Leakage reduction is an extremely important goal in the management of Water Distribution Networks (WDNs). Due to the dependence of leakage on pressure, Real Time Control (RTC) represents an effective tool to mitigate leakage by dynamically removing pressure excess as the users demand varies during the day. Cost-effectiveness of RTC can be improved by transforming part of the pressure excess into electric energy. This result can be achieved by means of Pump-as-Turbines (PATs). This paper proposes a novel RTC scheme able to simultaneously regulate pressure at multiple WDN nodes, and recover part of the excess energy, which can be sent directly to the main electrical grid. The control scheme is based on a Kalman Filter for joint state and disturbance estimation, a tracking Model Predictive Controller for multi-output regulation, and an Actuator Settings Optimiser to adjust the PAT settings. Thanks to its modular structure, the algorithm is scalable, flexible, and computationally cheap. The effectiveness of the proposed RTC scheme is demonstrated with several closed-loop simulations, carried out on a detailed, pressure-driven, unsteady flow model of an existing WDN. Moreover, the proposed algorithm can be directly applied to real WDNs, as it does not require any hydraulic model of the entire plant to be designed and tuned.

In this paper, we analyze stability of nonlinear model predictive control (MPC) using data-driven surrogate models in the optimization step. First, we establish asymptotic stability of the origin, a controlled steady state, w.r.t. the MPC closed loop without stabilizing terminal conditions for sufficiently long prediction horizons. To this end, we prove that cost controllability of the original system is preserved if sufficiently accurate proportional bounds on the approximation error hold. Here, proportional refers to state and control. The proportionality of the error bounds is a key element to derive asymptotic stability in presence of modeling errors and not only practical asymptotic stability. Second, we exemplarily verify the imposed assumptions for data-driven surrogates generated with kernel extended dynamic mode decomposition based on Koopman operator theory. Hereby, we do not impose invariance assumptions on finite dictionaries, but rather derive all conditions under non-restrictive conditions. Finally, we demonstrate our findings with numerical simulations.

Type 1 diabetes (T1Ds) is a chronic disease characterized by the absence of insulin production, causing high blood glucose (BG) levels. Although exogenous insulin therapy helps control glucose, its effectiveness is limited by patient-specific variability, particularly the circadian rhythm of insulin sensitivity. This work investigates periodic glycemic trajectories (basal trajectories) as stabilizable targets for model predictive control (MPC). Two stable control formulations are proposed: one using basal trajectories as terminal constraints, and another introducing auxiliary trajectories to expand the domain of attraction. Preclinical in silico tests under varying insulin sensitivity patterns demonstrate the benefits of explicitly incorporating circadian variability into the controller design.

This article proposes an offset-free, output feedback, tracking model predictive control (MPC) stabilizing formulation, specifically designed to handle incrementally input-to-state stable (delta ISS) systems subject to input and input rate constraints. Recursive feasibility and stability are guaranteed by means of suitable terminal ingredients, and an extended region of attraction is provided by means of artificial reference variables. Moreover, the knowledge of a suitable cost detectability function enables the use of a positive semidefinite stage cost (e.g., for output weighting), which can greatly simplify the controller tuning in case of high dimensional and/or black-box systems. Furthermore, offset-free tracking of asymptotically constant reference signals can be achieved even in presence of asymptotically constant disturbances, by means of a state observer that estimates the system state and output disturbances. The proposed MPC formulation is applied to control both linear and nonlinear systems, with two case studies inspired by remote pressure control in water systems and pH neutralization processes. In particular, the second case study also discusses how to combine the proposed formulation with recurrent equilibrium network (REN) models.

This paper derives conditions under which Model Predictive Control (MPC) with terminal conditions, using a data-driven surrogate model as a prediction model, asymptotically stabilizes the plant despite approximation errors. In particular, we prove recursive feasibility and asymptotic stability if a proportional error bound holds, where proportional means that the bound is linear in the norm of the state and the input. For a broad class of nonlinear systems, this condition can be satisfied using data-driven surrogate models generated by kernel Extended Dynamic Mode Decomposition (kEDMD) using the Koopman operator. Last, the applicability of the proposed framework is demonstrated in a numerical case study.

Glucose dynamics in type 1 diabetes are highly variable both across individuals and within the same individual throughout the day, due to factors such as meals, insulin sensitivity, and daily routines. This variability poses significant challenges for accurate prediction and control, limiting the effectiveness of single-model approaches. The aim of this work is to develop control-oriented models to provide accurate predictions of glucose dynamics, to be used within control strategies, such as model predictive control. The proposed approach adopts a periodic structure, based on multiple models, that accounts for both individual differences, in terms of metabolic response, and daily fluctuations in patient dynamics. Models are identified in different daily periods using an impulse response method applied to data generated by the UVA/Padova simulator. The day is divided into three time segments corresponding to breakfast, lunch, and dinner, with a separate model trained for each period. These models are integrated through a soft-switching mechanism. Two state estimation techniques, the Kalman filter and the moving horizon estimator, are compared to enable multi-step glucose prediction. Results indicate that the periodic predictor consistently outperforms the invariant predictor approaches across the entire virtual adult population. This modeling framework shows strong potential for integration into advanced insulin delivery systems based on predictive control. (c) 2026 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

