Search

Recently, connected and autonomous vehicle (CAV) technology is gaining attentions around the world. A connected vehicle (CV) is equipped with sensors to collect real-time vehicle data and communication devices to ‘talk’ with other surrounding vehicles and/or infrastructures (such as traffic signals). With the communication capability, a CV can obtain real-time traffic information that was not available today, such as preceding vehicles’ speeds, locations, and signal phase and timing (SPaT). This newly available information enables a CV to anticipate future driving conditions. With partial or full vehicle automation, the vehicle’s motion (acceleration, speed, etc.) can be controlled in real-time. With both connectivity and automation, the vehicle can respond to traffic conditions proactively and operate in the most energy efficient manner. This is achieved by an effective prediction of future driving conditions enabled by connectivity and being able to plan and adapt vehicle speed and powertrain operation in a more flexible and optimal fashion enabled by the vehicle automation. The target CAV can accelerate and decelerate smoothly, avoid unnecessary braking and idling, and operate the powertrain system more efficiently. HASH(0x40df3a8) This dissertation aims to explore the potential of energy savings through vehicle automation and connectivity. Real-time implementable optimal control strategies are developed for energy savings of CAVs with three most common powertrain types today: internal combustion engine based vehicles (ICVs), hybrid electric vehicles (HEVs), and electric vehicles (EVs). Both the vehicle speed trajectory and powertrain operation (e.g. transmission gear position or engine-battery power-split) are optimized to maximize the energy benefits. The optimization is designed for each specific powertrain type considering impacts of powertrain operation, constraints and dynamics on the energy consumption. This ensures the optimal control law is realistic and can be potentially implemented on an actual vehicle. A systematic control framework, which combines both traffic prediction and energy optimization, is developed to implement the optimal control in the model predictive control (MPC) fashion. The traffic prediction method can be applied to scenarios where both connected and non-connected vehicles are on the road. A traffic flow model is used to describe dynamics of the traffic states (traffic density and traffic speed). Current traffic states are estimated using an observer based on real-time information communicated from connected vehicles and signal lights. Future traffic states can be predicted by propagating the traffic flow model forward in time. Uncertainties in the traffic prediction are systematically quantified and considered during the optimization to ensure the performance of the optimal control. To experimentally validate the optimal control strategies, a hardware-in-the-loop (HIL) testbed developed previously in our group has been enhanced. The HIL testbed has an actual engine loaded by a hydrostatic dynamometer and emulates a virtual target vehicle with simulated vehicle dynamics. To increase the fidelity of the testbed, the same type of engine as the actual vehicle is installed and both vehicle and powertrain models are calibrated using actual vehicle testing data. The resulting HIL testbed matches well with the performance of the actual testing vehicle with only about 1% error. In addition, a living lab is developed with instrumented on-road testing vehicles and intersection at TH55 in Minnesota. This brings in real-world traffic data and enables the HIL testbed to interact with real traffic. The results significantly improve the credibility of the HIL testbed. The inclusion of real-world traffic information extends the capabilities of the HIL testbed to evaluate the real-time capability and robustness of various CAV applications in realistic roadway conditions. For ICVs, the optimal control problem is a hybrid one with both continuous (vehicle speed, braking force) and discrete (gear position) control inputs. The problem is formulated and simplified to a mixed integer programming problem with a convex quadratic objective function and linear constraints. The optimal control solutions are obtained in real-time using an efficient numerical solver. Two traffic scenarios are studied in both simulation and experiment: 1) a rolling terrain scenario without a nearby preceding vehicle. The target vehicle anticipates the future roadway slopes and cruises with the optimal speed and gear position. The results have shown that the fuel benefit is 16.1% compared to a baseline vehicle using constant speed cruising control. 2) a vehicle platooning scenario on a signalized urban road. The target vehicle is at the end of a vehicle platoon and follows the preceding vehicle. The target vehicle can achieve 10.6% fuel benefits compared to the immediate preceding vehicle. For HEVs, the vehicle speed and powertrain operation optimization algorithms are developed and solved in a consecutive order. The vehicle level optimization solves an optimal vehicle speed trajectory. The optimal speed profile is then sent to the powertrain controller to optimize both engine operating points and the power-split (engine power vs battery power). Again, the control is evaluated in both a rolling terrain scenario and an urban driving scenario. In the rolling terrain scenario, the energy benefits from the proposed optimal controller are 5.0% to 8.9% on major arterials and 15.7% to 16.9% on collector roads, compared to a regular HEV cruising at constant speed. In the urban driving scenario, the target vehicle can achieve 17.3% fuel benefit with vehicle level optimization alone. With both vehicle and powertrain optimization, the fuel benefit is 22.8% comparing to the preceding vehicle. For EVs, the efficiencies of the electrical powertrain (including the electric motor, battery, etc.) are considered during the vehicle speed trajectory optimization. The battery aging effects are considered to ensure a satisfactory battery life and several regenerative braking constraints are included to make the optimal control strategy realistic for practical implementation. The optimal control strategy is evaluated in a vehicle platooning scenario on a signalized roadway with two intersections. For a vehicle platoon of 16 vehicles, with 50% penetration rate of connectivity, the target vehicle can achieve 9.1% energy improvement. The performance is satisfactory compared to the 14.3% energy saving with perfect traffic prediction. This demonstrates the effectiveness of the proposed optimization method.