Novel piecewise compensation method for FOG temperature error

FU Jun, JIANG Sai, QIN Fang-jun, FENG Ka-li

(1. Electrical Engineering College, Naval University of Engineering, Wuhan 430033, China; 2. Office of R&D Naval University of Engineering, Wuhan 430033, China)

Novel piecewise compensation method for FOG temperature error

FU Jun1, JIANG Sai2, QIN Fang-jun1, FENG Ka-li1

(1. Electrical Engineering College, Naval University of Engineering, Wuhan 430033, China; 2. Office of R&D Naval University of Engineering, Wuhan 430033, China)

The temperature characteristics of fiber optic gyro(FOG) is analyzed, and its large-range temperature test is designed. The effects of different temperatures and temperature rates on the output of FOG are studied, and different temperature characteristics in different temperature ranges of FOG are achieved. In order to improve the compensation precision for temperature error, a novel idea of multi-model piecewise fitting is proposed based on the gyro temperature characteristics in low, middle and high temperature ranges respectively by using an artificial neutral net. The temperature errors are compensated based on the established models, and the results indicate that the model can effectively reduce the FOG temperature bias, and the precision is improved by one order of magnitude.

fiber optic gyro; temperature bias; multi-model; artificial neutral net

Fiber optic gyro is a kind of inertial measurement instrument based on the sagnac effect, which forms the core of marine strapdown inertial navigation system, and makes use of the full optical fiber structure of solid carrier rotating angular velocity measurement[1-2]. Compared with the traditional mechanical gyroscope, FOG has many outstanding advantages, such as high precision, impact resistance, good vibration resistance, large dynamic range, not sensitive to acceleration of gravity, etc.[2]. FOG has broad and bright prospect. In recent years because of its potential advantages, fiber optic gyroscope is widely used in car navigation, attitude control rocket, arms control and other fields[2-5].

For practical application of fiber optic gyro, it is generally required to have wide working temperature range. As the core of the inertial navigation system device, the main components such as optical fiber ring of fiber optic gyro, integrated optical device, coupler and so on are very sensitive to temperatures[6], when the environment temperature changes, the gyro bias significantly increases, the scale factor linearity is also significantly worse[7]. Therefore, to study the temperature characteristic of fiber optic gyroscope, modeling and compensation in order to improve the precision of gyro is very necessary. Domestic research on fiber optic gyro temperature modeling mainly concentrates in the single model, ignoring the fact that the influence of temperatures on the optical fiber gyro in different temperature range is different. Based on the study , the temperature models can be divided into three differentmodels according to temperature changes from low, medium to high scope. neural network based on piecewise multiple model of fiber optic gyro temperature error compensation method is put forward, which can effectively improve the precision of fiber optic gyro temperature error compensation.

1 Temperature influence mechanism analysis

The influence of temperature on the optical fiber gyro includes two aspects. One is the gyro working environment temperature influence on the working state of gyro. The second is the gyro material properties sensitive to temperature[6]. The core component of fiber optic gyroscope is sensitive to temperature, temperature change, which will lead to ring optical fiber refractive index changes. The two beams back propagation of optical fiber will have a different optical path to form the reciprocal effect. Gyro bias is mainly caused by the mutual out phase shift[8].

When the two beam interference light ,which is called by clockwise (CW) light and counterclockwise (CCW)light respectively, transmitting through the length of L, the phase delay of clockwise and counter-clockwise is respectively as follows[9]:

Where n is the refractive index of fiber ring, c is the speed of light propagation in optical fiber ring, z is the distance to the endpoint, time T for light through the optical fiber ring.

The reciprocal phase delay of fiber optic ring induced by temperature change is:

Where waveguide.

From the above analysis, the temperature error of fiber optic gyroscope can be quantitatively measured.nc is the speed of light in the

2 Modeling method and data processing

Fiber optic gyro temperature changes and causes for the nonlinear relationship between the temperature and gyro bias. So nonlinear model is adopted to improve the modeling accuracy. Artificial Neural Network (ANN) is a kind of simple calculation parameters and nonparametric model. The RBF neural network is a kind of three layer forward network, which has a simple structure, concise training and fast convergence rate. RBF can approximate any nonlinear function[9-10].

Temperature compensation is to use the method of software to modify the temperature error of gyro, compensation for gyro bias caused by temperature change[10-11]. So it is necessary to establish the relevant temperature and drift model. Based on the model, according to the measurement of temperature we predict the value of gyro bias with respect to the corresponding temperature, then get the temperature induced gyro bias from actual measured drift and predicted one.

In experiment a gyros labeled as X is put into a temperature box. The temperature is changed from 15～45 ℃ gradually. Test data is saved every 1 ms. Using MATLAB to average the test data, we find out three kinds of interesting phenomenon. Within the scope of the 15 ～0 ℃, gyro bias is approximately proportional to the temperature changes. Within the range of 0～30 ℃, gyro bias is approximately inverse proportional to the temperature changes. Within the scope of the 30 ～45 ℃, gyro bias is approximated as a quadratic function relation with the temperature changes. According to the findings, the temperature is divided into the following three interval: -15～0℃, 0～30℃and 30～45 ℃. The temperature value and temperature change rate are inputs, gyro bias is the output of RBF network. By connecting three models of different temperature scope, a complete model of the temperature bias compensation is formed. For an example, gyro, the effect of temperature on the X gyro bias and compensation effect is shown in Fig.1.

Fig.1 Compensation effect

Fig.2 Compensation effect of the first temperature scope

Before the compensation, the overall average for optical fiber gyro bias was 11.49 (°)/h, the maximum was 11.64 (°)/h. After compensation the average becomes 0.011 (°)/h, maximum becomes 0.14 (°)/h. It can be seen that the compensation effect is obvious.

