A 6-Speed Automatic Transmission Plant Dynamics Model for HIL Test Bench

Transcription

1 SAE TECHNICAL PAPER SERIES A 6-Speed Automatic Transmission Plant Dynamics for HIL Test Bench Quan Zheng, Asif Habeebullah, Woowon Chung and Andrew Herman Delphi Corporation Reprinted From: Transmission and Driveline, 28 (SP-2147) 28 World Congress Detroit, Michigan April 14-17, 28 4 Commonwealth Drive, Warrendale, PA U.S.A. Tel: (724) Fax: (724) Web:

2 By mandate of the Engineering Meetings Board, this paper has been approved for SAE publication upon completion of a peer review process by a minimum of three (3) industry experts under the supervision of the session organizer. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of SAE. For permission and licensing requests contact: SAE Permissions 4 Commonwealth Drive Warrendale, PA USA permissions@sae.org Tel: Fax: For multiple print copies contact: SAE Customer Service Tel: (inside USA and Canada) Tel: (outside USA) Fax: CustomerService@sae.org ISSN Copyright 28 SAE International Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE. The author is solely responsible for the content of the paper. A process is available by which discussions will be printed with the paper if it is published in SAE Transactions. Persons wishing to submit papers to be considered for presentation or publication by SAE should send the manuscript or a 3 word abstract of a proposed manuscript to: Secretary, Engineering Meetings Board, SAE. Printed in USA

3 A 6-Speed Automatic Transmission Plant Dynamics for HIL Test Bench Quan Zheng, Asif Habeebullah, Woowon Chung and Andrew Herman Delphi Corporation Copyright 28 SAE International ABSTRACT During the production controller and software development process, one critical step is the controller and software verification. There are various ways to perform this verification. One of the commonly used methods is to utilize an HIL (hardware-in-the-loop) test bench to emulate powertrain hardware for development and validation of powertrain controllers and software. A key piece of an HIL bench is the plant dynamics model used to emulate the external environment of a modern controller, such as engine (ECM), transmission (TCM) or powertrain controller (PCM), so that the algorithms and their software implementation can be exercised to confirm the desired results. This paper presents a 6-speed automatic transmission plant dynamics model development for hardware-in-theloop (HIL) test bench for the validation of production transmission controls software. The modeling method, model validation, and application in an HIL test environment are described in details. By developing a transmission plant dynamics model, test cases can be created to validate targeted areas of the control software for confirmation of the expected results from software release to release. This is especially important when algorithm/software development timing is aggressive and the management of development activities in a global work environment requires high quality, and timely test results. INTRODUCTION In the paper presented by Zheng, et al. [1], five areas were discussed for the extension of the Delphi transmission HIL capability. Those areas were: 1. Increased test coverage for current TCM software applications. 2. Automation of existing and newly-developed test cases. 3. Development of more sophisticated pass/fail determination of test cases. 4. Integration of higher-fidelity hydraulic models to offer a more realistic transmission response. 5. Data capture method using production vehicle calibration tools to re-create conditions in the HIL environment similar to vehicle conditions. Focus has been invested in #4 and #5, in order to provide the appropriate foundation to align existing data (#5) with simulation fidelity (#4) such that #'s 1-3 can be achieved. The following sections describe the system architecture, powertrain plant model, 6-speed model validation, bench test capability, and future development. TEST BENCH CONFIGURATION Figure 1 shows the Delphi HIL Virtual Car Test Bench setup, which consists of the Host PC, HIL simulator (TestDrive), M-Node, Break-Out Boxes (1 AutoBoB + 1 manual BoB) and the controller (TCM) itself. Figure 2 shows the system architecture and key signal flow diagram. Host PC: It runs the simulator's console software (user interface) to communicate with TestDrive. TESTDRIVE SIMULATOR: TestDrive simulator provides the following functions: The I/O interface and power to the controller Stimulates controller inputs by simulating various sensors like (TRANSOILTEMP, etc.) Communicates with the M-Node

