Gear shift strategies for automotive transmissions

Abstract

The development history of automotive engineering has shown the essential role of transmissions in road vehicles primarily powered by internal combustion engines. The engine with its physical constraints on the torque and speed requires a transmission to have its power converted to the drive power demand at the vehicle wheels. Under dynamic driving conditions, the transmission is required to shift in order to match the engine power with the changing drive power. Furthermore, a gear shift decision is expected to be consistent such that vehicle can remain in the next gear for a period of time without deteriorating the acceleration capability. Therefore, an optimal conversion of the engine power plays a key role in improving the fuel economy and driveability. Moreover, the consequences of the assumptions related to the discrete state variable-dependent losses, e.g. gear shifting, clutch slippage and engine starting, and their e?ect on the gear shift control strategy are necessary to be analyzed to yield insights into the fuel usage. The ?rst part of the thesis deals with the design of gear shift strategies for electronically controlled discrete ratio transmissions used in both conventional vehicles and Hybrid Electric Vehicles (HEVs). For conventional vehicles, together with the fuel economy, the driveability is systematically addressed in a Dynamic Programming (DP) based optimal gear shift strategy by three methods: i) the weighted inverse of the power re¬serve, ii) the constant power reserve, and iii) the variable power reserve. In addition, a Stochastic Dynamic Programming (SDP) algorithm is utilized to optimize the gear shift strategy, subject to a stochastic distribution of the power request, in order to minimize the expected fuel consumption over an in?nite horizon. Hence, the SDP-based gear shift strategy intrinsically respects the driveability and is realtime implementable. By per¬forming a comparative analysis of all proposed gear shift methods, it is shown that the variable power reserve method achieves the highest fuel economy without deteriorating the driveability. Moreover, for HEVs, a novel fuel-optimal control algorithm, consist-ing of the continuous power split and discrete gear shift, engine on-o? problems, based on a combination of DP and Pontryagin’s Minimum Principle (PMP) is developed for the corresponding hybrid dynamical system. This so-called DP-PMP gear shift control approach benchmarks the development of an online implementable control strategy in terms of the optimal tradeo? between calculation accuracy and computational e?ciency. Driven by an ultimate goal of realizing an online gear shift strategy, a gear shift map design methodology for discrete ratio transmissions is developed, which is applied for both conventional vehicles and HEVs. The design methodology uses an optimal gear shift algorithm as a basis to derive the optimal gear shift patterns. Accordingly, statis¬tical theory is applied to analyze the optimal gear shift pattern in order to extract the time-invariant shift rules. This alternative two-step design procedure makes the gear shift map: i) respect the fuel economy and driveability, ii) be consistent and robust with respect to shift busyness, and iii) be realtime implementation. The design process is ?exible and time e?cient such that an applicability to various powertrain systems con?gured with discrete ratio transmissions is possible. Furthermore, the study in this thesis addresses the trend of utilizing the route information in the powertrain control system by proposing an integrated predictive gear shift strategy concept, consisting of a velocity algorithm and a predictive algorithm. The velocity algorithm improves the fuel economy in simulation considerably by proposing a fuel-optimal velocity trajectory over a certain driving horizon for the vehicle to follow. The predictive algorithm suc¬cessfully utilizes a prede?ned velocity pro?le over a certain horizon in order to realize a fuel economy improvement very close to that of the globally optimal algorithm (DP). In the second part of the thesis, the energetic losses, involved with the gear shift and engine start events in an automated manual transmission-based HEV, are modeled. The e?ect of these losses on the control strategies and fuel consumption for (non-)powershift transmission technologies is investigated. Regarding the gear shift loss, the study ?rstly ever discloses a perception of a fuel-e?cient advantage of the powershift transmissions over the non-powershift ones applied for commercial vehicles. It is also shown that the engine start loss can not be ignored in seeking for a fair evaluation of the fuel economy. Moreover, the sensitivity study of the fuel consumption with respect to the prediction horizon reveals that a predictive energy management strategy can realize the highest achievable fuel economy with a horizon of a few seconds ahead. The last part of the thesis focuses on investigating the sensitivity of an optimal gear shift strategy to the relevant control design objectives, i.e. fuel economy, driveability and comfort. A singu¬lar value decomposition based method is introduced to analyze the possible correlations and interdependencies among the design objectives. This allows that some of the pos¬sible dependent design objective(s) can be removed from the objective function of the corresponding optimal control problem, hence thereby reducing the design complexity.