Research On Shift Quality Polytechnic Assessment System

Shift quality is an important issue in the field of vehicle drivertrain system, especially for the vehicles equipped with automatic transmission. This is primarily due to two reasons: the first one is that the shift quality is a kind of performance which keeps a close relationship with the driver's driving feeling. So, shift quality actually builds a bridge between subjective rating and objective metrics. On the other hand, shift quality is also a complex problem due to its dynamic characteristics. Shift mechanism takes an important position in the powertrain system and often causes vehicle dynamic performance changes. This challenges researchers to do more work in this field. These two important reasons make the shift quality research become an increasingly topic considered by most automobile manufacturers and research institutes.On the other hand, the traditional research on shift quality pay more attention on the longitudinal shock which is important but only one aspect of shift quality effects. This paper improves the concept of the shift quality and expands the scope of the research. The effects on three different fields including longitudinal shock, noise and fuel economy and emission, have been studied deeply and some basic and useful knowledge were developed for the future research. Shift quality evaluation system based on metrics,was built cored with the algorithm of Support Vector Machines(SVMs). The qualified generic ability provides a guarantee to the system to acquire a good rating function.1. The algorithm research and applicationFor shift quality, the important aspect is to establish a relationship between the shift performance and subjective feeling. Because of its complex nonlinearity, traditional methods could not provide ideal results. In the past decades, some nonlinear methods such as multiple linear regressions, artificial neural network (ANN) and so on, have been used for study. ANN also used by author for estimating the shift quality but unreasonable results often happened because of ANN's generalization ability. Support Vector Machine is a statistical theory based on the small sample of theory. It maps the nonlinear problem in the lower dimensional space to high-dimensional space, thus by which low-dimensional nonlinear problem is transformed to high-dimensional linear problem. With SVMs, the author built an evaluation system. Responded from experiments and simulations, SVM has shown a good ability of prediction, and build a solid basis for discovering the relationship between subjective rating and objective performance.2. Shift Quality Evaluation StudyThis paper studies the effects of shifting action in three fields: longitudinal impact, shift noise, fuel economy and emissions. The longitudinal impact is the traditional areas studied, and set of evaluation criteria including jerk, acceleration. Based on vehicles equipped with AMT, the jerk and the acceleration were studied in-depth as well as the factor of speed. Generally, people believed that the speed of vehicles is not sensitive to the shift action. But tests showed that impact from jerk, acceleration and velocity together always contributes more uncomfortable feeling than they works respectively. To make the metrics more extensive to the shift quality, in this paper, metrics are provided with the actual operating conditions. This makes the course of the assessment calculations integrated the classification conditions and operating conditions, so the better genetic ability could be acquired in this way.Shift noise was defined as the noise changes triggered by shift action. As the shift engine noise will cause the noise change in time domain and the frequency domain. Based on the time-frequency analysis, both the noise and the relevant operating conditions are expressed in the shift noise chart. The subjective rating is used for make sure of the positions worst feeling happens. Research shows that noise amplitudes and frequency act on in different area which provides a clue to improve the noise quality by matching engine and transmission without changing engine structures. The paper build a model based on experiment data to discover the metrics for shift noise. The metrics includes noise rising time rate, falling time and rate, noise peak value and relevant frequency.An important research area shift quality works on is the fuel economy and emission. This paper studies the effects on the fuel economy and emissions caused by shift quality. The concept of shift fuel concept was built based on the general knowledge of fuel economy. To clear the relationship between shift quality and fuel economy, this paper considers two aspects: the disturbance of the shift schedule and transient response to the shift action. The economy matching map method was provided to find"economy area". The test based on the"ECE+EUDC"was implemented to research effect from shift quality, and metric of fuel consumption peak was used for donating this king of effect. The software of ADVISOR and experiments were used for analyzing the effects on emission. The effects caused by the different shift quality have been researched in the way of simulation. Based on comparison of different experiments and simulations, the value of emission of fuel consumption per100 kilometer was taken as a metric for evaluating the shift quality.3. Shift quality evaluation systemShift quality evaluation system was built based on the theory of Support Vector Machine. This paper introduces the two functions of the evaluation system: positive evaluation and reverse forecasts. The positive development is completed on the real vehicles tests. Different data were collected under the all kinds of operating conditions which consist of the basis of evaluation. As the part of research, shift quality integrated data acquirement and processing devices which made the system as a complete and in dependent tool. The evaluation system also works along with other hardware and so can be used widely.The evaluation system evaluation is calculated based on the experimental data. In order to enable the system to acquire better comparison capabilities, a proprietary data formats was created to meet the evaluation. Evaluation system can absorb the results as well as tests. The important component-sampling library enables the system the ability of self-modified and comparison based on the reference vehicle. The virtual environment was built for showing the shift performance directly and clearly. From the database corresponding reference vehicle, the virtual simulation will provide the real feeling, so the more correct evaluation would be acquired for the development.