A Simulation-Based Digital Twin for Model-Driven Health Monitoring and Predictive Maintenance of an Automotive Braking System

Abstract

This paper describes a model-driven approach to support heat monitoring and predictive maintenance of an automotive braking system. This approach includes the creation of a simulation-based digital twin, or numerical model, that combines different modeling formalisms into an integrated model of the braking system that can be used for monitoring, diagnostics, and prognostics. The paper provides an overview of the basic models including Modelica models, reduced order models for various key components of the system model, and controls and sensor models. The Modelica models are implemented in the ANSYS Simplorer simulation to leverage existing modeling work and connections with other ANSYS finite element software to utilize reduced order models. The simulation results include both baseline results for the system and the results of injecting failures into the system for monitoring and predictive maintenance.

Understanding the role of a Digital Twin in Integrated Vehicle Health Management (IVHM)

“…Tesla Motors creates a DT for every car it makes and uses it Figure 3: Digital Twin in an IVHM framework (adapted from [5]) to provide fixes, software updates, and to schedule maintenance based on the car's usage and performance [29]. Likewise, Magargle et al [30] from Ansys Inc. and Modelon Inc., developed a physics-based DT of automotive braking systems by combining 0-D and 3-D models with controls. They demonstrated the ability of this DT to differentiate the wear rates with normal and abnormal conditions while disconnecting the ABS controller, which would help in suggesting suitable maintenance schedules in advance for the braking system.…”

Automatic Generation of a Simulation-Based Digital Twin of an Industrial Process Plant

“…While the physical structure of the plant can be determined in a DT by the plant's design material, model-based approaches are required for simulating the dynamics of the physical assets in order to predict, optimize and analyse its changes [19]. Existing commercially-available DT solutions for process plants [20], [21] are commonly based on data-driven models (DDMs) developed purely from the measured data of the targeted industrial plant [14], [22]. DTs based on data-driven approaches rely on black-box models built to capture relations between the inputs and outputs of the plant [10].…”

“…In contrast, simulation-based DTs (SBDTs) are those based on online firstprinciples simulation models [10], [22]- [24]. First-principles models (FPMs) utilize physical, chemical or engineering descriptions to represent the behaviour of the modelled plant [9], [25].…”

“…SBDTs of production plants, also known as tracking simulators [6], [26], [27], are a relatively new simulation-based application that has started to be developed, driven by recent trends in plant digitalization. SBDTs are digital replicas of the targeted plant, which rely on first-principles simulation models to represent its dynamics [22], [23], [54], [74]. These applications leverage the persistent communication between the simulation and physical systems in order to permanently synchronize their underlying dynamic simulation model with the operational plant.…”

“…In the automotive and aerospace industries, DTs have been used as an ultra-high fidelity simulated replica of a vehicle. They have been applied for anomaly diagnoses and for predicting future states and remaining useful life [22], [24]. In the construction industry, DTs have been implemented after combining virtual building models with physical data in order to obtain more accurate information regarding structure fatigue [101].…”