4 Oscilloscope Power Supply TCM Breakout Boxes Manual Breakout Box AutoBOB CAN Ethernet TestDrive Chassis IEEE 1394 (FireWire) Laptop Running INCA Host PC Ethernet M-Node Figure 1 Delphi s HIL Virtual Car Test Bench Provides the Throttle and Brake inputs as commanded by the user Reads the gear-output from the TCM and feeds this to the M-Node plant model Measures the solenoid currents commanded by the TCM and passes them as inputs to the M- Node for solenoid currents to clutch pressures conversion. Retrieves the plant-model generated TISS, TOSS signals and stimulates the appropriate controller inputs. It also gets Vehicle Speed information from the plant-model to display to the console. M-NODE: M-Node performs the drive-train plant-model calculations. It receives the Throttle, Brake inputs from the user and the commanded Gear and TRANSOILTEMP from the controller via TestDrive. It calculates the TISS and TOSS values from the driveline model and outputs them to TestDrive so as to be converted to real-world sensor outputs. The current to pressure conversion subsystem accepts the commanded/measured solenoid currents and outputs the respective clutch pressures to the transmission clutches. BREAK-OUT-BOXES (BOB): BoBs are used to simulate electronic/manual faults for diagnostic purposes. TRANSMISSION CONTROL MODULE (TCM): The controller reads the TISS, TOSS sensor signals from the TestDrive simulator, performs the control algorithm calculations, determines the appropriate gear for the automatic transmission based on various input signals (throttle, brake switch status, PRNDL switch status, transmission oil temperature and other sensor inputs), calculates the solenoid currents and outputs the current waveforms onto the solenoids. LAPTOP: The laptop as such doesn't form part of the closed-loop simulation setup. However, it is used for flashing the software and calibrations into the controller, recording/displaying the internal controller variables and calibrating the controller through a calibration tool like ETAS INCA USER INPUTS: User inputs to HIL bench are throttle and brake.

5 PRNDL Throttle Brake TFT Host PC with TestDrive Console Outputs Inputs TestDrive Gear PWM Solenoid currents Actuators Sensors CAN bus TCM Throttle Brake Gear TFT Solenoid currents TISS TOSS VKPH CCP Laptop with INCA M-Node Figure 2 HIL Bench Architecture 6-SPEED AUTOMATIC TRANSMISSION PLANT DYNAMICS MODEL Powertrain dynamics modeling has been an active research and development field for many years. In general, a complete powertrain model has representations of engine, torque converter, transmission, and driveline/ vehicle dynamics. Depending on the application, the powertrain model may have different levels of complexity for a certain part of the powertrain model. For example, if the emphasis is on engine applications, then the engine model may be of high-fidelity. Similarly, for chassis applications, the vehicle dynamics model is very detailed. In our case, the motivation for the development of automatic transmission plant model is to enable the algorithm/test engineer(s) to check the controller performance with a representative transmission plant model in the absence of the real transmission hardware. The powertrain plant dynamics model is a key element of the Virtual Car Test Bench as shown in Figure 1, which requires a powertrain model with high-fidelity representation of the transmission dynamics. Such Virtual Car Test Bench is used for controller and software testing, control algorithm development and testing. The initial model is developed by starting with an example SimDriveline model from the MathWorks. This model represents the major components of the drivetrain (engine, torque converter, transmission, and vehicle load) by specialized blocks from the SimDriveline blockset of the Simulink library. The 6-speed AT mechanical model is then configured using SimDriveline toolbox building blocks. Unlike the conventional modeling approach of deriving the transmission dynamics state equations and transitions, SimDriveline toolbox offers components with built in state equations and transition conditions. Therefore, the state transitions and boundary conditions are taken care of by the physical model. Table 1 shows the clutch state table for the 6-speed automatic transmission of interest. Figure 3 shows the powertrain model developed using SimDriveline toolbox. The model parameters of various blocks are updated to reflect the vehicle design parameters. The engine speedtorque map, the torque converter parameters (speed and torque ratios, capacity factor), transmission gear ratios, clutch design parameters, transmission inertias, the final drive gear ratio, the vehicle inertia and the road-load values. The model also uses certain parameters that are not readily available due to the lack of proprietary design information. These parameters are tuned in an iterative fashion and this is found to be an arduous process. The key transmission dynamics model parameters are tuned by inputting the actual vehicle engine speed in the model. The modeled transmission input speed and output speed are then compared with vehicle data. The powertrain model is developed with the following stages to deliver different levels of model fidelity at each stage: Stage 1: Develop and validate 6-speed AT mechanical model.

