Experiment Research And Mechanism Analysis On Relation Between Car Transmission Gear Noise And Manufacturing Errors

Car noise, an important index to measure the quality of the car, is one of the resources of environmental noise pollutions. Gearbox is one of the main car noise sources and therefore the control of car gearbox noise is one of the important subjects in the world car industry. Gear pairs as one of the transmission components, the mechanism of its engagement noise is a very complex problem. At present no mathematical model can manifest it perfectly and accurately and the present level of gear manufacturing can not avoid manufacturing errors. The study of gear tooth is the key to reduce gear noise, therefore, the study of gearbox errors is of great importance to the gearbox noise system. To identify and control manufacturing errors, which have the biggest effect on noise is an important way to reduce gearbox noise.1. Experimental study on the relationship between gear manufacturing errors and gear noiseGear manufacturing errors have direct effect on gear noise, but different kinds of gear manufacturing errors affect gear noise to different degrees. This paper makes experimental studies of different gear manufacturing errors, such as profile slope deviation, helix slope deviation, profile form deviation, helix form deviation, single pitch deviation, radial jumped error, base pitch error, total profile deviation, total helix deviation, pitch accumulated deviation. The Experiment Program is established by Means of the Orthogonal Test. On the gear noise test bench, gears of different manufacturing errors are tested and the curve of the relationship between single manufacturing error and gear noise is obtained. And theoretical explanation is provided to the relationship.2. Establishing a model for gears noise prediction based on gear manufacturing errorsBased on the analysis of effects of single gear manufacturing error on gear noise, a multiple linear regression model is developed showing the relationship between the gear manufacturing errors and transmission gear noise. Ridge regression is used to eliminate the problem of colinearity. In the established fridge regression model for gear noise prediction, the coefficient of each parameter reflects the impact of the deviation on the noise in the prediction model. Profile slope deviation and helix slope deviation have the biggest influence on gear noise; Profile form deviation and helix form deviation have bigger influence; base pitch error and pitch accumulated deviation have less influence; adjacent pitch error and runout error have the least influence. These conclusions are in accordance with the experimental results.The prediction model for the gear noise was established based on the principal component analysis (PCA) and BP neural network method. Using the principal components of original variables as the input of network can cut down the dimensions of input, eliminate the colinearity between variables, greatly simplify neural network structure and reduce the difficulty of neural network training. These conclusions are in accordance with the experimental results. The results indicate that the model has better predictive capability. This model can be used in production to optimize gear tooth micro-geometrical errors caused by machining to realize the control on gear noise.The prediction model is used to predict gear noise. The comparison between predicted results and experimental results show that the prediction value is close to the actual value. The accuracy of the model was verified. It means that the model can truly and correctly reflect the value of gear noise.3. Experimental study on the gearbox noise of car based on order analyzing methodGearbox is one of the main car noise sources and the control of its noise is increasingly important. In this paper, Rotas noise analysis system, developed by German DISCOM company, is used to test the gearbox noise in the car and investigate the contribution of gearbox noise to the whole car noise under the moving condition. The relation between sound press level and rotation speed has been obtained. Order analyzing method is used to identify the correlation between the whole car noise and gearbox noise. This method has the capacity to identify whole car noise source,and has broad prospects for application.Most systems can measure the noise difference between a very bad gearbox and a good gearbox, and it is not difficult. The real challenge is to distinguish the gearboxes that are just acceptable from those that are just unacceptable. In fact many gear boxes are close to the acceptable threshold. To set the acceptable threshold on the test bench, the best method is to establish good correlation between test rig measurements and in-vehicle noise. 4. The experimental study on gear noise reductionThis paper makes an experimental study of the effect of different machining methods (Shaving, grinding, honing,) on gear noise. Grinding can improve the macro geometric accuracy, but its micro surface quality is not satisfactory enough. Its meshing noise is higher than that of honing. Shaving is mostly used in low-torque manual gear box. It can realize tooth dressing and take compensation measures by means of heat treatment to improve greatly the gear accuracy and surface roughness,thus to reduce gear transmission noise. In addition, this paper also makes an experimental study of gear structure parameters. Such as modification coefficient,tooth addendum coefficient. The experiments show that the bigger tooth addendum coefficient, the bigger the contact ratio, the smoother the transmission is which has obvious effect. The change of modification coefficient can reduce gear noise without increasing cost.This paper provides new thinking, new approach and new lightening for the research and control of gear noise.