Failure prediction and maintenance decision-making are two core activities in a prognostics and health management (PHM) system. However, they are often studied independently and hierarchically. The main goal of this paper is to combine system failure prediction with maintenance decision-making to develop a predictive maintenance strategy. System failure prediction is achieved by constructing an ensemble model of deep autoencoder (DAE), long short-term memory (LSTM), quantile regression (QR) and kernel density estimation (KDE), namely DAE-LSTMQR-KDE. Then, on the basis of the probability density of system failure time obtained from the ensemble model, replacement cost function (RCF) and ordering cost function (OCF) are proposed to support maintenance and inventory decisions. Finally, optimal decisions are determined by minimizing the two cost functions. A score equal to 246.59 and a coverage width-based criterion (CWC) index equal to 0.35 were obtained when the DAE-LSTMQR-KDE ensemble model was applied to C-MAPSS dataset, while the average maintenance cost rate (MCR) of the proposed maintenance strategy was 0.74. The results demonstrated that the proposed prediction and maintenance method outperforms several state-of-the-art methods. In addition, different cost structure scenarios are also investigated to illustrate the flexibility of maintenance decisions based on failure prediction information.

A wheel force transducer is a vital instrument in the vehicle-testing field which provides a means for determining experimentally the forces and the moments transmitted to a vehicle through the tyre contact patch. However, existing wheel force transducers are almost commercial products. They are expensive and, for business reasons, technical information is not made public. This slows down the development of the wheel force transducer itself to a certain extent. Moreover, research studies on the wheel force transducer mainly focus on utilization of this tool for vehicle simulations, traction control and anti-lock braking evaluation as it provides information on the characteristics, the performance and the limitations of the vehicle. From the perspective of the sensor itself, although improvement in the transducer accuracy and information extraction using signal-processing methods have been successfully developed in the literature, few investigations on the design of the wheel force transducer have been reported yet. Accordingly, a general design and optimization procedure for the wheel force transducer with a practical case study will be of great significance. Since technical difficulties, admittedly, are primarily about the issues of how to design and optimize the transducer, in this paper a new self-decoupled six-axis wheel force transducer for a heavy truck is designed and fabricated. An easy-to-understand design procedure is highlighted mainly including the conceptual design with the aims of a universal-purpose self-decoupled transducer with an eight-spoke structure, an optimization design which depends on the proposed comprehensive performance criteria in the computer-aided procedure, and a collectivity design focus on the mechatronics assembly for a heavy-truck application. Finally, the wheel force transducer is manufactured, calibrated and tested. The results show that the mean errors of the six-axis forces are about 3% of the full scale with 1% non-linearity, 0.5% repeatability and 1% hysteresis for the sensor only, that there are increased mean values of about 2% non-linearity, 1% repeatability and 1.4% for a certain tyre assembly and that the observed dynamic response is about 2.5–3.5 ms. Both the self-decoupled characteristics and the sensor performance are verified to meet the design and optimization, and the transducer with a real-time wheel force output is also confirmed by road tests.