Estimation of brake pad wear and remaining useful life from fused sensor system, statistical data processing, and passenger car longitudinal dynamics

Abstract

The problem of estimating brake pad wear of cars has been considered, using friction work approximated from the product of longitudinal braking force and distance travelled during braking. The braking force is estimated from a longitudinal dynamics model with inputs from an inertial measurement sensing unit with Kalman filter based sensor fusion, and standard vehicle on-board diagnostics data. Wear coefficients are fitted with Bayesian linear regression on measurements from two series: (I) 16 measurements of previously worn pads over approx. 4,800 km with average wear of approx. -0.5 mm and (II) seven measurements of almost new pads over approx. 1,850 km and approx. -0.15 mm of wear. The approximated friction work appears to correlate well with the wear. The wear coefficients converge quickly after just 4-5 measurements. Coefficients ranging from −4.95⋅10−5mm3/J to −2.08⋅10−5mm3/J are obtained across the cases, when assuming a coefficient of friction μ=0.6 obtained through pin-on-disc (POD) tests. The wear coefficients are validated using POD tests, showing good agreements with the indirectly estimated wear coefficients from vehicle longitudinal dynamics. Data from series (II) is used for prediction of remaining useful life (RUL), where one additional measurement is carried out after 16,552 km. The RUL prediction requires knowledge of the relationship between usage feature and distance travelled. Different approaches are investigated for the predictions: wear based on only distance travelled, linear relationship between usage feature and distance, and three levels of braking severity based on the usage data. Four of the approaches underestimate the distance significantly at around 4,000 to 8,000 km against the actual distance of 16,552 km, while the low severity estimate is closest at 20,460 km, agreeing with expectations due to the driving being primarily on highways.