The dynamic response of gear transmissions holds essential information for the recognition of faults in the system. A realistic nonlinear dynamic model is used to predict the vibration regime of spur gear transmissions. This model was validated experimentally for both healthy and damaged conditions. The model allows to simulate each combination of rotational speed, load and surface roughness, and to isolate its influence on the dynamic response. An experimental study would be limited due to the large amount of cases to consider. The simulated surface roughness enables a more realistic evaluation of the condition indicators robustness. The first part of the study examines the effects of the operating conditions, e.g. speed, load and surface roughness, on the vibration signature of a healthy gearbox. It was found that the levels at the gear mesh frequencies (GMF) have a strong dependency on speed and load. The energy of the frequency modulation (FM) sidebands depends mainly on speed but hardly affected by load. The second part of the study examines the manifestation of a single tooth face fault in the vibration signature. Based on the conclusions of the first part, the analysis was focused on the synchronous average spectrum and the difference signal. The investigation using hundreds of simulations shed a new light on the physical phenomena in the dynamic response - the fault’s expression is amplified around the natural frequencies of the gears. Based on this insight, it was found that the analysis of the spectrum around the natural frequencies allows the detection of incipient faults that cannot be detected using the analysis of the difference signal. The understanding of the physical phenomena, based on the dynamic model, enabled the generalization of this new finding.

齿轮传动的动态响应具有识别系统故障的基本信息。实际的非线性动力学模型用于预测正齿轮传动的振动状态。该模型已针对健康和受损条件进行了实验验证。该模型允许模拟转速，负载和表面粗糙度的每种组合，并隔离其对动态响应的影响。由于需要考虑大量案例，因此实验研究将受到限制。通过模拟的表面粗糙度，可以更实际地评估状态指示器的坚固性。研究的第一部分研究了运行状况（例如速度，负载和表面粗糙度）对健康变速箱振动信号的影响。已经发现，齿轮啮合频率（GMF）的电平与速度和负载有很大的相关性。调频（FM）边带的能量主要取决于速度，但几乎不受负载影响。研究的第二部分检查了振动特征中单齿面缺陷的表现。基于第一部分的结论，分析主要集中在同步平均频谱和差分信号上。使用数百个模拟进行的调查揭示了动态响应中的物理现象-故障的表达在齿轮的固有频率附近被放大。基于这种见解，已经发现，通过对固有频率周围的频谱进行分析，可以检测出使用差信号分析无法检测到的初期故障。基于动态模型对物理现象的理解，使这一新发现得以推广。