Adaptive Policy Evaluation With Adjustable Step Sizes for Active Quarter-Vehicle Suspension Systems under IoT Environment

The rapid development of the Internet of Things (IoT), along with the widespread adoption of 5G and time-sensitive networking (TSN), has provided reliable communication support for the development of Internet of Vehicles technologies. As a critical component of intelligent vehicles, active suspension systems under IoT environment play a vital role in enhancing ride comfort and vehicle safety. However, failure to process data uploaded to the cloud promptly may lead to data accumulation, which can subsequently cause data loss, resource wastage, or system response delays. In this article, an adaptive step value iteration (ASVI) algorithm is designed for solving the optimal control problem of active quarter-vehicle suspension systems (AQVSSs), significantly improving the data analysis efficiency of the cloud layer. To prevent divergence caused by excessive policy evaluation step sizes under immature policies, this algorithm incorporates an adaptive step-size adjustment function that is upper bounded and monotonically nondecreasing. Based on convergence of the value function and stability criteria of control policies, an integrated ASVI (IASVI) algorithm is proposed, which avoids the need of admissible control policies and greatly improves learning efficiency. Feasibility and superiority of the IASVI algorithm are verified through a hardware-in-the-loop (HIL) simulation.