Scholarlink

<p>Adaptive brake pressure control on low and split friction surfaces constitutes a nonlinear stochastic optimal control problem with stringent real‑time and safety constraints. This paper presents an integrated hierarchical framework for intelligent adaptive brake pressure control designed to sustain vehicle stability across stochastic driving environments. Conventional methods treat environmental variations either as deterministic parameters or as unknown disturbances, thereby lacking mechanisms for anticipatory adaptation (i.e., responding before friction changes manifest in slip) and long-term learning from operational experience. To overcome these limitations, the proposed approach probabilistically models driving conditions as a Markov chain, enabling pre-emptive control actions before friction transitions induce instability. At the anticipatory layer, sequential convex programming-based model predictive control (SCP-MPC) computes front/rear wheel-cylinder pressure references from a nonlinear vehicle model, while a discrete Markov environment estimator updates road-state probabilities from measured slip, deceleration, and yaw signals. At the regulatory layer, a gain-scheduled PID loop tracks the commanded pressure and compensates hydraulic spool dynamics. Adaptive dynamic programming (ADP) is used to learn a warm-start policy and to shape the MPC cost, whereby particle swarm optimization (PSO) is used periodically to tune the multi-objective weights to maintain Pareto-optimal performance across varying driving conditions. Simulation results demonstrate that the proposed controller outperforms fixed‑gain PID and deterministic MPC baselines, achieving reductions of 11.4% in stopping distance, 24.2% in slip RMS error, and 35.7% in maximum yaw rate during braking instances with friction transitions. Markov adaptation further reduces brake pressure modulation errors by 41–44% and settling time by 23–27% during state transitions. Lyapunov analysis establishes stochastic input-to-state stability and recursive feasibility, confirming the suitability of the proposed framework for safety–critical intelligent braking applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Intelligent adaptive brake pressure control for sustaining vehicle stability across stochastic driving environments

  • 期刊论文

  • 发表日期：

    2026年7月9日

查看全文
  • Indranil Banik,
  • Avik Chatterjee,
  • Arup Kumar Nandi

摘要

Adaptive brake pressure control on low and split friction surfaces constitutes a nonlinear stochastic optimal control problem with stringent real‑time and safety constraints. This paper presents an integrated hierarchical framework for intelligent adaptive brake pressure control designed to sustain vehicle stability across stochastic driving environments. Conventional methods treat environmental variations either as deterministic parameters or as unknown disturbances, thereby lacking mechanisms for anticipatory adaptation (i.e., responding before friction changes manifest in slip) and long-term learning from operational experience. To overcome these limitations, the proposed approach probabilistically models driving conditions as a Markov chain, enabling pre-emptive control actions before friction transitions induce instability. At the anticipatory layer, sequential convex programming-based model predictive control (SCP-MPC) computes front/rear wheel-cylinder pressure references from a nonlinear vehicle model, while a discrete Markov environment estimator updates road-state probabilities from measured slip, deceleration, and yaw signals. At the regulatory layer, a gain-scheduled PID loop tracks the commanded pressure and compensates hydraulic spool dynamics. Adaptive dynamic programming (ADP) is used to learn a warm-start policy and to shape the MPC cost, whereby particle swarm optimization (PSO) is used periodically to tune the multi-objective weights to maintain Pareto-optimal performance across varying driving conditions. Simulation results demonstrate that the proposed controller outperforms fixed‑gain PID and deterministic MPC baselines, achieving reductions of 11.4% in stopping distance, 24.2% in slip RMS error, and 35.7% in maximum yaw rate during braking instances with friction transitions. Markov adaptation further reduces brake pressure modulation errors by 41–44% and settling time by 23–27% during state transitions. Lyapunov analysis establishes stochastic input-to-state stability and recursive feasibility, confirming the suitability of the proposed framework for safety–critical intelligent braking applications.