Alternative vehicle electronic architecture for individual wheel control
Electronic control systems have become an integral part of the modern vehicle and their installation rate is still on a sharp rise. Their application areas range from powertrain, chassis and body control to entertainment. Each system is conventionally control led by a centralised controller with hard-wired links to sensors and actuators. As systems have become more complex, a rise in the number of system components and amount of wiring harness has followed. This leads to serious problems on safety, reliability and space limitation. Different networking and vehicle electronic architectures have been developed by others to ease these problems. The thesis proposes an alternative architecture namely Distributed Wheel Architecture, for its potential benefits in terms of vehicle dynamics, safety and ease of functional addition. The architecture would have a networked controller on each wheel to perform its dynamic control including braking, suspension and steering. The project involves conducting a preliminary study and comparing the proposed architecture with four alternative existing or high potential architectures. The areas of study are functionality, complexity, and reliability. Existing ABS, active suspension and four wheel steering systems are evaluated in this work by simulation of their operations using road test data. They are used as exemplary systems, for modelling of the new electronic architecture together with the four alternatives. A prediction technique is developed, based on the derivation of software pseudo code from system specifications, to estimate the microcontroller specifications of all the system ECUs. The estimate indicates the feasibility of implementing the architectures using current microcontrollers. Message transfer on the Controller Area Network (CAN) of each architecture is simulated to find its associated delays, and hence the feasibility of installing CAN in the architectures. Architecture component costs are estimated from the costs of wires, ECUs, sensors and actuators. The number of wires is obtained from the wiring models derived from exemplary system data. ECU peripheral component counts are estimated from their statistical plot against the number of ECU pins of collected ECUs. Architecture component reliability is estimated based on two established reliability handbooks. The results suggest that all of the five architectures could be implemented using present microcontrollers. In addition, critical data transfer via CAN is made within time limits under current levels of message load, indicating the possibility of installing CAN in these architectures. The proposed architecture is expected to· be costlier in terms of components than the rest of the architectures, while it is among the leaders for wiring weight saving. However, it is expected to suffer from a relatively higher probability of system component failure. The proposed architecture is found not economically viable at present, but shows potential in reducing vehicle wire and weight problems.