A novel end-of-line acoustic–vibration-based testing tool for quality assurance and supplier control of gasoline direct injection fuel pump

Gasoline Direct Injection (GDI) systems offer enhanced fuel efficiency but have introduced significant noise challenges during high-pressure fuel pump operation. This study presents a novel, non-invasive end-of-line (EOL) testing methodology combining acoustic and vibration data to identify anomalously noisy pumps. Using cross-power spectral density analysis and a confidence interval-based statistical framework, the method enables precise detection of fuel pumps emitting above 60 dBA SPL. A small sample size, derived through power analysis, was used to ensure efficiency without compromising statistical validity. Experimental testing on Best-of-Best (BOB) and Worst-of-Worst (WOW) pump categories demonstrated reliable fault classification with over 95% confidence. The system’s integration of laser vibrometer and microphone data allows robust real-time testing, improves manufacturing consistency, and reduces quality control overhead. This approach contributes to advanced diagnostics and supplier control in GDI fuel systems. • Non-invasive EOL testing using combined acoustic and vibration signatures • CPSD-based order analysis isolates correlated pump-borne noise • Statistical thresholding enables automated pass/fail classification • Method improves robustness over conventional SPL-based EOL tests