In the test experiment, compensation effect of the first temperature scope is shown in the Fig.2.

Fig.3 Compensation effect of the second temperature scope

Before compensation the average gyro bias was 11.50 (°)/h, the maximum was 11.64 (°)/h. After compensation the average becomes 0.024 (°)/h, maximum becomes 0.18 (°)/h.

In the test experiment, compensation effect of the second temperature scope is shown in the Fig.3.

Before compensation the average gyro bias was 11.49 (°)/h the maximum was 11.62 (°)/h. After compensation the average becomes -0.019 (°)/h, maximum becomes 0.13 (°)/h.

In the test experiment, compensation effect of the third temperature scope is shown in the Fig.4.

Fig.4 Compensation effect of the third temperature scope

Before compensation the average gyro zero-bias was 11.48 (°)/h, the maximum was 11.63 (°)/h. After compensation the average becomes 0.028 (°)/h, maximum becomes 0.16 (°)/h. Experiment results are listed in the following table 1.

Tab.1 Compensation for temperature bias (°)/h

Temperature errors are compensated based on the established models. The results indicate that the model can effectively reduce the FOG temperature bias, the precision is greatly improved by one order of magnitude.

3 Summary

On the basis of a large number of experimental data analysis, the study found that the temperature influence on the gyro bias in different models with the different temperature range. Using RBF neural network fitting, we use Piecewise Compensation Method with different models. Test results show that within the temperature range -15～45 ℃, before compensation the mean overall gyro bias was 11.49 (°)/h. After compensation the mean overall gyro bias becomes 0.11 (°)/h. The compensation effect is obvious, and the precision is greatly improved by one order of magnitude.

From the results of compensation, we see obvious residual temperature bias, which cannot be fully eliminated by the compensation method. So in future research work, we will focus on analyzing the influence factors of residual temperature bias, with more factors, and a more detailed model of the temperature bias.

[1] Cheng Jian-hua, Cheng Li, Li Ming-yue. Design and realization of precision temperature control system for shipborne FOG[J]. Journal of Chinese Inertial Technology, 2011, 19(4): 403-407.程建华, 陈李, 李明月. 船用光纤陀螺精密温控系统的设计与实现[J]. 中国惯性技术学报, 2011, 19(4): 403-407.

[2] Sanders G A. Progress in high performance fiber optic gyroscopes[C]//The 12th International Conference on Optical Fiber Sensors. Virgina, USA, 1997: 116-120.

[3] Heckman D W, Baretela M. Interferometric fiber optic gyro technology (IFOG)[J]. IEEE Aerospace and Electronic Systems Magazine, 2000, 15(12): 23-28.

[4] Hays K M, Schmidt R G, Wilson W A, et al. A submarine navigator for the 21st century[C]//Position Location and Navigation Symposium, 2002: 179-188.

[5] Sanders S J, Strandjord L K, Mead D. Fiber optic gyro technology trends - a Honeywell perspective[C]//Optical Fiber Sensors Conference Technical Digest, 2002, Vol.1: 5-8.

[6] Yang Meng-xing, Chen Jun-jie. Analysis and experiment on static temperature characteristic of FOG[J]. Journal of Chinese Inertial Technology, 2011, 18(6): 751-755.杨孟兴, 陈俊杰. 光纤陀螺静态温度特性的分析及实验研究[J]. 中国惯性技术学报, 2011, 18(6): 751-755.

[7] Han Bing, Lin Yu-rong, Deng Zheng-long. Overview on modeling and compensation of FOG temperature drift[J]. Journal of Chinese Inertial Technology, 2009, 17(2): 218-224.韩冰, 林玉荣, 邓正隆. 光纤陀螺温度漂移误差的建模与补偿综述[J]. 中国惯性技术学报, 2009, 17(2): 218-224.

[8] Shupe D M. Thermally induced nonreciprocity in the fiberoptic interferometer[J]. Appl. Opt., 1980, 19(5): 654-655.

[9] Lofts C M, Ruffin P B, Parker M D, et al. Investigation of the effects of temporal thermal gradients in fiber optic gyro scope sensing coils - Part 2[J]. Opt. Eng, 1993, 36(1): 29-34.

[10] El-Sheimy N, Hou H Y, Niu X J. Analysis and modeling of inertial sensors using Allan variance[J]. IEEE Transactions on Instrumentation and Measurement, 2008, 57(1): 148-149.

[11] Fattah S A, Zhu W P, Ahmad M O. An algorithm for ARMA model parameter estimation from noisy observations[C]//IEEE Circuits and Systems Symposium. Seattle, USA, 2008: 3202-3205.

1005-6734(2016)02-0242-03

一种光纤陀螺温度误差分段补偿新方法

傅 军1，江 赛2，覃方君1，冯卡力1

（1. 海军工程大学 电气工程学院，武汉 430033；2. 海军工程大学科研部，武汉 430033）

分析了光纤陀螺的温度特性，设计了大范围的温度测试，研究了不同温度和温度变化率对光纤陀螺输出的影响，研究了光纤陀螺在不同温度范围内的温度特性。为了提高温度误差补偿精度，根据陀螺温度特性将温度分为低、中、高三个区间，分别利用人工神经网络进行误差建模，提出了一种多模型分段拟合的新方法。根据建立的模型进行温度误差补偿，补偿结果表明，建立的模型能有效地减小了光纤陀螺的温度漂移，精度提高了一个量级。

光纤陀螺；温度漂移；多模型；人工神经网络

A

2015-12-23

国家自然科学基金（41404002）；总装预研基金（14JB11378）

傅军（1975—），男，讲师，从事惯性导航技术及应用、组合导航研究。E-mail: fjsd@21cn.com

10.13695/j.cnki.12-1222/o3.2016.02.019