6 Stage 2: Develop and validate the hydraulic pressure generation model Stage 3: Validate vehicle dynamics model Stage 4: Develop and validate engine model to have proper torque generation. This staged development process enables us to focus on the key modules needed for controller and software testing purpose, and at the same time, the model development is continued to improve the model fidelity of other modules. R D1 D2 D3 D4 D5 D6 C1234 UD C456 OD C35R C26 1 Used for engine braking when on. CBLR LR/B 1 Table 1 Clutch state table of 6-Speed AT F1 measure engine speed is used as input to the powertrain model. The modeled transmission input speed is compared with the measure transmission input speed. Another source of variation is the vehicle loading on the output shaft of the transmission. There are several steps to develop a set of representative parameters for vehicle loading. After the load model is derived, the simulated output speed is compared with the measured output speed. Figure 4 - Figure 6 show the model validation results. These results are organized by recording signals from a typical test vehicle driven under selected driving conditions (this constitutes the test-data), running the model with its inputs being driven from the test-data and comparing the model outputs against the test-data outputs. Figure 4 shows the vehicle data of throttle, brake and gear requests, which are inputs to the model. Figure 5 shows the engine speed and engine torque traces. The engine torque graph shows the comparison engine torque calculated by model and the data taken from vehicle. It is obvious that engine model needs additional improvements. For our model validation purpose, engine speed is fed into the powertrain model using SimDriveline speed input block. This allows us to focus on transmission dynamics directly without focusing heavily on the fidelity of engine model. Figure 6 shows the comparison of transmission input speed (TISS), transmission output speed (TOSS) and vehicle speed calculated by the model and measured from vehicle. As shown in Figure 6, the model and vehicle data show a close correspondence for the most part with the transmission up-shifts and down-shifts. However, there are a few areas that the speed levels of modeled and measured data do not line up. This is due to the vehicle loading part of the model being not accurate. This is part of our future work to improve the vehicle dynamics model in order to add proper loading to the transmission. Figure 3 Drive-train Plant 6-SPEED AT MODEL VALIDATION The automatic transmission model is validated by vehicle data. One challenge of the model development is the tuning of the model parameters. The original design parameters are limited, and are sometime proprietary. Therefore, model parameter identification and tuning becomes an important task. Another challenge of the model development is that a detailed engine dynamics model is necessary to fine tune the transmission dynamics response. The focus of the model is on the transmission dynamics. Therefore, actual measured vehicle data is used to tune the transmission model parameters. In this case, Throttle Brake Gear request x Time (secs) Figure 4 (User) Inputs

