The growth of a new car involves endless exam drives totalling several million kilometres with prototypes in a several stages of development. It is a usually approach to put all a systems by their paces and an constituent partial of a altogether growth process. However, advances in complicated motorist assistance systems call for opposite strategies. For instance, engineers would have to exam a semi-autonomous pushing duty on a highway for several years, though mechanism simulations are increasingly entrance into concentration – including during Porsche Engineering
The group uses hardware-in-the-loop (HiL) opposite many projects. It is a routine that has proven itself in member tests and is increasingly being used for rarely formidable systems as well.
HiL is a couple between program and a aim hardware in a vehicle. The HiL exam dais acts as a reproduction of a genuine sourroundings of an electronic control section or an ECU network. The wiring of a connected ECU are granted with electrical control signals, while a signals of a train systems are granted by a residual train simulation. For a ECU, this state can't be renowned from operation in a vehicle. For exam purposes, it is also probable to selectively request improper information to a vehicle’s sensor interfaces and information buses, for instance to copy a effects of a poor crankshaft or a finish disaster of another car system, as a means of validating a ECU’s diagnostics.
Library with make-believe models
The basement for a HiL simulations is minute mathematical program models, that are combined with Matlab/Simulink and imitate a earthy processes in a vehicle. They are stored in libraries from that a modellers can remove suitable modules – formed on a engine several or framework type, for instance – and mix them to form an altogether make-believe model. Among other things, a library contains models for a poise of an intake complement or a explosion chamber, though also formidable models of several framework with or but adaptive atmosphere suspension, anti-roll stabilisation, and a far-reaching accumulation of wheels.
Access to a libraries means that not each HiL make-believe indication has to be reprogrammed. Instead, it is sufficient to configure and parameterise a existent modules. However, replicating a earthy processes in a car is really computationally finish and requires an suitable infrastructure. “We have commissioned several really absolute computers in a HiL exam benches and can now simply imitate sold sub-models physically and rightly in genuine time. There are also models of a whole car network, a driver, and a environment. But in some formidable processes, such as a whole explosion cycle in a engine, we now still have to find a workaround,” says Jörg Turowski, who is a dilettante plan manager during Porsche Engineering obliged for powertrain HiL systems.
“Some of a Hil tests would be compared with a high risk of personal repairs and skill repairs in a vehicle.”
Jörg Turowski, Specialist Project Manager for powertrain HiL systems
That’s because a HiL experts use neural networks, that need significantly reduction computing energy than earthy models. Among other things, they are used to facilitate earthy models. In a make-believe indication of an engine, for example, a method of a explosion routine is distributed from a control signals of a engine control unit. The make-believe provides values such as a generated torques and sensor signals as outlay variables, that are fed behind into a engine control unit. By regulating a neural network, these outlay values do not have to be distributed regulating a physically scold formula. Instead, a neural network provides them in genuine time. To do this, however, it contingency initial be lerned with a information from a real antecedent engine or a earthy indication that is not able of real-time execution. Porsche Engineering has been regulating neural networks really successfully in a make-believe models for years.
Measurement and control
Sensors supply a submit for a electronic control unit, that afterwards compares a totalled tangible values with aim values. If a tangible values deviating from a aim values, a control section triggers corrections around actuators.
The possibilities for validating program are roughly sum interjection to HiL exam benches. “With hardware-in-the-loop, even exam cases that are really formidable to inspect on a highway or in margin tests can be analysed,” explains Heiko Junker, Senior Manager Powertrain Electronics during Porsche Engineering. This relates in sold to a investigate of impassioned situations.
“Some of these tests would engage a high risk of personal repairs and skill repairs in a vehicle,” adds Turowski. “But with hardware-in-the-loop, they can be steady as mostly and accurately as desired.” As examples, he cites tests during high speeds, deliberately surpassing limits, and contrast program responses to signals that are usually triggered in a eventuality of an accident. “This is important, for example, to countenance protecting functions,” says Turowski.
When regulating hardware-in-the-loop, Porsche Engineering advantages from a many years of knowledge in a margin of simulation. In addition, a matching simula- tion models are used in a finish validation and formation process. “This allows us to feat many synergies,” says Junker. “We are also aided by entirely programmed processes, from a specifications to a exam result. We use exam selection generators and exam box generators for this.”
Customers advantage from a one-stop service: Porsche Engineering offers them a finish package – from a judgment to pattern and commissioning to residual train make-believe and earthy modelling. Added to this is a minute research of a exam results, including recommendations on probable calming measures.
Several Porsche Engineering locations work closely together on HiL projects. This intercontinental team-work on a basement of a same processes, techniques, and methods gives Junker and Turowski a event to discharge tasks worldwide according to a accessibility of resources and imagination and with due courtesy for a special facilities of a markets.
“We are invariably expanding a Hil capacities.”
Heiko Junker, Senior Manager Powertrain Electronics
For example, while a site in Mönsheim provides a HiL exam benches for member and duty developers and develops a concepts for destiny HiL systems, a collection for programmed exam box doing and analysis are invariably optimised in Prague, as good as in-house HiL systems for exam box execution. The group in Shanghai, on a other hand, specialises in a origination of earthy make-believe models and a formation and training of neural networks.
“We can also take advantage of a time disproportion between Germany and China,” Junker says. “If we find an blunder in a make-believe indication here during HiL tests, it can be analysed overnight in Shanghai.” There, as good as during a other general locations in Cluj, Prague, and Ostrava, Porsche Engineering maintains laboratory space with a sum of some-more than 60 HiL exam benches. In addition, there are another 40 or so exam benches in Germany, creation it probable to accommodate a usually augmenting demand. In a future, a use of HiL record will continue to increase, as a trend towards fewer and fewer genuine prototypes will continue. “This is because we are invariably expanding a HiL capacities,” says Junker.
The computing energy compulsory for rarely formidable make-believe models is available, as it can be increasing roughly indefinitely by relocating tools of a indication to a cloud. “Moreover, a ECUs during a exam benches are already connected to a backend today,” says Turowski. “So cloud functions can also be certified during an early stage.” Even a finish deputy of prototypes seems conceivable: “We are operative on joining several member HiL systems in a cloud to form a practical finish vehicle,” says Junker.
Text: Andreas Burkert
Text initial published in a Magazine Porsche Engineering, emanate 2/2021.