Hardware-In-the Loop Simulation of Commercial Vehicles
차량에 대한 HIL 시뮬레이션 자료
나의 결론은 테스트 효율화를 위해 스테이트 머신 형태로 테스트가 진행되고 있다는 것
---
ABSTRACT
Modern day commercial vehicles are controlled by various Electronic
Control Units (ECU). They are not only tested as single units, but also by
networking them in Controlled Area Network bus (CAN) to form a complete
electrical control system. This is achieved using Hardware In the Loop (HIL)
Integration Lab. In HIL, the electrical system is connected to a real time
mathematical model of the vehicle plus it’s environment so as to form a loop.
Testing functionality of the electrical system begins by defining functional
tests. An example would be testing cruise control activation. Executing each test
is made possible by parameterizing variables in the vehicle dynamic model and
externally controlling them.
HIL based Verification and Validation (V&V) is moving towards
automation.
This is because of the complexity of electrical control systems is increasing and manual V&V is time consuming. In an automated test environment, a Test Engineer develops test scripts to implement functional tests. These test scripts execute the vehicle model in real time, control parameterized variables, and observe the electrical system response. This is compared to the expected response to decide if a functional test passed or failed.
Tests are designed to remain independent of each other. Scheduling of tests is done by the Test Engineer, which is a difficult task owing to their large number and possible combinations. Hence, the normal practice is to execute tests in a predefined sequence.
To solve the test scheduling problem in Hardware In the Loop simulation,
two solutions are proposed.
Both the solutions exploit relationship between test case and state of the vehicle in a dynamic simulation environment. An example of such relationship is engaging cruise control only when vehicle speed is above 20 km/h. It can be proved that a test process that is sensitive to the simulation environment will be more realistic and hence efficient.
One solution is to model the test execution as a state machine. Tests are
treated as states. Entry conditions for each state are defined using state variables
of the dynamic model. When a simulation is run, state variables of the dynamic
model are sampled in real time. One sample of state variables trigger a transition
from one state to another in the state machine. When the state machine is in one
state, a test case corresponding to that state is selected and executed. A sequence
of these transitions results in a test process evolving in time.
The second proposed solution is functionally similar to a state machine but
it’s implementation is derived from logic design. Here, one sample of state
variables is compared with entry conditions of each test case. Test cases whose
entry conditions match with the current sample are selected for execution.
Both the solutions use Failure Mode Effective Analysis (FMEA) to resolve test selection conflicts, that is, situations where more than one test is selected.
Results show that test execution using this approach is sensitive to the simulation environment and comparable to that of a real test drive scenario. An
improvement in test efficiency both in Qualitative and Quantitative terms is also
achieved. Test runs show how the new method of test execution allows faults to
propagate from one test to another like in a real test drive.