7 Engine Torque Engine Speed Time (secs) Figure 5 Outputs: Engine torque and speed TISS 4 2 TOSS, Engine RPM, vehicle speed etc based on user inputs such as throttle and brake. As shown in Figure 7, the plant model is running on Host PC. Based Throttle and brake inputs from the user, the plant model generates transmission shifting control related signals (TISS, TOSS etc.). These signals are converted to actual sensor signals that are inputs to TCM. Based on these speed signals and user inputs, TCM control software determines shift control strategy and generates gear change commands. HIL bench reads this gear change request (gear ratio change command) from the controller, which triggers the actuation of proper solenoids to generate pressure of corresponding clutches. When certain clutches are actuated, shifting happens. Plant dynamics model generates TISS and TOSS, which are read by the controller. By such, HIL bench mimics realistic transmission shifting and generates proper signals to the controller. TOSS Vehicle speed Time (secs) Figure 6 Transmission Outputs Figure 7 System architecture of Delphi's HIL Closed loop Bench BENCH TEST CAPABILITIES CLOSED LOOP HIL BENCH Conventional test benches are open loop benches. Open loop here means that there is no plant model running to provide realistic feedback signals. Engineers need to manipulate certain inputs to test controller and control functions. However, this type of testing is not sufficient for certain applications such as transmission control algorithm development and verification. To manually generate speed traces during shifting is not a trivial task. Figure 7 shows the simplified representation of the closed loop HIL bench. The difference between open loop and closed loop HIL bench is that closed loop bench generates realistic powertrain outputs such as TISS, The closed loop bench concept explained above is clearly shown in Figure 8. As shown in the figure, HIL bench reads the solenoid current output from TCM. Plant model converts this current reading to the corresponding solenoid pressure and calculated clutch pressures. These clutch pressures are then converted to the actuation force on the corresponding clutches. Based on gear ratio change mechanics, the plant dynamics model generates realistic TISS (Input shaft speed) and TOSS (Output shaft) RPM, which are fed to TCM. Thus, transmission control algorithms can be verified with realistic TISS, TOSS and VRPM relations during shifting.

8 Command Current Control Pressure Clutch Braking Force= uf F=P*A Figure 9 Example Power off 1->2 shift Closed Loop Plant Generate TOSS,TISS and VRPM relation Figure 8 Current, Pressure and Closed Loop structure As pointed out earlier, engine RPM signal needs to be fed to the plant model because a validated high fidelity engine plant dynamics model is not yet available at this stage of development. Engine torque is estimated by using measured engine speeds. CONTROL ALGORITHM TESTING The new generation of Delphi's transmission control methodology is to provide the customer with the best calibration authority. This means that transmission control can be achieved by very general calibration setting. For those familiar with the art of transmission controls, a key point of the art is to properly control transmission actuators at each phase of shifting process. In order to test this new generation of transmission control strategy, closed loop test bench is a necessity. An example of the testing needs is given below. One important aspect of transmission shift control is to properly determine the shift progress, i.e. shift phases. For the apply and release clutch phase transitions, phase duration time is used as well as certain event triggers, such as Shift Begin, Shift Finish, etc. As shown in Figure 9, during 1->2 power off shifting, Apply1 Clutch SYR point can be set to true by model the HIL bench. Therefore, the phase transition algorithm can be verified. PLAY BACK MODE Play Back Mode is an important function of the HIL Virtual Car bench. Play Back enables test engineers to playback test data repeatedly to test certain areas of controller functionality. This functionality also allows data collected in a test vehicle or dyno to be repeated, which is important to reproduce errors that are difficult, damaging or maybe even dangerous in a test vehicle (Zheng, et al., 27). The concept of Play Back Mode is to make HIL test bench re-produce the recorded (vehicle) signals repeatedly, i.e. to re-create the vehicle test scenario on the test bench. In this section, we will detail the data capture method using production vehicle calibration tools to re-create conditions in the HIL environment similar to vehicle conditions. Since the bench model is running in Matlab environment, it is necessary to convert all data file to this Matlab.mat data format. Figure 1 shows the process of data format conversion, which can be summarized at the following steps: Collect input signals to TCM using production vehicle calibration tool; Convert.dat file to Matlab.mat format (mdfimport developed by Mathworks, Figure 12); Convert data to the right format (loop rate, row, column, etc.)

