E-mail Alert
RSS
FOG temperature drift compensation method based on wavelet denoising and neural network
-
摘要:
光纤陀螺(FOG)输出易受环境温度的影响，发生漂移导致光纤陀螺测量精度降低。采用传统的BP神经网络容易陷入局部极小值，导致网络训练失败。为了优化BP神经网络，本文提出了一种粒子群(PSO)优化BP神经网络与小波降噪相结合的光纤陀螺温度漂移补偿方法。首先分析了光纤陀螺温度漂移产生的原因；然后在不同温度下对光纤陀螺进行测试，最后采用该方法建立了光纤陀螺温度漂移模型并根据模型对光纤陀螺进行补偿，结果表明采用该方法补偿后光纤陀螺在不同温度下的输出标准差降低了60.19%，与传统的BP神经网络相比补偿效果显著提高。
Abstract:The output of fiber optic gyroscope (FOG) is easily affected by the temperature variations, so it leads to produce drift and the measurement accuracy of FOG is reduced. The traditional BP neural network is an optimization method of local search, which is easy to fall into local minimum, leading to the failure of network training. In order to optimize BP neural network, a temperature drift compensation method for FOG based on particle swarm optimization (PSO) and wavelet denoising is proposed. Firstly, the mechanism of FOG temperature drift is analyzed. Next, FOG static state test in different temperatures is finished. Finally, the FOG temperature drift model has been built by the method and compensate. The results show that the output standard deviation of FOG at different temperatures is reduced by 60.19%, and the compensation effect is better than traditional BP neural network.
-
Overview: Fiber optic gyroscope (FOG) is a new solid-state optoelectronic gyroscope based on Sagnac effect and it is widely used in servo control, flight control and inertial navigation. The output characteristics of the main components of FOG (such as optical fiber ring, light source, and photoelectric detector, etc.) are vulnerable to the influence of ambient temperature and self-heating. Ultimately, the output of FOG produces temperature drift, which is the comprehensive effect of temperature on the components of FOG. This drift greatly affects the measurement accuracy of FOG, so measures must be taken to suppress the temperature drift of FOG. Restrict by technology and cost, ameliorate the construction of FOG and winding craft of fiber coil can't overcome the influence by temperature completely. Building the temperature drift compensate model can restrain the temperature drift primely and unrestrictedly.
BP neural network is often used in temperature drift modeling of FOG, but the traditional BP neural network is an optimization method of local search. The weights and thresholds of the network are gradually adjusted along the direction of local improvement, which is easy to fall into local minimum, leading to the failure of network training. If the direction of maximum descent gradient is found on a larger scale and the connection weights and thresholds of each layer are adjusted, the local minimum can be avoided to a certain extent, and the fitting effect of the neural network can be improved. At the same time, the noise in the FOG signal will also cause disadvantage to the establishment of temperature drift model, so the signal must be filtered before the model is built.
Guided by the above ideas, a temperature drift compensation method for FOG based on particle swarm optimization BP neural network and wavelet denoising is proposed. Firstly, according to the operating principle of the FOG, the mechanism and the temperature characteristic of FOG temperature drift are analyzed and state the temperature characteristic of the FOG drift. Then, FOG temperature drift static state test within the limits of -40 ℃~60 ℃ is designed and record temperature in real time. The results show that temperature gradient will impact the FOG temperature drift. Next, using heuristic threshold filtering can reduce high frequency noise and eliminate abnormal change data. Using the filtered experimental data, the temperature drift model is established by optimizing BP neural network fitting with particle swarm optimization algorithm. The model can predict the temperature drift in different states. Finally, compensate results are verified by simulation experiments. The results show standard deviation of FOG outputs in different temperatures is descend by 60.19%, and the compensation effect is better than traditional BP neural network.
-
-
图 8 PSO-BP神经网络拟合、BP神经网络拟合和光纤陀螺原始数据
Figure 8. The data of PSO_BP neural network, BP neural network and primeval output
表 1 不同温度下光纤陀螺输出平均值和标准差
Table 1. Average and standard deviation of FOG in different temperatures
温度/(℃) 输出均值/(°/h) 输出方差/(°/h) 20 0.719659 0.028345 20~-40 0.646126 0.058365 -40 0.616971 0.023955 -40~60 0.757383 0.065029 60 0.662305 0.023636 60~20 0.662601 0.059026 全过程 0.689201 0.064910 
下载: 导出CSV
表 2 采用启发式阈值法滤波后的光纤陀螺输出平均值和标准差
Table 2. Average and standard deviation of FOG output by heuristic threshold filtering
温度/(℃) 输出均值/(°/h) 输出标准差/(°/h) 20 0.7197154 0.0183977 20~-40 0.6461299 0.0509969 -40 0.6170071 0.0084288 -40~60 0.7574017 0.0612815 60 0.6622794 0.0103533 60~20 0.6622794 0.0371164 全过程 0.6891953 0.0600332
下载: 导出CSV
表 3 补偿后的光纤陀螺输出平均值和标准差
Table 3. Average and standard deviation of FOG output after compensation
温度/(℃) BP神经网络补偿 PSO化BP神经网络补偿 输出均值/(°/h) 输出标准差/(°/h) 输出均值/(°/h) 输出标准差/(°/h) 20 0.0033827 0.0274135 -0.0022721 0.0242832 20~-40 -0.0063835 0.0377248 -0.0012272 0.0385154 -40 -0.0021987 0.0246365 0.0020123 0.0232413 -40~60 0.0008027 0.0292569 0.0007977 0.0248314 60 0.0079882 0.02841471 -0.0011717 0.0246126 60~20 -0.0032244 0.0304272 0.0001394 0.0239154 全过程 0.0004328 0.0411212 0.0002119 0.0258433
下载: 导出CSV
-
[1] 国防科学技术工业委员会.光纤陀螺仪测试方法: GJB 2426A-2004[S].北京: 国防科学技术工业委员会, 2004: 1–16.
