A Neural Network Multiagent Architecture Applied to Industrial Networks for Dynamic Allocation of Control Strategies Using Standard Function Blocks

Abstract

This paper presents a multiagent architecture applied to factory automation. These agents detect faults in automated processes and allocate intelligent algorithms in field device function blocks (FBs) to solve these faults. We also present a dynamic FB parameter exchange strategy that allows agent fieldbus allocation. This architecture is a foundation for intelligent physical agents standard-based agent platform developed using Foundation Fieldbus technology. The aim is to enable problem detection activities, independent of user intervention. The use of artificial neural network (ANN)-based algorithms enables the agents to learn about fault patterns and adapt an algorithm that can be used in fault situations. Thus, we intend to reduce supervisor intervention in selecting and implementing an appropriate structure for FB algorithms. Furthermore, these algorithms, when implemented in device FBs, provide a solution at the fieldbus level, reducing data traffic between gateway and device, and speeding up the process of problem resolution. We also show some examples of our approach. The first is a neural network architecture change that allocates different types of neural networks in field devices without interrupting the fieldbus network operation. The second shows a multiagent architecture that implements the neural network change in a laboratory test process, where fault scenarios have been simulated.

Review on Application Research of Multi-Agent Theory in Industrial Process Control

“…To problems that in complex industrial process control, it is hard to create accurate system model, we can utilize learning, adaptation, self-organization and optimization characteristics of intelligent control, to design the control system based on rules from experiences, reasoning and intuition instead of on rigor analysis too much [17]. Algorithms in control theories based on multi-agent include reinforcement learning, evolution algorithm, neural network, fuzzy algorithm and simulated annealing algorithm [18][19][20][21][22]. Machado et al [20] use ANN to enable agents to learn about fault patterns and adapt the algorithm which can be used in fault situations in the presented multi-agent architecture applied to industry automation.…”

“…Algorithms in control theories based on multi-agent include reinforcement learning, evolution algorithm, neural network, fuzzy algorithm and simulated annealing algorithm [18][19][20][21][22]. Machado et al [20] use ANN to enable agents to learn about fault patterns and adapt the algorithm which can be used in fault situations in the presented multi-agent architecture applied to industry automation. Anderson et al [23] utilize reinforcement learning to make controllers learn by themselves, and obtain optimal control actions finally.…”

Implementação de uma rede neural em ambiente foundation fieldbus para computação de vazão simulando um instrumento multivariável

“…Ainda neste capítulo será explicado o queé o foundation fieldbus e como este foi utilizado no contexto de medição de vazão. O 42 treinamento das redes neurais foi desenvolvido fora do ambiente foundation fieldbus e, então, a topologia da rede neural mais os parâmetros de aprendizagem como os pesos sinápticos e limiares foram transportados para blocos funcionais em foundation fieldbus sendo que os principais blocos foram o aritmético e o caracterizador.O principal objetivo do trabalhoMachado et al (2010) foi criar uma arquitetura multiagente que permite incorporar nas redes industriais foundation fieldbus, aplicações inteligentes baseadas nos blocos funcionais de dispositivos de campo. Essa arquitetura viabiliza a alocação de recursos (algoritmos) em sensores e atuadores que permitem realizar algumas tarefas para auxiliar os supervisores a detectar e solucionar alguns problemas na rede como filtragem de ruído, predição A variação dos valores da temperatura e pressão também resultam em variações na densidade e viscosidade do fluido, isso acarretará uma variação no número de Reynolds que irá, por sua vez, provocar mudanças no valor do coeficiente de descarga.…”