Virtual Application for Volkswagen
Hardware-in-the-loop simulation makes it possible to make a full test of control units in a virtual driving test. At Volkswagen, HIL tests are currently being carried out for the application of a new ESP system for the Golf V platform. A CarMaker/HIL test bench from IPG Automotive is being used as the simulation environment.
At VW, the sign of the times was recognised very early. Long before hardware-in-the-loop simulation (HIL simulation) had made a name for itself with developers and application engineers of electronic control systems, ECUs were being tested at VW in simulation environments. The purpose of this discussion is to introduce the use of hardware-in-the-loop simulation for the application of a new ESP system for the Golf V platform (PQ35).
The Simulation Environment
The real-time platform CarMaker/HIL from IPG Automotive is being used as the simulation environment. This meets two essential criteria: as regards software, the simulation of dynamic driving characteristics in real time, and as regards hardware, the coupling of an ESP system with sensor stimulation and simulation of the brake hydraulics.
The CarMaker simulation programme provides an accurate detailed virtual driving environment to simulate driving dynamics. Central to this driving environment is a vehicle model capable of real-time performance and which is valid up to the limits of driving dynamics. This is a Multi-Body-System (MBS) in which the kinematic and elastokinematic coupling of the wheel carriers is described in parameter data sets. In order to set the parameters for the vehicle to be examined, data from measurements or from simulations with a Multi-Body-Simulation programme (e.g. IPG KINEMATICS) can be used. At VW, the Golf V chassis is simulated with Adams. Then an automatic data conversion generates the parameter data sets for the CarMaker vehicle model.
The MBS model forms the core model and integration platform for all other sub-components of the vehicle, such as the steering system, tyres, brake system, drivetrain and aerodynamics. Overall, the model environment of the virtual vehicle has a uniform, modular structure: the models of the vehicle environment are arranged manageably in terms of the individual sub-assemblies and can be modified or exchanged for in-house models. The way that parameters are set for the vehicle models is also linked to individual sub-assemblies and therefore to functional units and structures to which test drivers and application engineers are accustomed. When constructing the overall vehicle model, there are various model variants to choose from for the individual vehicle components. A special ESP brake system is provided as a hydraulics model for testing the ESP control. Parameter setting for the hydraulics model had to be adapted for similarly controllable valves in order to meet the criteria of the analog valve controls of the new brake system. Parameters were set for the hydraulics model in accordance with the ESP system to be tested on the basis of data on the behaviour of the hydraulic valves in various operating conditions.
The virtual vehicle environment also includes a robust driver model with full parameters (IPG-DRIVER) as well as the virtual vehicle. This means that all operating conditions necessary for the application of ESP control can be illustrated, e.g. various braking manoeuvres or ISO lane changes at different speeds. Three-dimensional road models (IPG-ROAD) together with the virtual driver and vehicle complete the virtual vehicle environment. The road models can either be constructed from individual sections of route or they represent real test routes (e.g. the Hockenheim Ring or Nürburg Ring northern loop), which were defined from measurement data. In order to be able to make as good a comparison as possible between test and simulation, in-house test routes were digitized at VW and then incorporated into the virtual vehicle environment.
The hardware components of the HIL simulator include a power supply unit, a real-time computer, I/O modules and slots for the ECUs. The slots contain the signal conditioning for the ECUs and the recording unit for the magnet coils of the ESP hydraulics.
The ESP ECU in the Simulation Loop
The new ESP ECU is connected in one of these slots of the HIL simulator. A special feature of this control unit is that the sensors for the yaw rate and lateral acceleration are not fitted in an external sensor cluster but are integrated in the control unit itself. This means that it is not possible to separate the sensors from the system in order to transmit the yaw rate and lateral acceleration calculated by the models straight into the ESP control unit as input values. To solve this problem various approaches have been discussed. One approach under discussion would be to mount the control unit on a motion platform and to generate the lateral acceleration and yaw rate measurement values with the movements of this platform. Another idea was to circumvent the sensors by manipulating the ECU code. In the end, the second solution option was chosen. The ECU supplier provided a special sensor interface for this. Using this, the yaw rate and lateral acceleration can be fed into the control unit via CAN. Using a software switch, the internal yaw rate and lateral acceleration sensor can be switched off so that the control unit can await the input signals from the CAN bus.
The ECU picks up the signals, tests them for plausibility and executes logical and mathematical links. Depending on the simulated driving condition, the control intervention then takes place by control of the solenoid valve coils. At the same time, the HIL simulator records the magnetic fields in the coils and transfers the data via an A/D converter to the simulation environment. The brake hydraulics model calculates the brake pressures and transfers these to the vehicle model. By means of feedback to the control unit, the simulation loop between the ESP control and the HIL simulator is closed.
