On Board Charger algorithm design compliant with the Model Based approach

This thesis aims at modelling an On Board Charger for an electric vehicle following the Model Based Design. This method allows to address complex problems, while making use of system models throughout the design, by means of their simulations it leads to a rapid prototyping, software testing and verification. The aforementioned task saw the realization and simulation of more than one system model, based on the provided requirements by Teoresi. The latter ones in turn were designed following the customer requests to implement a Charging Process Management solution. MATLAB and Simulink were used throughout the whole project, moreover, within each step of the Model Based Design workflow, it was made use of additional tools of the MATLAB environment. In particular the modelling phase hinged on Stateflow, whose graphical language allowed carrying out the logic of the Charger by way of a state transition diagram. As for the final one for each realised model at least one test harness was implemented, hence a group of properly designed inputs stimulated the Device Under Test (DUT) to take down a collection of outputs. By means of Simulink Test and the creation of test suites, running these ones against the model made possible to measure the model coverage thus performing the MIL testing. Meeting satisfactory results led up to SIL testing, that is to say running the same suites of tests against the auto-generated code of the model in order to measure the coverage of the code. Therefore the SIL and MIL testings were executed choosing three different coverage metrics, i.e. decision, condition and MCDC. Simultaneously with the testings, Simulink Check was employed to verify if the model complied with the ISO 26262 and to make the right adjustments for the purpose of being consistent with the safety standards related to an automotive application. Once ISO 26262 requisites were achieved, linking what follows with the previous statements, MIL and SIL testings coverage were compared to illustrate the missing coverage objectives for the auto-generated code and to eventually find a strategy to fill them. In the end the generated code was tested by exploiting its functions with the aim of programming a micro-controller where the code was loaded into.