Abstract

The increase of applications for industrial smart sensors is booming, mainly due to the use of distributed automation architectures, industrial evolution and recent technological advances, which guide the industry to a greater degree of automation, integration and globalization. In this research work, an architecture for deliberative-type intelligent industrial sensors is proposed, based on the BDI (Belief Desire Intentions) model, adaptable to the measurement of different variables of the filtering process of a water purification plant. An intelligent sensor with functions of signal digitalization, self-calibration, alarm generation, communication with PLC, user interface for parameter adjustment, and analysis with data extrapolation have been arranged. For decision making, the use of fuzzy logic techniques has been considered, which allows imprecise parameters to be appropriately represented, simplifying decision problem solving in the industrial environment, generating stable and fast systems with low processing requirements. The proposed architecture has been modelled, simulated and validated using UML language in conjunction with Petri nets, which facilitate the representation of discrete system events, presenting them clearly and precisely. In the implementation and testing of the prototype, C/C ++ language has been used in an 8-bit microcontroller, experimentally corroborating the operation of the device, which allowed evaluating the behavior of a pseudo-intelligent agent based on the requirements of the water treatment plant, and also through comparisons with similar works developed by other researchers.

Highlights

  • Advances in industrial automation, along with computing, digital communications and microelectronics, are evolving towards a new industrial revolution that promises to lead profound changes in the industrial and manufacturing sectors [1], becoming a priority of many companies [2]

  • The paradigms of industrial automation are oriented to the distribution of artificial intelligence (AI) among all the components of the factory [7], which have been very useful supports in decision-making, demanding tasks or in case of risks for workers or operators [10], reasons for which they focus on models that adapt to the conditions of an intelligent distributed network [11], Within distributed AI, agents and multi-agent systems represent an ideal solution to design systems based on this paradigm [12,13], supported by the BDI model (Beliefs, Desires and Intentions) allow to systematically describe the behavior of each intelligent agent in the network [14]

  • The process begins with the definition of the initial considerations of the smart sensor, the modeling and verification of the system is developed using Unified Modeling Language (UML) and Petri Nets (PN) and the implementation of the level sensor using the fundamentals of fuzzy logic

Read more

Summary

Introduction

Along with computing, digital communications and microelectronics, are evolving towards a new industrial revolution that promises to lead profound changes in the industrial and manufacturing sectors [1], becoming a priority of many companies [2]. The proposed smart sensor behaves like a sensor node, because, despite being autonomous, it receives instructions and is supervised by a parent node [21], which in this case is a programmable logic controller (PLC) This sensor, due to its architecture, is not an intelligent agent, but it has many of its characteristics, for being considered a pseudo agent, and for this reason, in its projection and design, an intelligent agent methodology is used. This architecture consists of a belief database that constitutes a set of parameters stored within the sensor, of which, one group is configurable to adapt to the type of sensor, while the other group is fixed or general At this stage, the behavior selection and event dispatcher functions have been arranged, which correspond to the wishes or objectives according to the BDI architecture, while the functions of self-calibration, analysis and average of values, extrapolation and data generation, respond to the intentions you have, based on the objectives. The sensor output to the PLC is done through the MODBUS RTU protocol

MODBUS RTU
Filter level
Input Variables
Me Hi
Conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call