Research On Automobile Gap Detection Curve Identification Based On Cluster Method

Automotive has become the major means of transport in the people's production and daily life. Along with the increase in vehicle mileage, the technical situation of the automotive assembly and components will gradually deteriorate. Therefore, in order to assure the safe operation of vehicles, extend the life of the vehicle, and enhance transport capacity, it has became extremely important that detecting the comprehensive performance of the vehicles in use. Under the situation of the auto's non break up, the technique condition and the usage function of the organization,system and parts can be detected. Find the failure and the hidden danger.Driving department is an important component of the vehicle system. In the process of the use of vehicles, the gap among the parts of the wheel and shaft and suspension will be increased gradually, and the gap will lead the connecting parts to loose or translocation. It would serious impact on the vehicle handling and stability, driving safety and working life.Currently, the detection of the item is still in the way of manual detection. The test equipment developed by Jilin University can intelligently detect the gap among the wheel and shaft and suspension. Testing process, measurement and control system measured tire force and the displacement of plate. Based on the judgement of inflection point on the measured curve the algorithm achieves recognition of the value of the gap. However, there's not a good algorithm to judge the inflection point on the curve. The purpose of this paper is to find a reasonable and effective method of achieving sub-curve and extracting accurate clearance value. Cluster Analysis is an effective way for solving the problem.Clustering analysis is one of the main functions in Data Mining and Knowledge Discovery, which groups data sets into classes by nature and gives a character depiction for every class. Clustering is the things in accordance with certain attribute, things will be gathered into categories. The similarity of classes is as small as possible, within the category of similar as big as possible. Fisher Ordered Sample Cluster Analysis is a cluster analysis. This approach requires the sample according to a certain order, classification can not disrupt the order that the same type of samples must be mutually adjacent. This approach use a deviation square as diameter, classification made by the various types of square deviation to a minimum. Therefore, this paper will focus on the application of the Fisher ordered samples cluster in detecting the gap among the wheel and shaft and suspension, in order to achieve accurate identification of the gap.In the practical application, the effect of the recognition of this gap is not satisfactory in using the original Fisher ordered samples cluster analysis. This is because the square deviation in the algorithm is the dispersion of the measurements from the mean of the samples. But for this paper, the curve is a trend identified in the orderly samples, the mean of samples is no practical significance, and the trend of samples is hided. Therefore, this paper introduces a line fitting method for optimally partitioning an ordered sample. The basic idea of this method is to sample indicators and the corresponding sample points on the one-on-one as the data used piecewise linear function fitting these data, fitting accuracy through to the highest determine the optimal section.This approach considered the needs of practical problems in this paper, the application in detecting the gap among wheel and shaft and suspension, the results of the identification are good, and the value of the gap accurately extract. The car is tested to improve the classification of Fisher ordered samples to achieve a good recognition of clearance.The main constituents are as follows:1. A domestic and international developing situation of the auto inspection especially domestic status quo is introduced. The current detection of the gap among wheel and shaft and suspension has not been fully realized intelligent, this paper presents a new ideas that Fisher ordered sample cluster analysis applied to the gap detection.2. Through the vehicle suspension, steering system performance analysis, deduced the impact on vehicle handling and stability with the existence of gap. This paper studied the methods and principles of intelligent detection, the mechanical structure and data acquisition system of intelligent detection equipment, so as to provide theoretical and equipment support to achieve the goal of accurate extract gap value.3. Based on the practical problems of automobile wheel axle suspension gap recognition, this paper improve the original Fisher Ordered Sample Cluster Analysis, that is a line fitting method for optimally partitioning an ordered sample.Then this paper gives the specific structure and algorithm of the model. The model applied to the gap detection. Theoretical analysis and experimental results show that the algorithm is reasonable and effective. This method overcomes the defaults of the mutual influence in similar data and through the line fitting method to the exclusion of the impact of singular points. Achieve accurate extraction of the value of the gap, we have had good results.4. This paper analyses the software environment and the specific algorithm processes of the improvement Fisher ordered samples clustering algorithm. The algorithm realized through the S language and R software platform so as to achieve our purpose that identify the value of the sub-gap.5. Through the experimental to validate the feasibility and accuracy of algorithm. By comparing the experimental results and the actual parameters, this paper validates the accuracy of algorithm. The results show that the algorithm can be used to achieve the gap identification among wheel shaft and suspension. Recognition results are satisfactory.