Research On Online Automobile Transmission Intelligent Quality Detection System

As an important component of automotive, automotive transmission plays asignificant impact on the overall performance of the car. Its technical parameters andworking condition are important indicators to evaluate performance of the car. In theresearch areas of the automotive industry, the improving of the technical performanceand reliability of the transmission is always the topics at the forefront of theautomotive research. In order to ensure the quality of transmission and improve thedetection rate and accuracy of quality, it is necessary to establish an online intelligentquality detection system of automobile transmission.The paper analyzes the development process and status of modern detectiontechnology and transmission detection technology, the system makes application ofcomputer detection technology, spectrum analysis, the standard extremal curve, SVMand database technology to establish online intelligent quality detection system ofautomobile transmission based on the study and summary of the defect mechanism ofthe transmission. The system establishes a full-featured data acquisition hardwareplatform through the reasonable selection of sensors and data acquisition card. Thesoftware system is based on mixed of VC++and MATLAB, including five modules:parameter setting, signal acquisition, signal analysis, quality testing and database. Itachieves a synchronous acquisition of vibration signal and speed signal, whileachieves online real-time analysis of vibration signal and speed signal takingadvantage of the processing speed of MATLAB. On the basis of analysis of the basicstructure and working principle of the auto transmission, the system combinescleverly the technique of order analysis with cepstrum to extract the characteristicsignal for the final quality inspection, and the module of quality detection providestwo detection methods: standard extremal curve and support vector machine. Firstly,put the characteristic signals as the test samples into the support vector machinemodel to make the first quality detection of the current transmission. When the resultis no defect by the method of SVM, the system will make the second detection by the method of the standard extremal. This method compares the order spectrum with thestandard extremal, if does not exceed the standard extremal, update the standardextremal using the order spectrum; if exceeded, find out the order of the exceeded part,and obtain the types and location of defects by querying the database, then update thetraining samples database by the test samples to make the model contain more defectinformation.The system enriches the SVM training samples by the extremal to get amore complete model of support vector machine, so that can improve the accuracyand detection efficiency of transmission’s online pre-delivery inspection. This systemhas completed the development of database systems combining ACCESS, in order toachieve a variety of data processing functions, including the result of order spectrumanalysis, Access to time-domain characteristic value, the parameters of qualitydetection, modified and access to the parameters of order analysis, test resultscorresponding to the defect type and location. It makes full use of an ActiveX DataObject (ADO) interface to shorten the development cycle for database applicationsand improve the speed of data communication between the database and application.The system has been experimentally verified its feasibility, it has an importantpractical significance to improve the rapidity and accuracy of transmission’s onlinepre-delivery inspection.