A Formal Approach to Requirements Engineering of Automated Systems: Facing the Challenge for New Automated Systems

Abstract

It is a consensus that intelligent manufacturing plants should be automated, especially in what concerns automated processes to Industry 4.0. This new manufacturing approach is based on multifunctional distributed systems that, for its turn, depend on a sound design requirements phase. Consequently, the requirement specifications became even more significant in the design of intelligent manufacturing systems, where the possibility to analyze requirements previously could save time and effort. Frequently, relevant requirements assumptions change during the engineering process due to emergent and volatile requirements—called “scope creep.” A revision of the requirements approach should be demanding to minimize this problem. The proposal presented in this work anticipates the formal representation of automated systems requirements to allow the proper analysis and validation. Also, since conventional production lines have been replaced by a product-service (production) systems (PSS), there is pressure to review the requirements phase, mainly to automated systems. To increase the accuracy, we propose a requirements cycle composed of modeling, analysis, and (formal) verification where formalization is anticipated by capturing visual modeled requirements. Alternative approaches to the functional method are also explored, considering the Goal-oriented Requirements Engineering approach, the best fit to PSS. A formal process to treat requirements in a functional approach, represented by UML diagrams, is transferred to Petri Nets and submitted to formal analysis. Traditional translation from UML to Petri Nets is replaced by an object-oriented match to Petri Nets extensions covered by the standard ISO/IEC 15.909. A comparison between this approach and the goal-oriented raises a discussion about the requirements efficiency dilemma, where functionality could be pointed as the main reason to scope creep. Therefore, a strictly functional approach is proposed with a new transference algorithm, and the results are used to support the discussion about using goal-oriented alternatives. A case study of the chemical industry illustrates the practical use of the proposal.