内燃机噪声源时域盲识别技术研究

发布时间：2018-06-16 04:51

本文选题：内燃机 + 噪声源识别　；参考：《吉林大学》2014年硕士论文

【摘要】：NVH(Noise Vibration and Harshness)水平是衡量机动车辆性能的重要指标之一，NVH直接反应机动车辆舒适性品质，而机动车辆舒适性程度直接影响该车辆的市场竞争力。车辆的噪声和振动特性的研究和控制重点在于对内燃机噪声进行控制，而噪声源识别对内燃机噪声控制起重要的作用。要想合理的控制内燃机噪声，应该对内燃机的振动噪声产生原因进行研究，，即对振动噪声的产生部位进行定位，准确测量和分析振源的特性，如振源的类型、频率特性、声压级大小、变化和传播规律等。然后根据振动噪声源的分析结果，采取相应的措施，降低内燃机的辐射声压级。近三十年来，内燃机科技工作者通过不懈的努力，使其在低噪声结构优化设计、低噪声燃烧系统开发、内燃机噪声分离以及阻尼降噪技术等领域均有多项技术突破，有关资料显示，与三十年前噪声整体辐射水平相对比，其总体噪声辐射水平下降了12dB。即便如此，内燃机噪声辐射水平仍有下降的空间和必要性。由于内燃机属于复杂机械系统，当内燃机系统运行时，其内部多个零部件在同时运转，这使待分析的目标部件响应即使可以直接测量也受到非目标零部件振动信号的影响，在频谱中多个振源信号往往存在重叠，再加上实验手段也十分复杂，实验成本高，这使得传统的内燃机噪声源识别技术具有一定的局限性。随着新理论的提出以及新技术的应用，采用现代信号处理方法对内燃机的噪声和振动特性的研究有着重要的理论意义和工程实用价值。针对上述相关问题，本文开展了相应的研究工作：首先，阐述了课题研究背景，介绍柴油机振声特性的危害性，以及当前柴油机振动信号处理过程中存在的技术问题。详尽介绍了盲分离算法发展过程以及研究现状，内燃机振动信号现代处理方法，并指出应用时域盲识别算法对内燃机噪声源识别理论意义及工程价值；其次，利用连续小波变换对内燃机表面振动信号进行时-频分析，通过对内燃机各个部位表面加速度振动信号的特征的分解，对内燃机在运转过程中的主要噪声源进行识别，为内燃机的噪声源的识别研究提供一些方法和途径；再次，建立了源信号线性瞬时混合模型和源信号线性卷积混合模型，介绍了有关信息论的基本概念，介绍了线性瞬时盲分离问题的常用解混算法，并详细的介绍了FastICA算法的流程，通过数值算例验证了FastICA算法的高效性和鲁棒性。同时，介绍线性卷积盲分离问题的常用解混算法，详尽介绍了MBLMS算法流程，并通过数值算例验证MBLMS算法的高效性和鲁棒性。最后，应用FastICA算法以及改进的MBLMS算法对柴油机表面混合信号进行分离，通过分离结果比较验证了改进的MBLMS算法的高效性和可靠性，也进一步对内燃机非平稳振声响应结构模型进行了验证。
[Abstract]:The level of NVH noise Vibration and Harshnessis is one of the important indexes to measure the performance of motor vehicles, which directly reflects the comfort quality of motor vehicles, and the degree of comfort of motor vehicles directly affects the market competitiveness of motor vehicles. The research and control of vehicle noise and vibration characteristics is focused on the control of internal combustion engine noise, and noise source identification plays an important role in internal combustion engine noise control. In order to control the internal combustion engine noise reasonably, it is necessary to study the cause of the vibration noise, that is, to locate the position of the vibration noise, to measure and analyze the characteristics of the vibration source, such as the type of the vibration source and the frequency characteristic. Sound pressure level, variation and propagation law, etc. Then according to the analysis results of vibration and noise source, the corresponding measures are taken to reduce the sound pressure level of internal combustion engine. In the past 30 years, the scientific and technological workers of internal combustion engine have made many technical breakthroughs in the fields of low noise structure optimization design, low noise combustion system development, internal combustion engine noise separation and damping noise reduction technology through unremitting efforts. The relative data show that compared with the total noise radiation level of 30 years ago, the overall noise radiation level has decreased by 12 dB. Even so, the internal combustion engine noise radiation level still has the space and the necessity. Since the internal combustion engine belongs to a complex mechanical system, when the internal parts of the internal combustion engine system are running at the same time, the response of the target component to be analyzed is affected by the vibration signal of the non-target component even if it can be measured directly. There is often overlap in the spectrum of multiple vibration source signals, in addition, the experimental means are also very complex and the experimental cost is high, which makes the traditional identification technology of internal combustion engine noise source has some limitations. With the development of new theory and the application of new technology, it is of great theoretical significance and practical value to study the noise and vibration characteristics of internal combustion engine by using modern signal processing method. In view of the above related problems, this paper has carried out the corresponding research work: first, elaborated the research background, introduced the diesel engine vibration sound characteristic harmfulness, And the technical problems existing in the process of diesel engine vibration signal processing. The development and research status of blind separation algorithm and the modern processing method of internal combustion engine vibration signal are introduced in detail, and the theoretical significance and engineering value of using time domain blind recognition algorithm to identify internal combustion engine noise source are pointed out. The time-frequency analysis of internal combustion engine surface vibration signal is carried out by using continuous wavelet transform, and the main noise sources of internal combustion engine during operation are identified by decomposing the characteristics of the acceleration vibration signal on each part of the internal combustion engine. It provides some methods and approaches for the identification of internal combustion engine noise sources. Thirdly, a linear instantaneous mixing model of source signals and a mixed model of linear convolution of source signals are established, and the basic concepts of information theory are introduced. In this paper, the common unmixing algorithms for linear instantaneous blind separation are introduced, and the flow chart of FastICA algorithm is introduced in detail. The efficiency and robustness of FastICA algorithm are verified by numerical examples. At the same time, the common de-mixing algorithms for linear convolution blind separation problem are introduced, and the flow chart of MBLMS algorithm is introduced in detail, and the efficiency and robustness of MBLMS algorithm are verified by numerical examples. Finally, the FastICA algorithm and the improved MBLMS algorithm are used to separate the diesel engine surface mixed signals. The efficiency and reliability of the improved MBLMS algorithm are verified by comparing the separation results. The structural model of non-stationary vibration response of internal combustion engine is also verified.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TB535

【参考文献】