Autotööstuse tarkvara mudelipõhine arendamine ja analüüs
Files
Date
2019
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Mudelipõhine arendamine ja analüüs on autotööstuses kasutatav uus meetod. Seda rakendatakse mootorsõidukite tootjate poolt, kuna hajusale komponentide arendusele sobib olemuslikult spetsifitseerimine musta-kasti printsiibil. Muud põhjused tulenevad survest toota kvaliteetset tarkvara, mis vastab kõigile regulatiivsetele standarditele, kuid mis sobib autotööstuse tootjate hinnamudeliga. Mudeli kasutamisel saab komponentide kehtivuse ja standardse vastavuse kontrollida enne, kui tegelik tarkvara on autosse paigaldatud.Mudeli kasutamine tekitab ka väljakutseid, et toota lõpuks tarkvara, mis kajastab täpselt mudeli toimimist. Mudelist automaatselt genereeritud tarkvara loetakse vastuseks, kuna see on stabiilne ja pärit juba kontrollitud mudelist. Kuna tarkvara muutub autotööstuses üha olulisemaks, muutuvad tarkvara loomise mudel ja genereerimise protsess üha keerulisemaks.Käesolev töö uurib mudelipõhist autotööstuse tarkvara arendamise ja analüüsimise protsessi - teisendades MATLAB/Simulink mudel AUTOSAR mudeliks. Lõputöö raames loodud programmid teostavad analüüsi erinevate teisendussammude tarbeks. Protsessi analüüsides selgus, et teisenduse meetoodika mõjutab oluliselt mudeli esitust ning ka lõpptulemuseks saadud AUTOSAR mudeli struktuuri. Näeme erinevaid võimalikke alternatiive sellele, kuidas mudelit saab vaadata ja muuta AUTOSAR-failiks. Selles lõputöös vaadeldud iteratiivne protsess pole lõplik ja seda saab veel täiustada.
Model-driven development and analysis is the state of the art method in the automotive industry. One of the reasons for its heavy utilization is coming from the black box nature of the components developed by the automotive vehicle manufacturers. The other reasons are coming from the pressure to produce quality software that complies with all regulatory standards but can fit the pricing model of automotive vehicle manufacturers.Validity and standard compliance of the components can be verified using models before the actual piece of software is deployed into an automotive vehicle. The utilization of the model also creates challenges: how to produce final software that precisely reflects how the model works. An automatically generated software from a model is deemed as an answer since it is coming from the already verified model and also will inherently retain consistency with the model. As software gets more and more critical inside an automotive vehicle, a model to create the software is getting more and more complicated and along with the automated software generation process.This thesis examines the model-driven development and analysis process for automotive software by conducting model conversion from MATLAB/Simulink model into AUTOSAR. The application developed for this thesis provides analysis and insights for every step of the conversion process. From the insights gathered along the process, it shows that the different model and transformation method creates a different model representation that affects the final structure of the AUTOSAR result. In the end, there are several possible alternatives on the way a model can be seen and transformed into an AUTOSAR file. It is also concluded that the iterative process in this project is not final and can be further improved.
Model-driven development and analysis is the state of the art method in the automotive industry. One of the reasons for its heavy utilization is coming from the black box nature of the components developed by the automotive vehicle manufacturers. The other reasons are coming from the pressure to produce quality software that complies with all regulatory standards but can fit the pricing model of automotive vehicle manufacturers.Validity and standard compliance of the components can be verified using models before the actual piece of software is deployed into an automotive vehicle. The utilization of the model also creates challenges: how to produce final software that precisely reflects how the model works. An automatically generated software from a model is deemed as an answer since it is coming from the already verified model and also will inherently retain consistency with the model. As software gets more and more critical inside an automotive vehicle, a model to create the software is getting more and more complicated and along with the automated software generation process.This thesis examines the model-driven development and analysis process for automotive software by conducting model conversion from MATLAB/Simulink model into AUTOSAR. The application developed for this thesis provides analysis and insights for every step of the conversion process. From the insights gathered along the process, it shows that the different model and transformation method creates a different model representation that affects the final structure of the AUTOSAR result. In the end, there are several possible alternatives on the way a model can be seen and transformed into an AUTOSAR file. It is also concluded that the iterative process in this project is not final and can be further improved.