Learning-based tuning of supervisory model predictive control for drinking water networks

Abstract

This paper presents a constrained Model Predictive Control (MPC) strategy enriched with soft-control techniques as neural networks and fuzzy logic, to incorporate self-tuning capabilities and reliability aspects for the management of drinking water networks (DWNs). The control system architecture consists in a multilayer controller with three hierarchical layers: learning and planning layer, supervision and adaptation layer, and feedback control layer. Results of applying the proposed approach to the Barcelona DWN show that the quasi-explicit nature of the proposed adaptive predictive controller leads to improve the computational time, especially when the complexity of the problem structure can vary while tuning the receding horizons.

A service reliability model predictive control with dynamic safety stocks and actuators health monitoring for drinking water networks

“…Principal component analysis (PCA) preprocessing is applied to the training patterns. See, [4] for more details.…”

“…This approach is too conservative and reduces the manoeuvrability space for economic optimization since the full tank excursion is limited and its capacity is not usable to save energetic costs in pumping actions [4]. On the other hand, lead time do vary over when capacity is limited or time varying, nevertheless, models that take into account non-stationary behaviours are not completely helpful if they use this information to just calculate safety stocks, especially when variations are caused by network supply components ageing.…”

Water demand forecasting for the optimal operation of large-scale drinking water networks: The Barcelona Case Study.

“…Drinking Water Networks (DWN) are large-scale, multipleinput multiple-output systems whose operation is liable to a set of operating, safety and quality-of-service constraints while at the same time their dynamics is affected by disturbances of stochastic nature (see Brdys and Ulanicki [1994], Grosso et al [2013]). All these characteristics render their control a challenging problem.…”

On the Comparison of Predictive Control and Command Governor Approaches for Operational Management of Drinking Water Networks: A Case Study

“…As discussed in Pulido-Calvo and Gutiérrez-Estrada [2011], during the last years optimal operation of water supply systems has been addressed by a wide variety of methods, ranging from heuristics and expert systems to more advanced mathematical modelling and optimization techniques such as linear programming, dynamic programming, non-linear programming, hierarchical-decompositions, combinatorial schemes, and more recently, model predictive control (MPC). Examples of some of the aforesaid methods are reported in Castelletti et al [2012], Cembrano et al [2011], El Mouatasim et al [2012], Ocampo-Martinez et al [2013b], Pulido-Calvo et al [2012], Vieira et al [2011], Grosso et al [2013], among many others.…”