Synthèse automatique de systhèmes de contrôle-commande spécifier en Graf-cet sur multi-cibles microcontrôleurs

Abstract

Embedded systems are ubiquitous in modern systems and have a built in logic controller that requires precision and meticulous design. These SCCs can be specified using several standard languages including Grafcet (IEC60848), which has become a widely used international formalism. As a for mal specification language, the manual implementation of the Grafcet is prone to errors and costly in time, hence the interest of automatic methods of generating code from Grafcet models..Generally, the researchers proposeto transform the Grafcet into PLC code of IEC 61131-3 standard, which makes the control solution still expensive because of the high cost of the PLCs. However, the continuous development in software engineering faci litates the emergence and rapid proliferation of a wide variety of low-cost processors for running programs in complex embedded applications. Thus, to reduce the cost of the control solution, the use of microprocessors may be preferred to ordinary PLCs in some embedded applications. The main difficulty lies in the production of the control software for a chosen micropro cessor target, given the wide variety of existing microcontrollers. Our goal is to use the Grafcet language to provide a programming environment for microcontroller based logic controllers. For that, it is important to find a mo deling taking into account all the aspects of the Grafcet, as well as the main characteristics of the microcontrollers targeted. Then, it will be a question of proceeding by successive refinements of abstraction levels starting from a Grafcet model to result in an implementation on a chosen microcontroller. First, we proposed a Grafcet matrix model, as well as a representation of the characteristics of the microcontroller targets, which were implemented through a multi-target synthesis environment. This environment combines the Grafcet model interpreter code generation mainly based on the interpre tation algorithm with Grafcet code generation, according to the model of Grafcet algebraic equations. A profiling of the codes generated in Arduino is carried out on the ATmega328P microcontroller at the end of which the estimated duration of the scanning cycle of the programs generated is of the order of 10 ms. However, this first solution presented limits mainly de pendent on the exhaustive taking into account of Grafcet elements in a matrix code and the implementation in a general environment. It is about the lack of genericity, the difficulties of maintenance and extension of the solutions of synthesis. For this, we carried out a Grafcet multi-target syn thesis by IDM approach, based on a proposed Grafcet metamodel and a metamodel of microcontrollers, which captures all the characteristics use ful for code generation. The instances of these two metamodels were then used as inputs of the transformation process (formalized by transformation rules) performed for the generation of the control code. To make it easier to edit Grafcet models, Grafcet expressions have been studied and forma lized through a free-context algebraic grammar, whose parser is integrated into the model editing and code generation process. For the implementation of the approach, we have developed a platform based on Eclipse EMF, the generator of ANTLR parsers, the OCL language and the Acceleo code gene ration engine. Initially, the validation of the proposed solutions was done by simulation and experimentation on the realization of a traffic light controller at a crossroads. The controller is then based on the EASYdsPIC4A micro controller and the application is emulated using an expansion card. Finally, we carried out a case study on an autonomous domestic water supply sys tem at the heart of which was developed and realized a switching model of energy sources, implemented on the Atmega1280 microcontroller after the generation of the corresponding Arduino code. It emerges that compared to Grafcet’s matrix-based synthesis approach, the IDM approach has many advantages: no need to calculate the encoding matrices, possibility to check the conformity of the model to the Grafcet language and great flexibility in the representation of Grafcet model expressions. However, there are ways to improve this work, here are some of them identified: taking into account Grafcet’s hierarchical structures in modeling, the production of a Grafcet graphic editor based on the proposed metamodel and the code generation for microcontrollers that can operate in disturbed environments.