Research On Vehicle Comfort Evaluation Methods Based On Neural Network

Vehicle ride comfort is a measure of the overall performance of automotive, one of the most important indicators, which means the passenger motor vehicle to be driven to adapt to the level of vibration. With the automobile industry is becoming more competitive and the increase of vehicles performance,the vehicle ride comfort needed accurate forecasting and evaluation. To shorten the design cycle and reduce production costs, it become an important factor in the market where a car manufacturer to win the competition .Therefore, the research of the vehicle ride comfort and its evaluation methods is very important.Study on Vehicle Ride Comfort started in half a century ago, to adopt a large number of vibration test and Ergonomics research and summed up a number of evaluation methods and criteria.60s of the last century, the International Organization for Standardization (ISO) began development of "human subject guidelines for the evaluation of whole body vibration", after several changes and additions, ISO announced the ISO2631-1: 1997 (E) "Evaluation of the human body to bear the whole body vibration -- Part I: General requirements". In the evaluation of comfort, it takes three contacts with the seat position and a total of 12 axial vibrations. According to the center frequency weighting coefficient of the axis and the axial weighting coefficient, the total root-mean-square value is calculated form 12 of axial acceleration .Finally, reference to total value of acceleration with the form of subjective feeling, to draw people's subjective feelings.However, in recent years, studies have found that when the human body in a random vibration, even if the weighted RMS acceleration calculated in accordance with the ISO2631 standard, and if the spectrum of two types vibration is inconsistent, comfort do not feel the same. The consistency of subjective evaluation of ride comfort with the objective measurement not only has been the focus of the car at home and abroad scholars and testing personnel, but also the issue of this article needs to explore.Artificial is a multi-disciplinary forefront technology which developed recently, and which simulate the human brain structure and smart features, and whose problem-solving are not required to accurately model, only requires a large amount of raw data. Vibration comfort is subjective in nature with a fuzzy feeling. There are great difficulties in using traditional mathematical model to describe, but neural network has a huge advantage in this respect. In this paper, the basic theory of the neural network technology and improved BP algorithm are applied to the evaluation of the vehicle comfort ,combining the objective vibration and the subjective feeling of human. It provides a new thinking of evaluation of the vehicle comfort. According to the principle of neural network modeling, the actual vehicle vibration signals are processed as a model input, the actual feeling as the model output to establish .Using the actual data of vibration comfort train the neural network model and verify the generalization ability of the model. The predict of models and actual results are generally consistent, it can able to establish mapping relations between objective vibration quantity and the actual subjective feeling of the staff.By research the basic working process of the system, using the neural network model which is established before, we have developed a neural network-based evaluation system of vibration comfort by using LabVIEW and Matlab programming. We carry out a simple design on the system hardware resources needed, including the acceleration sensor, the choice and application of data acquisition cards. The software progamming based on the idea of LabVIEW modular design, it describes the realization process of the function modules in detail. Through the system, you can save a great deal of experimental and professional evaluation of the subjective evaluation of professional evaluation staff, and will be able to get a vehicle's value of vibration comfort which is combined objective and subjective succinctly.