Modeling extreme loads acting on steering components using driving events

• To identify driving events we use hidden Markov models (HMMs).
• We use a multivariate generalized Laplace model (GAL) as the distribution of the observation process in the HMM.
• We model the sequence of detected driving events (left and right turns) with a Markov chain.
• We define a reduced load by keeping the extreme value for each left and right turn and zero for the straight forward event.
• An explicit formula for the expected damage is computed using the reduced load.

Forces during steering events, such as curves and maneuvers, cause large stresses on steering components. In this paper, we formulate a model for the lateral loads causing fatigue damage of the steering components. Steering events are identified using a Hidden Markov model on the CAN (Controller Area Network) bus data. The CAN data is available on all vehicles, thus the model is applicable across many types of vehicles. To identify the events, the observation from CAN data is modeled by a multivariate generalized Laplace (GAL) distribution. An explicit formula for the expected fatigue damage is given. Results are validated using measured lateral acceleration.