基于FBG加速度传感器的内燃机典型故障识别系统与方法研究

发布时间：2018-06-06 20:38

本文选题：光纤光栅加速度传感器 + 内燃机健康监测　；参考：《山东大学》2017年博士论文

【摘要】：内燃机结构健康监测与不解体故障识别技术是确保内燃机及以其为动力源的大型机械装备安全稳定运行的关键保障之一。内燃机表面振动信号中包含了丰富的状态信息,因此,基于振动分析技术的内燃机结构健康监测引起了国内外学者的广泛关注,并取得了一定的研究成果,但仍然面临一系列挑战,诸如传统电类振动传感器受内燃机结构和工作环境影响,所获取的信号表现为非平稳、非线性特性,难以直接作为内燃机故障识别的依据;内燃机复合故障识别方法尚处于初步探索阶段,远未成熟等。针对以上问题,本课题以内燃机典型故障为研究对象,通过FBG加速度传感器设计、解调系统构建、信号处理、特征提取、故障识别方法等研究,建立高可靠性的内燃机典型故障识别模型,实现内燃机典型故障准确识别。具体研究内容阐述如下:(1)针对现有FBG加速度传感器灵敏度较低、平坦响应区窄和抗交叉干扰能力差的问题,在理论分析其传感机理的基础上,提出了基于柔性铰链结构的高灵敏度FBG加速度传感模型,利用ANSYS软件仿真分析了其响应特性,优化传感器结构,研制了基于柔性铰链结构的传感器,进而利用可调谐F-P滤波器,构建了基于边缘滤波原理的高频解调系统。(2)提出了一种基于多分类支持向量机的内燃机气门故障识别方法。在分析气门状态监测FBG传感器优化布置的基础上,构建了内燃机气门运行状态监测系统,实现了气门故障信号的准确检测。利用小波分解与重构算法和振荡能量提取气门故障信号特征,建立了信号特征与气门故障之间的关系模型,利用具有良好分类精度的支持向量机算法,在少量模型训练样本条件下,实现了内燃机气门故障的高准确率识别。(3)提出了一种基于变分模态分解算法和多分类支持向量机的内燃机喷油提前角故障识别方法。利用变分模态分解算法和振荡能量提取喷油提前角故障信号特征,建立了信号特征与喷油提前角故障之间的关系模型,利用具有良好分类精度的支持向量机算法,在少量模型训练样本条件下,实现了内燃机喷油提前角的高准确率故障识别。(4)提出了一种基于扩散映射算法和多分类支持向量机的内燃机复合故障识别方法。利用小波分解与重构和VMD算法提取内燃机复合故障信号特征,提出了基于扩散映射算法的特征向量降维方法,对特征向量进行降维,去除冗余信息,建立了信号特征与复合故障之间的关系模型,利用具有良好分类精度的支持向量机算法,在少量模型训练样本条件下,实现了内燃机气门和喷油提前角复合故障的高准确率识别。
[Abstract]:The technology of structural health monitoring and failure identification of internal combustion engine is one of the key guarantees to ensure the safe and stable operation of internal combustion engine and its large mechanical equipment. The vibration signal on the surface of internal combustion engine contains abundant state information. Therefore, the structural health monitoring of internal combustion engine based on vibration analysis technology has attracted wide attention from scholars at home and abroad, and some research results have been obtained. However, it is still facing a series of challenges, such as the traditional electric vibration sensor is affected by the internal combustion engine structure and working environment, and the signals obtained are non-stationary and nonlinear, so it is difficult to directly serve as the basis of internal combustion engine fault identification. The complex fault identification method of internal combustion engine is still in the initial stage of exploration, far from mature and so on. In view of the above problems, this paper takes typical internal combustion engine faults as the research object, through the FBG acceleration sensor design, demodulation system construction, signal processing, feature extraction, fault identification methods and so on. The high reliability typical fault identification model of internal combustion engine is established to realize the accurate identification of typical internal combustion engine faults. The specific research contents are as follows: (1) aiming at the problems of low sensitivity, narrow flat response area and poor ability to resist cross-interference of the existing FBG acceleration sensors, based on the theoretical analysis of its sensing mechanism, A high sensitivity FBG acceleration sensing model based on flexure hinge structure is proposed. The response characteristics of FBG are simulated and analyzed by ANSYS software, the sensor structure is optimized, the sensor based on flexure hinge structure is developed, and the tunable F-P filter is used. In this paper, a high frequency demodulation system based on edge filtering principle is constructed. A multi-class support vector machine based valve fault identification method for internal combustion engine is proposed. On the basis of analyzing the optimal arrangement of FBG sensors for valve state monitoring, a valve operating condition monitoring system of internal combustion engine is constructed, which realizes the accurate detection of valve fault signal. Wavelet decomposition and reconstruction algorithm and oscillatory energy are used to extract the valve fault signal feature, and the relationship model between the signal feature and valve fault is established, and the support vector machine algorithm with good classification accuracy is used. Under the condition of a small number of model training samples, the high accuracy identification of internal combustion engine valve faults is realized. A new method based on variational mode decomposition algorithm and multi-classification support vector machine is proposed to identify the fuel injection advance angle fault of internal combustion engine. Using variational mode decomposition algorithm and oscillation energy to extract the fault signal features of fuel injection advance angle, the relationship model between the signal characteristics and the injection advance angle fault is established, and the support vector machine algorithm with good classification accuracy is used. Under the condition of a small number of model training samples, the high accuracy fault identification of the injection advance angle of internal combustion engine is realized. A new method of internal combustion engine complex fault identification based on diffusion mapping algorithm and multi-classification support vector machine is proposed. Wavelet decomposition and reconstruction and VMD algorithm are used to extract the features of internal combustion engine complex fault signals. A method of dimension reduction based on diffusion mapping algorithm is proposed to reduce the dimension of feature vectors and remove redundant information. The relationship model between signal features and complex faults is established. The support vector machine (SVM) algorithm with good classification accuracy is used to train samples with a small number of models. The high accuracy recognition of the composite fault of valve and injection advance angle of internal combustion engine is realized.
【学位授予单位】：山东大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TK407;TP212

【相似文献】