Transfer Learning in Wastewater Treatment Plant Control Design: From Conventional to Long Short-Term Memory-Based Controllers.

Ivan Pisa,Jose Lopez Vicario,Ramón Vilanova,Antoni Morell

doi:10.3390/s21186315

Ivan Pisa, Jose Lopez Vicario + Show 2 more

Open Access

https://doi.org/10.3390/s21186315

Copy DOI

Abstract

In the last decade, industrial environments have been experiencing a change in their control processes. It is more frequent that control strategies adopt Artificial Neural Networks (ANNs) to support control operations, or even as the main control structure. Thus, control structures can be directly obtained from input and output measurements without requiring a huge knowledge of the processes under control. However, ANNs have to be designed, implemented, and trained, which can become complex and time-demanding processes. This can be alleviated by means of Transfer Learning (TL) methodologies, where the knowledge obtained from a unique ANN is transferred to the remaining nets reducing the ANN design time. From the control viewpoint, the first ANN can be easily obtained and then transferred to the remaining control loops. In this manuscript, the application of TL methodologies to design and implement the control loops of a Wastewater Treatment Plant (WWTP) is analysed. Results show that the adoption of this TL-based methodology allows the development of new control loops without requiring a huge knowledge of the processes under control. Besides, a wide improvement in terms of the control performance with respect to conventional control structures is also obtained. For instance, results have shown that less oscillations in the tracking of desired set-points are produced by achieving improvements in the Integrated Absolute Error and Integrated Square Error which go from 40.17% to 94.29% and from 34.27% to 99.71%, respectively.

Highlights

Industrial environments are characterised by running complex and repetitive processes which are sometimes maintained over time
The performance of the TL-based Control Design approach is determined by means of analysing the control performance of each one of the proposed TL approaches: (i) the Transfer Learning from Dissolved oxygen (DO) to NO, (ii) the Transfer Learning from NO to DO, and (iii) the Long Short-Term Memory (LSTM)-based controller Fine-tuning and Transfer
Results will show which is the best option to obtain a complete and good control approach mainly based on data, and to speed-up the design process of the complete Wastewater Treatment Plant (WWTP) control strategy

Summary

Introduction

Industrial environments are characterised by running complex and repetitive processes which are sometimes maintained over time. The incursion of the Industry 4.0 paradigm and Artificial Neural Network (ANNs) applications are changing the way we control and manage industrial environments. Their main aim is to provide the industries with solutions mainly based on measurements obtained from their systems [2]. Some of these solutions go from basic forecasting systems to more complex solutions, like predictive maintenance ([3], Chapter 9). But the industrial control domain is experiencing a change in its tendency: ANNs are used more and more as the main control structures than conventional controllers

Methods

Results

Conclusion