Uma proposta de ferramenta para controle de qualidade em linha de produção e montagem de câmbios automotivos utilizando filtragem adaptativa

Abstract

Normally, automotive gearbox quality control in assembly line is made in subjective manner by specialists. The human ear, despite being a highly sensitive tool and widely used in quality control is also sensitive to emotional and environmental problems (back noises), besides not giving measurable values. In order to solve this problem, the traditional techniques of signal analysis, along with the monitoring of vibrations and noise of the tested gearbox, may be used to define objective values of acceptance or refusal in a quality program. But, due to the complexity of the dynamics of a automotive gearbox and to the noises inherent of a production line, it s difficult to define the control systems and determinate the limit values. In this work, it is introduced a study about the use of adaptive filtering through the least mean square algorithm (LMS), as a support tool in the procedures of quality control by vibrations and noise monitoring. The methodology is based on the use of the signals, measured in a standardized test, as a former signal and of reference signal in the filtering procedure. Initially, numeric simulations were done, in order to evaluate the performance of the LMS algorithm in cancellation of external interferences of the mechanical system under study. In the next step, acceleration and sound pressure signals were collected of approved gearbox in the actual proceedures of quality control (good gearbox). Another distinct group of gearbox signals, in which were inserted ordinary manufacturing and assembly defects, were divided in the groups of acceptable gearbox and bad gearbox. From the filtered signals of the good gearbox group, statistic techniques were used in order to define the limiting vibroacoustic symptoms for quality control. Finally, the procedure was validated using the signals from acceptable gearbox and bad gearbox groups.