Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An Actor-Critic-Identifier Control Design for Increasing Energy Efficiency of Automated Electric Vehicles

Hamed Faghihian, Arman Sargolzaei

TL;DR

This work addresses energy efficiency in automated electric vehicles with unknown dynamics by proposing an Actor-Critic-Identifier (ACI) framework that learns the mapping from control inputs to power online and optimizes drive-cycle performance without an explicit power model. The architecture combines a critic (value function), an actor (control policy), and an online identifier (unknown drift) and uses the Hamilton–Jacobi–Bellman residual $\delta_{HJB}$ to couple learning across the networks, with Lyapunov-based update laws guaranteeing stability. Stability is established via a two-tier Lyapunov analysis and Filippov handling of discontinuities, ensuring asymptotic drive-cycle tracking and uniform ultimate boundedness of the learning weights under appropriate gain choices. In simulation with a realistic EV powertrain, the ACI controller achieves significantly lower net traction energy than a tuned PID baseline (a 42% reduction) and improved tracking, demonstrating a model-free, online approach to enhance EV energy efficiency and energy recovery.

Abstract

Electric vehicles (EVs) are increasingly deployed, yet range limitations remain a key barrier. Improving energy efficiency via advanced control is therefore essential, and emerging vehicle automation offers a promising avenue. However, many existing strategies rely on indirect surrogates because linking power consumption to control inputs is difficult. We propose a neural-network (NN) identifier that learns this mapping online and couples it with an actor-critic reinforcement learning (RL) framework to generate optimal control commands. The resulting actor-critic-identifier architecture removes dependence on explicit models relating total power, recovered energy, and inputs, while maintaining accurate speed tracking and maximizing efficiency. Update laws are derived using Lyapunov stability analysis, and performance is validated in simulation. Compared to a traditional controller, the method increases total energy recovery by 12.84%, indicating strong potential for improving EV energy efficiency.

An Actor-Critic-Identifier Control Design for Increasing Energy Efficiency of Automated Electric Vehicles

TL;DR

This work addresses energy efficiency in automated electric vehicles with unknown dynamics by proposing an Actor-Critic-Identifier (ACI) framework that learns the mapping from control inputs to power online and optimizes drive-cycle performance without an explicit power model. The architecture combines a critic (value function), an actor (control policy), and an online identifier (unknown drift) and uses the Hamilton–Jacobi–Bellman residual to couple learning across the networks, with Lyapunov-based update laws guaranteeing stability. Stability is established via a two-tier Lyapunov analysis and Filippov handling of discontinuities, ensuring asymptotic drive-cycle tracking and uniform ultimate boundedness of the learning weights under appropriate gain choices. In simulation with a realistic EV powertrain, the ACI controller achieves significantly lower net traction energy than a tuned PID baseline (a 42% reduction) and improved tracking, demonstrating a model-free, online approach to enhance EV energy efficiency and energy recovery.

Abstract

Electric vehicles (EVs) are increasingly deployed, yet range limitations remain a key barrier. Improving energy efficiency via advanced control is therefore essential, and emerging vehicle automation offers a promising avenue. However, many existing strategies rely on indirect surrogates because linking power consumption to control inputs is difficult. We propose a neural-network (NN) identifier that learns this mapping online and couples it with an actor-critic reinforcement learning (RL) framework to generate optimal control commands. The resulting actor-critic-identifier architecture removes dependence on explicit models relating total power, recovered energy, and inputs, while maintaining accurate speed tracking and maximizing efficiency. Update laws are derived using Lyapunov stability analysis, and performance is validated in simulation. Compared to a traditional controller, the method increases total energy recovery by 12.84%, indicating strong potential for improving EV energy efficiency.
Paper Structure (16 sections, 2 theorems, 54 equations, 4 figures, 2 tables)

This paper contains 16 sections, 2 theorems, 54 equations, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Problem Formulation and Objectives
  3. Dynamic model
  4. Problem formulation
  5. Method: Actor-Critic-Identifier (ACI)
  6. Challenges
  7. Solution Architecture
  8. Critic Network
  9. Actor Network
  10. Identifier Network Design
  11. HJB Residual based on NNs
  12. Stability Analysis
  13. Necessary Assumptions and Calculations
  14. Proof of Stability
  15. Results
  16. ...and 1 more sections

Key Result

Theorem 1

With the update laws identifier5, the identifier identifier3 guarantees asymptotic tracking $\|\tilde{x}(t)\|,\,\|\dot{\tilde{x}}(t)\|\!\to\!0$ as $t\!\to\!\infty$ whenever

Figures (4)

  • Figure 1: The controller architecture
  • Figure 2: Desired profile versus trajectories under the proposed ACI controller and the baseline controller.
  • Figure 3: Speed-tracking error over the drivecycle for the proposed ACI controller and the baseline controller.
  • Figure 4: Net traction power over the drive cycle for the proposed ACI controller and the baseline controller.

Theorems & Definitions (8)

  • Remark 1
  • Theorem 1
  • proof
  • Remark 2
  • Theorem 2
  • proof
  • Remark 3
  • Remark 4