Introduction of an ABS ECU hardware-in-the-loop test bench

frozenviolet

Introduction of an ABS ECU hardware-in-the-loop test bench [Copy link]

Abstract: This paper introduces the dSPACE real-time simulation system and Tesis simulation model, and introduces the ABS ECU hardware-in-the-loop test bench based on dSPACE of Audi AG.
Keywords: ABS; dSPACE; hardware-in-the-loop simulation

1 Introduction

For modern cars, the safety of the car depends heavily on the performance of the electronic control unit of the anti-lock braking system (ABS). The software of the anti-lock braking system controller (ABS ECU) is becoming more and more complex, so detailed testing is required in the early stages of ABS development. The cost of testing the ABS ECU on a real car is very expensive. In order to test under extreme conditions, a cold or hot environment is usually required, which is difficult for testers to achieve and will pose a certain threat to life safety. In addition, if there is a lack of prototype vehicles, the test will have to be postponed, which is contrary to the principle of concurrent engineering. In addition, testing with real cars has the disadvantages of poor repeatability and inability to reproduce the same test conditions. However, in any case, facing the market's strict requirements on vehicle reliability and safety and avoiding the risk of vehicle recalls, ABS ECU testing occupies a key position in the ABS development process. Faced with such a difficult problem, many manufacturers at home and abroad have used hardware-in-the-loop simulation (HILS) systems to implement ABS ECU testing. This article mainly introduces an industrial hardware-in-the-loop test bench for the anti-lock braking system of Audi Automobile Company in Germany based on dSPACE.

2 dSPACE Real-Time Simulation System

The dSPACE real-time simulation system is a set of software and hardware work platforms for control system development and testing based on MATLAB/Simulink developed by the German dSPACE company.
The hardware of the dSPACE system mainly includes a processor board based on PowerPC or DSP (such as TI's TMS320C40 DSP and TMS320F240 DSP, etc.), which is used to realize the real-time operation of the algorithm or simulation model designed by the user; in addition to the processing board, dSPACE provides a set of feature-rich I/O boards to facilitate users to process external physical system signals. The functions of these I/Os include analog input port (ADC), analog output port (DAC), digital I/0, serial interface, CAN interface, PWM output port, incremental encoder subsystem, etc. Users can arbitrarily combine between processors and between processors and I/O boards to form a standard component system as needed, or use a single-board system that integrates the two.
The dSPACE software system makes full use of the powerful functions of Matlab and expands on this basis. Its main software components include RTI (Real-time Interface), ControlDesk and MotionDesk. RTI is the link between dSPACE real-time system and software development tools MATLAB/Simulink. ControlDesk is a new generation of experimental tool software developed by dSPACE. ControlDesk can be used to achieve graphical management of real-time hardware, easy establishment of user virtual instruments, visual management of variables, visual management of parameters, and automation of experimental processes. MotionDesk can provide 3D animation effects for objects simulated online on dSPACE processor boards.
dSPACE real-time simulation system provides a hardware-in-the-loop simulation test platform. HILS is that after the controller is developed, the actual controller and the simulation model used to replace the real environment or equipment together form a closed-loop test system. Components that are difficult to establish mathematical simulation models (such as hydraulic systems) can be retained in the closed loop to perform simulation tests on the entire system. dSPACE's Simulator is the most widely used hardware-in-the-loop simulator. It is an integrated test environment, including: system models (including engine, vehicle dynamics and road models, etc.), real-time hardware, signal conditioning, fault simulation unit, load simulation unit, experimental software (including experimental management, hardware management, automated testing, etc.). dSPACE Simulator can be used to build systems of different sizes and functions according to the task requirements. From portable systems to 19' cabinets, dSPACE has provided more than 800 Simulators. Hardware-in-the-loop simulation uses the simulator to simulate a virtual vehicle, ensuring the real-time nature of the simulation during the test, adding some real components and loads, and simulating some faults, thereby achieving simulation testing of the ABS ECU.

3 Tesis simulation model

In the hardware in-loop simulation test of ABS ECU, an accurate vehicle dynamics model is required. However, the strong nonlinearity of the vehicle dynamics model makes it extremely difficult to accurately model it. For testers, it is impossible to have the energy to build such a model. Germany's TESIS, Mathworks and dSPACE have worked closely together to develop a set of dynamic models suitable for user simulation. Its products are divided into two parts: vehicle dynamics model (ve-DYNA) and engine dynamics model (en-DYNA).
It is sufficient to use only ve-DYNA in ABS ECU testing. ve-DYNA is a fast vehicle dynamics simulation software specifically used for real-time applications and offline research. The core of the software includes a high-precision vehicle model, different handling controls, three-dimensional road surface and virtual driver. The vehicle dynamics model describes the kinematic characteristics of the vehicle, in which the drive system model describes the important speed and torque values required for vehicle dynamics and engine control (including ABS control unit); the maneuvering controller includes controllers for longitudinal and lateral open-loop control and simple closed-loop control; the advanced road model provides a complete three-dimensional road profile and fully considers the geometry and surface characteristics of the road; the driver model is a controller that guides the vehicle along a given trajectory (set in the form of trajectory points) on the road surface.
ve-DYNA has a developed modular structure, the kernel is written in C, and the data flow and interface are represented by Simulink modules. This makes it very easy to add external or user-defined models. Its model definition and simulation control are both carried out through a graphical user interface. The graphical pre-processing and post-processing functions are implemented through a Matlab-based user interface, making user operation very simple.
In the test, the ve-DYNA model is used in conjunction with dSPACE Simulator to be the best real-time application integration environment. The semi-implicit integration algorithm and optimized vehicle model equations can ensure the data stability and efficiency of the simulation. During
simulation , up to 1500 model parameters can be tracked and recorded for further data analysis and visualization. Most parameters can be modified in real time, and the simulation results are displayed in 3D animation with the help of MotionDesk.

