An identification method for PLC-based automated Discrete Event Systems

Abstract

Abstract-This paper deals with identification of automated Discrete Event Systems (DES). A method for processing sequences of input/output signals from PLC-based controlled DES is proposed; it yields an interpreted Petri net model describing the closed-loop behavior of the automated DES. This new method extends a previously presented approach by taking into account the technological characteristics of industrial controllers and the data collection requirements. Based on polynomial-time algorithms, the method is implemented as an efficient software tool; its use is illustrated through a case study dealing with an automated manufacturing system.

Stepwise identification of automated discrete manufacturing systems

“…In [4], [6], [10] are considered both inputs and outputs, but cyclic sequences are supposed to be known. Techniques in [7], [8], [9] process as input data a language generated by the system, which is given as events sequences, regardless how the events are obtained from I/O data.…”

“…Both controller's inputs and outputs are sampled from the initial state for building a single sequence of I/O vectors, which is processed yielding an interpreted Petri net (IPN) model. This approach is based on a previously presented efficient method for coping with concurrent partially observable DES [10] which processes a set of cyclic input/output sequences yielding models including silent transitions and non-labelled places. This technique has been extended and adapted for identifying actual PLCbased controlled discrete manufacturing systems, which operate during a long time period performing repetitive tasks.…”

“…This is due to the observation of input changes that do not affect the outputs. In order to obtain a more compact model a simplification strategy has been presented in [10] …”

Automated modelling of reactive discrete event systems from external behavioural data

“…In order to obtain a more compact model, a simplification strategy has been presented in [13] and is recalled below. It consists in merging several places, representing internal behaviour whose detected events do not have effect on the The application of this simplification procedure yields the IPN model showed in Figure 16.…”

“…Both controller's inputs and outputs are sampled from the initial state for building a single sequence of I/O vectors, which is processed yielding an interpreted Petri net (IPN) model. This approach is based on a previously presented efficient method for coping with concurrent partially observable DES [13]. The method is composed by several polynomial time algorithms that process a set of cyclic input/output sequences yielding IPN models including silent transitions and nonlabelled places.…”