9 mdfimport Vehicle test data(*.dat) mat file Save as mat file Time align as 5ms and Row,column change Run model 5 ms step By conversion m file Developed by Delphi Figure 1 Data format conversion Figure 12 MDFIMPORT program developed by Mathworks Figure 13 shows data comparison between input and output data. Here the compared data are TISS, TOSS and VRPM. B* represents bench generated signal. As shown, the Play Back Mode data matches the vehicle data very well. Figure 11 Input signals to the TCM Figure 11 shows input signals to the TCM that are necessary for the proper transmission control. In this case, TCM uses CAN for Eng related signal sharing, which are the necessary input to the model as well. Figure 12 shows MDFIMPORT program used to convert the recorded vehicle data (*.dat) to Matlab data format. Figure 14 shows the comparison of solenoid current generated on the bench and vehicle compared result, where variables AUTCURA (26Brake), AUTCURB(UD), AUTCURD(35R) and AUTCURE(OD) represent solenoid control current (from to 85mA), and B* represent bench test result with mat conversion. As shown in the figure, the two sets of data match very well, which means that same outputs from TCM can be expected via Play Back. In summary, Play Back Mode feature allows test engineers to replay vehicle test data on the HIL bench repeatedly, which is not possible by human driving a car. Therefore, Play Back Mode makes it easier to find the root cause of a problem and eventually fix the problem.

10 BVTINSPDF VTINSPDF FUTURE DEVELOPMENT The above described 6 Speed HIL environment offers the ability to evaluate real-world data against previously discussed model. The extensions of this capability will follow primary paths. Below are the areas of focus for future development: 1. Increased model fidelity, addressing the following: Add vehicle dynamics to the model by modeling a differential, longitudinal car dynamics and tires BVOUTSPDF VOUTSPDF BVRPM VRPM Exercise the current to pressure generation portion (already available in the model) so that it produces the clutch pressures for the transmission clutches. Add a hydraulic plant to model the transmission hydraulics; SimHydraulics physical modeling block-set could be utilized for this purpose. Use Simulink Parameter Estimation to dynamically tune the model parameters to further refine performance. Improved Engine model 2. Increased TCM HW and SW Test Coverage (items described in the INTRODUCTION not addressed in this paper): Increased test coverage for current TCM software applications Figure 13 VTINSPDF (TISS), VOUTSPDF (TOSS) and VRPM Automation of existing and newly-developed test cases. Development of more sophisticated pass/fail determination of test cases BAUTCURA BAUTCURB BAUTCURD BAUTCURE AUTCURA AUTCURB AUTCURD AUTCURE CONCLUSION In this paper, we presented the development of the 6- speed automatic transmission plant dynamics model for the Delphi s HIL Virtual Car Test Bench and how the HIL bench is used in the verification and validation of production controller and software. System architecture, plant model, bench capabilities and example test cases are presented. Finally, future development is given. REFERENCES Figure 14 Compared result 1. Q. Zheng, W. Chung, K. Defore and A. Herman, A Hardware-in-the-loop Test Bench for Production Transmission Controls Software Quality Validation, SAE Technical Paper, No , 27.

11 CONTACT Quan Zheng, Ph.D. Staff Research Engineer -Forward Transmission Controls Delphi Powertrain Systems Technical Center Brighton 1251 E. Grand River, Brighton, MI Phone: Asif Habeebullah Systems Engineer Delphi Powertrain Systems Technical Center Brighton 1251 E. Grand River, Brighton, MI Woowon Chung Senior Project Engineer - Forward Transmission Algorithms Delphi Powertrain Systems Technical Center Brighton 1251 E. Grand River, Brighton, MI Phone: woowon.w.chung@delphi.com DEFINITIONS, ACRONYMS, ABBREVIATIONS ECM: Engine Control Module TCM: Transmission Control Module PCM: Powertrain Control Module AutoBOB: Automated Break-Out-Box HAL: Hardware Abstraction Layer HWIO: Hardware Input Output Layer HIL: Hardware In the Loop Matlab, Simulink, Stateflow, SimDriveline are Registered Trademarks of MathWorks, Inc. TestDrive is a Registered Trademark of Opal-RT Technologies. INCA is a Registered Trademark of ETAS Inc. Andrew Herman Staff Engineer- Forward Transmission Algorithms Delphi Powertrain Systems Technical Center Brighton 1251 E. Grand River, Brighton, MI Phone: andrew.herman@delphi.com