Design Of Shaft Torque Estimator

Vehicle Transmissions generally include manual transmission (MT), automatic mechanicaltransmission (AMT), dual-clutch transmission (DCT), Automatic Transmission (AT) and continuouslyvariable transmission (CVT). AMT is in forefront use of vehicle transmissions with its unique advantageson high efficiency and low cost. Besides, AMT is not only simpler than DCT, AT and CVT in controlprograms. But also easier than MT in operations and obviously have higher transmission efficiency thanAT and CVT. However, the biggest drawback for AMT is the power interruption in gear shifting process.Therefore, the researches on AMT currently are all about reducing gear shifting time while ensuringcomfort. For AMT gear shifting process optimization issue, some scholars proposed engine speed activecontrol methods, while most of them through the design of shaft torque estimator to make it true.Vehicle development mainly follows on the V-mode steps, where the model building is anindispensable step for rapid prototyping and hardware in loop. The rapid prototyping and hardware inloop are all semi-physical simulations that require enough accurate simulation models. Therefore, modelvalidation and parameter debugging of simulation model become very important. The model aftervalidation can be used as a controller simulation platform, and also can instead actual structure to validateprocedures and optimize parameters.In this paper, we need to build a heavy truck simulation platform with12-AMT, and then validateshaft torque estimators on the validated platform. A designed estimator is mainly based on Input-to-StateStable (ISS) theory, and the other linear Kalman filter is just for comparison. The main contents of thispaper are as follows:1. The design of driving shaft torque estimator is based on ISS theory. A reduced-order drivingshaft torque estimator is designed and if the control inputs (engine torque and resistance torque)have limits, the torque tolerance will be limited to10%of its maximum transmitted valueaccording to ISS theory. As the ISS nonlinear estimation algorithm is complex, so designinganother simple linear Kalman filter algorithm torque estimator for comparison.2. Building the overall12-AMT heavy truck model, including engine, clutch and its actuator,transmission and its actuators and vehicle body. Carrying out driveline bench test and then validating the model by open-loop control. The results show that the accuracy of all parts of themodel and their actuators’ dynamic response time can reach more than80%, which illustrate themodel is available.3. Checking the ISS reduced-order shaft torque estimator on the whole vehicle simulation platformand the estimator can meet the requirements with switched estimator gain. The results ofKalman filter checking show that the filter has good stability if its initial value and initialcovariance are given and its system error and measurement error are defined properly. However,the ISS estimator is more useful than the simple Kalman filter due to that system error andmeasurement error are difficult to be given accurately.4. Finally, we developed a clutch disengaging strategy based on the ISS torque estimator. Thesimulation results show that if the clutch being disengaged on time, not only the gearshift timeand vehicle jerk will be shortened and reduced, but also the driveline backlash shocking will beprevented.