National Defense Science, Technology and Industry Commission. Test methods for fiber optic gyroscope: GJB 2426A-2004[S]. Beijing: National Defense Science, Technology and Industry Commission, 2004: 1–16.
[2] 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. http://d.old.wanfangdata.com.cn/Periodical/zggxjsxb200902022
Han B, Lin Y R, Deng Z L. Overview on modeling and compensation of FOG temperature drift[J]. Journal of Chinese Inertial Technology, 2009, 17(2): 218–224. http://d.old.wanfangdata.com.cn/Periodical/zggxjsxb200902022
[3] 王威, 陈熙源.一种基于多尺度和改进支持向量机的光纤陀螺温度漂移建模与补偿方法[J].中国惯性技术学报, 2016, 24(6): 793–797. http://d.old.wanfangdata.com.cn/Periodical/zggxjsxb201606017
Wang W, Chen X Y. Modeling and compensation method of FOG temperature drift based on multi-scale and improved support vector machine[J]. Journal of Chinese Inertial Technology, 2016, 24(6): 793–797. http://d.old.wanfangdata.com.cn/Periodical/zggxjsxb201606017
[4] 冯卡力, 李安, 覃方君, 等.光纤陀螺温度误差自适应神经模糊补偿方法[J].兵工学报, 2016, 37(4): 641–647. doi: 10.3969/j.issn.1000-1093.2016.04.010
Feng K L, Li A, Qin F J, et al. Temperature error compensation method based on adaptive neuro fuzzy inference for fiber-optic gyro[J]. Acta Armamentarii, 2016, 37(4): 641–647. doi: 10.3969/j.issn.1000-1093.2016.04.010
[5] 吕辰刚, 张瑞峰, 武星, 等.光纤环的热致非互易性噪声理论与实验研究[J].传感技术学报, 2009, 22(6): 798–802. doi: 10.3969/j.issn.1004-1699.2009.06.008
Lv C G, Zhang R F, Wu X, et al. A theoretical and experimental study on thermal-induced non-reciprocity noise of fiber optical coil[J]. Chinese Journal of Sensors and Actuators, 2009, 22(6): 798–802. doi: 10.3969/j.issn.1004-1699.2009.06.008
[6] 郭士荦, 许江宁, 李峰, 等.优化BP神经网络的光纤陀螺温度漂移建模与补偿[J].中国惯性技术学报, 2016, 24(1): 93–97. http://d.old.wanfangdata.com.cn/Periodical/zggxjsxb201601017
Guo S L, Xu J N, Li F, et al. Modeling and compensation algorithm of FOG temperature drift with optimized BP neural network[J]. Journal of Chinese Inertial Technology, 2016, 24(1): 93–97. http://d.old.wanfangdata.com.cn/Periodical/zggxjsxb201601017
[7] 刘文涛, 刘洁瑜, 沈强.光纤陀螺随机误差的集成建模及滤波处理[J].光电工程, 2018, 45(10): 180082. doi: 10.12086/oee.2018.180082
Liu W T, Liu J Y, Shen Q. Integrated modeling and filtering of fiber optic gyroscope's random errors[J]. Opto-Electronic Engineering, 2018, 45(10): 180082. doi: 10.12086/oee.2018.180082
[8] 高伟, 祖悦, 王伟, 等.基于二代小波的光纤陀螺实时降噪方法研究[J].仪器仪表学报, 2012, 33(4): 774–780. doi: 10.3969/j.issn.0254-3087.2012.04.009
Gao W, Zu Y, Wang W, et al. Research on real-time de-noising of FOG based on second generation wavelet transform[J]. Chinese Journal of Scientific Instrument, 2012, 33(4): 774–780. doi: 10.3969/j.issn.0254-3087.2012.04.009
[9] 霍炬, 王石静, 杨明, 等.基于小波变换阈值法处理光纤陀螺信号噪声[J].中国惯性技术学报, 2008, 16(3): 343–347. http://d.old.wanfangdata.com.cn/Periodical/zggxjsxb200803019
Huo J, Wang S J, Yang M, et al. Noise processing of FOG signal based on wavelet threshold-value[J]. Journal of Chinese Inertial Technology, 2008, 16(3): 343–347. http://d.old.wanfangdata.com.cn/Periodical/zggxjsxb200803019
[10] 顾春雷, 陆金桂, 王怡祖, 等.基于GA－BP神经网络的光纤陀螺温度补偿[J].仪表技术与传感器, 2018(3): 113–116. doi: 10.3969/j.issn.1002-1841.2018.03.026
Gu C L, Lu J G, Wang Y Z, et al. Temperature compensation of fog based on GA－BP neural network[J]. Instrument Technique and Sensor, 2018(3): 113–116. doi: 10.3969/j.issn.1002-1841.2018.03.026
[11] Kennedy J, Eberhart R C. Particle swarm optimization[C]//Proceedings of ICNN'95 - International Conference on Neural Networks, 1995: 1942–1948.
[12] Hadeler T T, Montvale N J. Fiber optic gyro drift rate compensation based on temperature: US5416585[P]. 1995-05-16.
-
点击扫一扫
图(10)
表(3)
计量
- 文章访问数: 7498
- PDF下载数: 2327
- 施引文献: 0
