NUMERICAL AND EXPERIMENTAL STUDY OF A HIL TEST BENCH FOR ELECTRIC WORK VEHICLES

The increasing use of electrification in both road vehicles and work vehicles has shown the need to use robust, efficient, and flexible testing methods to support the development and validation of electric vehicles (EVs). Today, validation methods involve physical prototyping and real-world testing, which is costly and slow. In this context, X-in-the-Loop (XIL) strategies play a key role in the design and testing cycle, providing tools for the analysis and verification of control units without the need for a physical prototype. Software-in-the-Loop (SIL) and Hardware-in-the-Loop (HIL) are key methodologies for validating complex systems, allowing the control software to be tested in a simulated environment and the real hardware to be integrated into a real-time simulation. The aim of this thesis work is to translate the functionalities of a real test bench into a SIL configuration, allowing tests to perform directly in virtual environment. The real test bench is the one built by the company Ecothea Srl, developed for the validation of electric powertrains designed for work vehicles (NRMM - Non-Road Mobile Machinery). Peak System's CAN-USB conversion devices are used to interface the test bench with the software on the PC. A SIL configuration of the test bench was required to test the accuracy of data exchange. Since the latter is the main objective of this work, the numerical model developed in Simulink/Simscape to reproduce the dynamic behavior of the test bench is very simple. It includes electric motors and inertial load to simulate the system dynamic, PI controllers for speed control, BUS DC for the power supply and energy recovery, and virtual CAN network management for data exchange. The MATLAB Vehicle Network Toolbox was used to configure the transmission and reception of CAN messages. To facilitate interaction with the system, a Graphical User Interface (GUI) has been developed in MATLAB App Designer, designed for setting the test parameters, monitoring the model’s variables in real time through the managing of the transmission and reception of CAN messages. The validation of the SIL configuration was carried out in different stages. After ensuring that the communication between Simulink model and GUI was correct, the model was enriched with strictly numerical models to simulate heavy data traffic on the CAN network, so that the data exchange between ECUs is accurately replicated as in the real test bench. The SIL model's behavior was evaluated with tests on real hardware using CAN-USB devices to transmit and receive messages after making modifications to ensure consistent data flow in real-time. The tests proved that the simulation and experimental data were consistent, highlighting the robustness of the software used for managing the message flow through CAN network. To sum up, this work illustrates how the switch from a physical test bench to a SIL configuration is an effective way to validate control and communication strategies based on CAN, decreasing development time and costs, and providing a versatile testing platform. For future development, the test bench model can be improved to accurately reproduce the behavior of the actual system. Integrating HIL architecture and replacing the Simulink model with a physical test bench will be the next step, allowing for direct testing of the control software under real operating conditions.