入党基本条件
Title：CALIBRATING METHOD FOR ELECTRONIC COMPASS
Electronic compass is an electronic compass by measuring the Earth's magnetic field to achieve directional navigation device, it is an important navigational tool that can provide real-time data objects heading and attitude, and has a small size, low cost, fast respon , no accumulated error, etc., are widely ud in mobile robots, vehicles, aircraft and other orientation subsystem; However, due to the electronic compass is bad on magnetic principles calculations magnetic bearings, their work environment, in addition to other external magnetic field Earth's magnetic field will inevitably affect the output of the electronic compass, resulting in errors affect measurement accuracy. Therefore, how to reduce outside interference precision of its output is engineering applications must be addresd.
常州竹海
Electronic compass the existing literature has propod veral possible compass error compensation method. Yuen Chee Wing paper, "three-axis magnetic heading nsors full attitude error compensation 'propod compass error into quadrature error, zero and nsiti
遇到挫折怎么办
vity errors were compensated, high compensation accuracy of the method, but in the compensation process requires non-magnetic turntable , but also the computer automatically when the measured rotation compass X and Y-axis nsor output maximum, minimum, the calibration process is complex and requires high equipment; HIGHWAY AND TRANSPORTATION Wang Jiaxin paper "Development of high-precision electronic compass" and Shao Tingting, ROCKETS paper "electronic compass tilt and Luo slip compensation algorithm rearch," the least square method to compensate for the electronic compass, the method requires Compass 0 ° ~ 360 ° rotation average sampling, operation is relatively cumbersome and The amount of data samples of different sizes will have a greater impact fitting result, the amount of data is too small, the compensation effect is not good, too much data can cau performance degradation fitting; Yangxin Yong, Huang Shengguo paper "magnetic heading measurement system Error Correction Method "and Qi Zhang, Liang-shui Lei, Jiang Fan, Song Liu's paper" Autocalibration of a magnetic compasswithout heading reference "error ellip hypothesis is propod bad on the model compensation method, becau only a hypothetical mod
el ellip bad on the experimental experience suggested that the lack of theoretical deduction, the compensation effect is not very satisfactory; Chao Min, Jiang Oriental, Wen rainbow paper, "magnetic compass error analysis and calibration" using an analytical method to establish the direction of the magnetic measurement system is more accurate models, and electronic compass compensation in the horizontal conditions, but the compensation process to identify more parameters (up 9), and the results show the compensation effect is similar to oval hypothetical model; Hao Zhenhai, Huang Shengguo paper, "bad on a combination of differential magnetic compass heading system." propod a "differential magnetic compass" (DMC, Differential Magnetic Compass) design program, which us two identical magnetic compass to determine whether the system is a combination of low-frequency interference occurs, if the system glitches did not happen, then the u of magnetic compass navigation, low-frequency interference will occur if the system switches to the gyroscope system navigation. The program does not substantially improve the magnetic compass measurement accuracy, and the navigation program consists of a plurality of magnetic compass and gyroscope structure,
will lead to greatly increa the cost; Lu Wang, Zhao Zhong and other paper "Magnetic Compass Error Analysis and Compensation" using BP neural network error model, and u LM learning algorithm to train the network, the method need not 0 ° ~ 360 ° within the average sampling, with a neural network can be approximated function with arbitrary precision characteristics, with high compensation accuracy, but the BP neural network convergence is slow, the right to t initial values need to be very careful, and easy to fall into local minimum.
In summary, the current electronic compass calibration methods are the main method of least squares, oval assumption France, BP neural network method, etc. The methods are more cumbersome calibration procedure or the existence of the necessary instruments for demanding calibration or calibration issues such as lack of precision.
It is an electronic compass by measuring the Earth's magnetic field to achieve directional navigation device, it is an important navigational tool that can provide real-time data objects heading and attitude, and has a small size, low cost, fast respon , no accumula
ted error, etc., are widely ud in mobile robots, vehicles, aircraft and other orientation subsystem; However, due to the electronic compass is bad on magnetic principles calculations magnetic bearings, their work environment, in addition to other external magnetic field Earth's magnetic field will inevitably affect the output of the electronic compass, resulting in errors affect measurement accuracy. Therefore, how to reduce outside interference precision of its output is engineering applications must be addresd.
In order to simplify the calibration procedure, reducing the requirements for the necessary equipment and to improve calibration accuracy, the prent invention designed an electronic compass calibration method, the calibration method bad on adaptive differential evolution method and Fourier neural network (Adaptive Differential Evolution and Fourier Neural Network, ADE-FNN), by means of Fourier neural network modeling of the electronic compass error, the neural network using orthogonal Fourier ries excitation function as a network, and us adaptive differential evolution algorithm to train the network weights, get more preci The error model to compensate for the measured
专业证书values of the compass, so as to improve the precision of the compass.
党员大会会议议程
峰回路转意思Differential evolution method bad on adaptive neural networks and Fourier electronic compass calibration methods, including access to training samples, the u of Fourier neural network (FNN) build error compensation model and the u of adaptive differential evolution algorithm (ADE) to train the neural network weight value of three parts, to achieve the specific steps are as follows:
半长轴Step one, get training samples;
Step two, identified neural network structure;
Step three, to create an electronic compass error model and choo the training targets;
Step four: According to the training sample using an improved adaptive differential evolution algorithm for training the neural network;
视死如归什么意思
Step Five, will optimize the neural network weights obtained values into the neural network to obtain a more accurate compensation models;
Step Six, the electronic compass output measurement values into compensate for neural network input and output compensation value.
The invention took into account the impact of measurement precision electronic compass, propos a differential evolution method bad on adaptive neural networks and Fourier concentration calibration method to compensate, on the one hand to simplify the calibration process, that is just the turntable rotate (no uniform rotation) electronic compass one week for training samples and the corresponding true value; hand with better results, ie, more precily offt outside interference, improve measurement accuracy.
