frequency multiplication circuit and A/D converter circuit of the breakdown device to enhance the rotary encoder resolution and processing speed.Second,the issues involved in Bayesian algorithm and neural networks are focused,the optimization parameters of objective function drawn through Bayesian inference to improve network generalization ability.Finally,data errors caused by various reasons are analyzed when the meter at work,in particular due to the rotating shaft,dynamic lag of follow-up system,and the data error decomposition model is established.Using Bayesian regularization neural networks to do error data correction experiments measured by the trajectory meter,the results are analyzed and summed up.Experimental results show that the Bayesian regularization neural networks can track the meter data error correction of the movement and achieve good results,up to raising the trajectory measurement accuracy of measurements.