无人机传感器故障诊断方法研究

发布时间：2018-11-11 17:33

【摘要】：小型无人机因其具有成本低、风险可控、机动性高等特点被广泛应用在商业和军事领域。在无人机系统中布设着数量众多的传感器,如垂直陀螺、角速率传感器、加速度计等。无人机平台上的传感器工作环境特殊,诱发故障的因素率较多。传感器一旦发生故障或不稳定,严重时可能导致无人机失控坠毁。因此,开展无人机传感器故障诊断研究具有重要的应用价值。本文以国产某小型无人机为研究对象,以典型机载传感器的故障诊断实用方法为研究目标,旨在提出一种诊断精确度高、泛化能力强的故障诊断方法。首先,总结了无人机传感器故障诊断技术的国内外研究现状,并且对典型机载无人机传感器作了简要介绍和故障分析。以某型无人机科研试验历史数据为基础,针对无人机传感器,研究基于模式识别的故障诊断方法。然后,将小波分析应用于特征提取方法中。仿真实现了小波包系数特征提取方法和小波包能量特征提取方法。并针对其不足做出改进,提出一种小波包复合特征提取方法。实验证明,该方法明显改善了算法性能,提高了特征向量的可分性。接着,研究基于决策树的分类诊断方法。采用ID3算法和CART算法构建分类模型,实现了对无人机传感器故障信号的分类识别。为提高故障诊断精度,引入梯度提升决策树(GBDT)算法,通过对弱分类模型的迭代与组合,构成诊断精度高的强分类模型。经参数调优后,算法性能得到进一步的提升。最后,基于上述研究成果,提出一种基于小波与GBDT的无人机传感器故障诊断方法。设计故障诊断验证平台,以无人机传感器地面测试模块与科研历史数据作为测试样本,对其进行仿真验证。实验结果表明,该方法具有诊断精确度高和泛化能力强的性能优势。
[Abstract]:Small UAVs are widely used in commercial and military fields because of their low cost, controllable risks and high mobility. There are many sensors in UAV system, such as vertical gyroscope, angular rate sensor, accelerometer and so on. The sensor working environment on UAV platform is special, and the factor rate of inducing malfunction is many. If the sensor fails or is unstable, it can cause the UAV to crash out of control. Therefore, the research of UAV sensor fault diagnosis has important application value. In this paper, a small UAV made in China is taken as the research object and the practical method of fault diagnosis of typical airborne sensors is taken as the research object. The purpose of this paper is to put forward a fault diagnosis method with high diagnostic accuracy and strong generalization ability. Firstly, the research status of UAV sensor fault diagnosis technology at home and abroad is summarized, and the typical airborne UAV sensor is briefly introduced and analyzed. A fault diagnosis method based on pattern recognition is studied for UAV sensors based on the historical data of scientific research and test of a certain type of UAV. Then, wavelet analysis is applied to feature extraction. The methods of wavelet packet coefficient feature extraction and wavelet packet energy feature extraction are realized by simulation. In order to improve the performance of wavelet packet, a new method of wavelet packet composite feature extraction is proposed. Experimental results show that the proposed method improves the performance of the algorithm and improves the separability of the eigenvector. Then, the classification and diagnosis method based on decision tree is studied. The classification model is constructed by using ID3 algorithm and CART algorithm to realize the classification and recognition of UAV sensor fault signals. In order to improve the accuracy of fault diagnosis, the gradient lifting decision tree (GBDT) algorithm is introduced. The strong classification model with high diagnostic accuracy is constructed by iterating and combining the weak classification model. After parameter tuning, the performance of the algorithm is further improved. Finally, based on the above research results, a fault diagnosis method for UAV sensors based on wavelet and GBDT is proposed. The fault diagnosis and verification platform is designed, and the UAV sensor ground test module and scientific research history data are used as test samples to simulate and verify it. The experimental results show that this method has the advantages of high diagnostic accuracy and strong generalization ability.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：V267;V279;TP212

【参考文献】