The cycle of signal pick up by the control unit, through calculation of all the models and signal feedback to the control unit takes places in real time with a cycle time of one millisecond. However, since the brake hydraulics model has a high degree of rigidity it was necessary to record the brake system signals at a frequency higher than 1 kHz. For this reason, signal pick up and calculation of the hydraulics model takes place with fourfold oversampling, i.e. a cycle time of 0.25 milliseconds.
Validation of the Simulation Environments
Before the actual HIL tests on the ESP control unit, the simulation environment of the HIL test bench is subjected to a thorough test. Simulation of the virtual Golf is validated using a selection of representative standardised driving cycles. For this, the results of the simulation are compared with those of the real driving test and model parameters are adjusted until a good match can be guaranteed between measurement and simulation. The same method is used to check the analogous valve data records of the hydraulics model.
The Test Programme using the Example of Vehicle-Trailer Combination Stabilisation
In the HIL simulation, the new ESP system is tested in numerous, virtually produced driving situations in order to make it compatible for use in the series production vehicle. Here, the virtual vehicle is sent for long periods on test routes such as the Nürburg Northern loop or one of VW's own test routes. During the test runs, the vehicle parameters, e.g. loading or road characteristics are systematically altered, which makes it possible to reveal possible weak points in the control units. The driving situations in which weak points can be revealed can be reproduced precisely in the simulation. This makes it easy to make optimum parameter settings for the control unit.
As an example of such performance tests, refer to the vehicle-trailer combination stabilisation tests, part of ESP functionality.
Lane changing, wheel ruts on the motorway or gusts of wind from the side can cause a trailer to sway. If a certain speed is exceeded - the critical speed - this swaying does not decrease, but instead increases. The consequence for the driver is generally that he can no longer control the vehicle-trailer combination (see the movie for this: Vehicle-Trailer Combination Simulation with and without Stabilisation).
The task of vehicle-trailer combination stabilisation is to dampen down the swaying movements through alternating braking operations and to brake the vehicle to a speed below the critical speed. Whether the ESP supplies this function in the required manner is tested in the virtual driving test. Various vehicle, trailer, ESP software and driving variables, which can lead to trailer swaying, are run and the relevant ESP control intervention is tested. Using simulation, the wide range of variables can be tested considerably quicker and with greater reproduction than by using real tests. The dangers to which drivers and vehicles are exposed in such tests on real routes can also be circumvented with the simulated driving test.
- Click on the image to play the movie (0:52 Minutes; 5,2 MB). The animation requires the installation of the DivX Codec - download for free under www.divx.com/divx/windows/.
Software-in-the-loop Simulation for the Steering System
At the moment, performance tests focus solely on braking interventions by the ESP system. A particular feature of the new ESP system is that the braking intervention by the control unit can be combined with a controlled steering moment intervention. Here, the ESP control unit sets steering moments which are converted by the electromechanical steering and superposed on the normal steering moments.
In order to exhaust the potential of this sort of functional integration of brake and steering, the available ESP HIL test bench is currently extended by adding the steering software-in-the-loop simulation (SIL simulation) function.
The steering system model will serve at first as a simple steering model in the CarMaker environment. The existing model is modified such that it takes into account the steering moment required by the ESP system as well as the steering moment intervention by the driver. At a later development stage, the steering system supplier will provide a model which will map even more precisely the electromechanical steering together with the controls.
So that the virtual driver reacts in exactly the same way to steering with interventions from the ESP system as a real driver, the driver model must also be extended in the CarMaker environment. For this, the behaviour of human drivers is measured in a real driving test and analysed for typical behaviour patterns. With the aid of the knowledge gained from this, a control strategy can be developed and implemented in IPG-DRIVER.
Once extensions to the test bench by adding the SIL simulation steering function have been completed, braking and steering interventions can be tested together. Here, it will be important to ensure that the interplay of both systems works perfectly.
Summary
HIL simulation is an important component of development work at VW. However, there is no desire to create the impression that simulation would replace test drives on real routes. Rather than that, the tests on both real and virtual test routes complement each other. If, during test drives, test drivers discover weak points in the setting of the control systems, these can be recreated in the simulation. The reproducibility of simulation results in virtual test drives makes it easy to investigate and correct these weaknesses systematically.
A testing tool that can be used consistently also enables the various development departments to work with the same tool in order to benefit from synergy effects. At VW, CarMaker is used in various development areas. This results in significantly reduced development times and creates an important competitive edge.
