Published: November 13, 2014
Amanjot Dhaliwal, Team Leader, dSPACE Inc.
Hardware-in-the-loop (HIL) test systems are increasingly being used in the development and testing of embedded software in new construction and agricultural vehicles. These off-highway vehicles are just as complex, if not more so, as their automotive counterparts. This is due to the increasing amount of embedded software used on-board the vehicle for the safe and reliable performance of these machines working in very demanding environments. In addition to the embedded control systems for powertrain, vehicle dynamics, and body electronics found in automobiles, off-highway vehicles also need to support:
Due to high torque demands, it is very common for off-highway vehicles to have diesel engines with 12 or more cylinders. These large engines include precise control systems like direct injection, EGR, etc., to meet the emission regulations, as well as fuel efficiency expectations. These technologies, powered by electronic control units (ECUs) and embedded software, require additional considerations in the design of the HIL test system hardware (e.g. current measurement on the injection lines and injection waveform capture). dSPACE developed special solutions like the DS5377 board, which provides current measurement and fault simulation on 32 channels, to meet such demanding requirements.
A key component of the HIL simulation technology for embedded software testing is the models of the physical systems like engines. In order for these models to be calculated, the dynamic and physical states of the system, as measured by sensor and actuator signals, have to be known. For example, the measured current drawn by an injector is an indication of its state, which makes it possible to calculate the amount of fuel injected in that particular cylinder. This further feeds into the overall engine torque calculation. However, in order for the calculated model to be accurate, such signals, which are quite complex, have to be measured using specialized hardware.
dSPACE HIL systems include specialized IO components like the complex comparator on a DS2211 board that reads the injector current waveform and infers the starting angle and duration of injection. This information can be used in an engine model for torque calculation.
It is also critical to validate the current levels and durations through the peak, boost, and the hold phases of the injection waveform. The injection current waveform can be monitored on an oscilloscope to check the profile, but that is a manual process and can be very time consuming. On a HIL, the injection current waveform can be routed to a high-speed ADC board like the DS2004. These captured waveforms can be validated directly in the model or post processed to compare against a desired/ideal profile.
With tightening emissions regulations, the exhaust aftertreatment systems for diesel engines have accumulated more active, electronically controlled components. HIL testing can save time and cost associated with number of dynamometers, operations and maintenance. Using HIL test systems, it is possible to recreate any test scenarios in the lab for automated verification of embedded software, and with high repeatability and reliability of tests.
While in the past most of the aftertreatment control could be handled by the engine controller itself, now most applications require a dedicated controller for the aftertreatment system. An HIL test system for exhaust system testing mainly requires the simulation of various aftertreatment components:
By using the exhaust systems models from the dSPACE ASM library, users can quickly assemble and parameterize a model of their aftertreatment system. Since these models are built and tested for real-time operation, further changes to the model are not required for execution in real-time on a HIL platform.
Implements are the various machinery components, such as the boom, arm or buckets on an excavator. These Implements have traditionally been controlled with hydraulic actuators, which can provide the fine control and high actuation forces/torques. The ECUs control the current through proportional or on/off type valves to direct the flow of hydraulic fluids through these actuators. For HIL simulation, this requires measurement of the current to infer the state of the corresponding valve. This information can then be used in modeling of the vehicle’s hydraulic system. In this respect, the implement and the transmission control signals are not very different.
Closed-loop modeling of the implement hydraulics is necessary to test the responsiveness of the system and the ripple effects of electrical faults to ensure safe operation. Hydraulic system modeling particularly presents difficult computational challenges because of complexity in modeling the fluid dynamics. Modeling of hydraulic systems typically results in complex, higher order differential equations that are computationally intensive and challenging using the fixed-step solver required by real-time systems. Tools like SimHydraulics™ and Amesim™ provide intuitive means to model a hydraulic actuation system with easy-to-use building blocks. Being able to use this approach in a HIL simulation model is very attractive, but also presents some challenges while running a real-time simulation. Because the system is modeled with the state-space approach, the state-space parameters have to be recalculated every time there is a change in the state resulting in a spike in the model execution time, which can cause the real-time simulation to overrun.
While engine simulation for off-highway vehicles is not very different from the automotive industry, modeling the vehicle dynamics presents a number of different challenges. Because the vehicles are mostly driven in dirt, the tire-soil or track-soil interaction needs to be modeled. While most automotive vehicle dynamics models just assume a fixed coefficient of friction on a patch of road, the tire-soil model is non-linear in nature and needs to account for factors like sinkage, contact patch variations, and combined slip.
Additionally, off-highway vehicles have a lot more variability in loading conditions because of the position of the attached implements and the amount of load carried by the vehicle. These variations need to be supported by the real-time models so the user can validate the embedded control system performance under all applicable loading conditions.
Not only are these factors are difficult to model, but including these effects in a dynamics model of the vehicle can make it very computationally intensive. dSPACE, through years of experience working with our customers, has developed techniques to address these challenges using one or more of the following methods:
Off-highway vehicles seldom use a Global Positioning
System (GPS) for navigation in the conventional sense as automobiles.
Instead, it is commonly used for complete autonomous driving or to
accurately position the end effectors of the vehicle. In agricultural
equipment, this allows the farmers to control the positioning of sowing
equipment and to evenly spray the optimal amount of fertilizer and
pesticides. In construction equipment, it could allow the engineers to
precisely control how a road gets laid out or prevent the vehicle from
accidentally crossing into a prohibited area.
Because the GPS is so integral to the operation of most off-highway vehicles, their embedded controllers rely heavily on the GPS signal for implement control functions. The GPS receiver is external to the ECU and sends the position information to the ECU via serial or CAN-based protocols. From a HIL point of view, we can either disconnect the GPS receiver and simulate the protocol or simulate the actual satellite signals in real-time for any defined path and location for the GPS receiver.
The dSPACE GPS simulator solution allows the users to communicate to a Spectracom™ GSG-62 unit via Ethernet. This interface can be used to send the vehicle position or velocity to the GSG-62 unit, which generates a GPS or a GLONASS signal. The GPS signal generator, in conjunction with a real-time vehicle dynamics model, generates a valid position signal for the ECU.
For many years, dSPACE has been closely working with global off-highway OEMs and suppliers to optimize their development and testing of embedded software. We have gained a lot of experience to address this challenging application domain and provide test systems that are used extensively to improve product quality and safety, at a relatively low cost and with high efficiency.