This example describes an industrial hardware-in-the-loop test bench for testing anti-lock brake systems, which has been installed at Audi AG in Germany.

4.1 Test bench principle

The Audi HIL test bench uses the ve-DYNA three-dimensional vehicle dynamics model developed by TESIS. The real Audi A8 hydraulic brake system or Audi A8 Quattro four-wheel drive hydraulic brake system is placed on a test bench, which is connected to the ABS ECU, dSPACE Simulator, and PC host to form a hardware in-the-loop test bench, as shown in Figure 2. In order to provide I/O signals to the ABS ECU like a real car when simulating the most complex vehicle configuration and operation, the simulation of the entire model must be completed within a 1ms step (less than the sampling time of the ABS controller). The dSPACE Simulator in this test bench uses five TMS320C40 DSP processors to jointly perform. The main DSP is responsible for calculating the drive trajectory model; two DSPs are used to establish the axis system; the other two DSPs write and read signals to the four slave processors. The five DSPs are in parallel. Under different configurations, the simulation execution time of the model ranges from 650μs to 940μs. In dSPACE Simulator, DDS is responsible for generating complex simulation signals for wheel speed sensors, and standard I/O is responsible for the input and output of other signals.

Figure 1 Schematic diagram of hardware-in-the-loop test bench

During the experiment, ve-DYNA was configured with the vehicle model and the car's operation. The model was first stationary for 1s, then accelerated for 5s, and after reaching a speed of 80km/h, it was kept at this speed for 5s, and then braked. Under this setting, hardware in-loop simulation, i.e., online testing, was performed. When braking begins, dSPACE Simulator transmits the braking signal to the brake pedal drive mechanism, executes the brake pedal action, the brake master cylinder supplies hydraulic pressure, and the pressure sensor measures the brake pressure signal of each brake wheel cylinder. Each brake pressure signal is converted into a corresponding digital quantity through the interface and sent back to dSPACE Simulator. These brake pressures will change the vehicle state, and then the new vehicle state is calculated through the model, and then converted into corresponding analog and digital quantities (such as wheel speed, lateral acceleration, engine speed, etc.) by the interface and sent to the ECU. The logical judgment and algorithm of the ECU's internal software issues a reasonable control command for the hydraulic regulator, and then changes the pressure of each brake wheel cylinder. The pressure sensor measures the new brake pressure signal and sends it back to dSPACE Simulator again. This cycle continues until the vehicle model is fully braked and the test experiment is completed.

4.2 Experimental results processing

During the experiment, the experiment file is created in ControlDesk, as shown in Figure 2. During the online test, various parameters of the vehicle (such as wheel speed, wheel acceleration, brake pressure, etc.) can be collected and monitored in ControlDesk. The performance of the ABS ECU can be analyzed through these collected data. In addition, it is also very convenient to change the vehicle's handling and road information in ControlDesk, such as changing the friction coefficient of the road, changing the vehicle's driving time, changing the road conditions, etc., so as to test the ABS ECU under various simulation conditions.
In ControlDesk, an automatic test tool - Test Automation is provided, which allows users to easily implement the entire process of experimental testing. The process of testing the ABS ECU on this test bench can be set as follows: start test - simulate driver behavior - check system response - diagnostic scan - automatic document generation - reset - change parameter variables - restart.
After a test is completed, the real-time data of the collected signal is transmitted to the MATLAB workspace. This data is sufficient for experimental analysis, but it is difficult to obtain a more intuitive vehicle movement situation. The MotionDesk 3D animation program is configured on the dSPACE simulator, as shown in Figure 3. The graphic images recorded during the test can be saved in a file. The recorded images can be replayed step by step to conduct a detailed analysis of the extreme test conditions.

Figure 2 Control diagram in ControlDesk Figure 3 Simulation animation reproduction diagram in MotionDesk

In addition to collecting real-time vehicle parameters for analysis, the experiment also tests the ABS ECU for faults (such as power short circuit faults) and the ABS ECU load working capacity (such as solenoid valve). The fault simulation part and the load simulation part are integrated on the dSPACE Simulator. Through the fault simulation module, 90% of the vehicle faults can be simulated. This test bench realizes the load working capacity test by connecting to a real load.

5 Conclusion

This article introduces the ABS industrial hardware-in-the-loop test bench of Audi AG, Germany. It uses the dSPACE software and hardware platform for developing and testing control systems combined with the Tesis model to realize the hardware-in-the-loop simulation function, and applies it to the testing of ABS ECU, realizing various extreme conditions, fault conditions and load tests of ABS ECU. The design idea of this test bench provides the feasibility for us to develop the test system of ABS ECU.