In this technical communique, the properties of interconnected Incrementally Input-to-State Stable (SISS) discrete-time systems are studied. In particular, a small gain theorem for SISS systems with feedback connection is derived with two formulations: one based on the definition of SISS, and one based on SISS-Lyapunov functions. (c) 2025 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

In this brief, the use of velocity form model predictive control (MPC) for the control of currents in synchronous reluctance motors is studied. Stability is guaranteed by means of a terminal equality constraint, and an artificial reference is introduced in the optimization problem to enlarge the feasibility region and to improve the performances of the closed loop. The velocity formulation of the MPC guarantees zero tracking error in the presence of model uncertainties. The velocity form MPC algorithm is compared, in simulation, with a standard control strategy based on decoupling and proportional-integral (PI) controllers and with a classic MPC, also in the presence of uncertainties in the inductance values of the motor model. The velocity form MPC shows better performances in the presence of model uncertainties and does not require the knowledge of the relationship between the inductances and the currents for its implementation.

The Dissolved Oxygen (DO) in the reactor plays a key role in modern wastewater treatment based on Activated Sludge Process (ASP). This paper proposes a methodology to design a DO controller based on simple black-box model identification that can be applied to any ASP plant starting from standard measurements produced applying a simple ON-OFF controller to the plant. The proposed control scheme is easily implementable on a commercial Programmable Logic Controller and SCADA system. It is tested on different technologies: a simulated Conventional Activated Sludge and a real full-size Thermophilic Aerobic Membrane Reactor (TAMR) plant. A six-month experiment in TAMR has shown an improvement in the reduction of both Chemical Oxygen Demand (COD) (80. 9% vs 66. 8%) and N0(3) (94. 5% vs 85. 3%).

In this paper we propose a robust Model Predictive Control where a Gated Recurrent Unit network model is used to learn the input–output dynamics of the system under control. Robust satisfaction of input and output constraints and recursive feasibility in presence of model uncertainties are achieved using a constraint tightening approach. Moreover, new terminal cost and terminal set are introduced in the Model Predictive Control formulation to guarantee Input-to-State Stability of the closed loop system with respect to the uncertainty term.

This article proposes a novel scheme for the real-time control (RTC) of service pressure in water distribution networks (WDNs), which is beneficial in terms of leakage reduction, energy recovery, pipe burst abatement, and extension of infrastructure lifetime. Compared with the other schemes previously proposed in the scientific literature, this novel scheme combines regulatory performance with proven guarantee of stability, which is obtained by framing gain scheduling in the context of internal model control (IMC) of linear parameter-varying (LPV) systems. Previous works relying on gain scheduling only prove stability for fixed scheduling parameter values, which is only a necessary condition for stability in case of possibly fast, time-varying parameters. The proposed RTC scheme guarantees instead stability of the closed loop for any admissible trajectory of the scheduling parameter. The novel control scheme is tested numerically against challenging operating conditions in a benchmark WDN, including two different demand patterns and four hydrant activation scenarios.

This paper develops a control scheme, based on the use of Long Short-Term Memory neural network models and Nonlinear Model Predictive Control, which guarantees recursive feasibility with slow time variant set-points and disturbances, input and output constraints and unmeasurable state. Moreover, if the set-point and the disturbance are asymptotically constant, offset-free tracking is guaranteed. Offset-free tracking is obtained by augmenting the model with a disturbance, to be estimated together with the states of the Long Short-Term Memory network model by a properly designed observer. Satisfaction of the output constraints in presence of observer estimation error, time variant set-points and disturbances is obtained using a constraint tightening approach.

This paper presents a novel pulsatile Zone Model Predictive Control (pZMPC) for glycemic control in type 1 diabetic patients, which is an extension of the one presented in literature. Its main characteristics are (i) the explicit inclusion of a time-varying insulin on board constraint to promote a non-zero insulin delivery after a standard bolus infusion and to increase the postprandial system controllability, and (ii) the softening of this constraint to ensure closed-loop stability for an enlarged domain of attraction. Additionally, configuration methods are proposed for the insulin on board constraint based on carb counting regression models and glycemia rate of change information. In-silico results in the FDA-approved UVA/Padova simulator, under over/underestimation of the meal amounts, and under insulin sensitivity variability, show a superiority of the proposed glucose control over postprandial periods.

This paper proposes a novel methodology for simultaneously addressing design of burst detection/localization machine learning algorithm and flow/pressure sensor placement in water distribution networks (WDNs). A preliminary spectral clustering is performed for defining a suitable WDN partitioning, in order to carry out burst localization at cluster level and reduce the computational burden. A grouped regularization approach is used to define the most suitable flow and pressure sensor placement, inter and intra clusters, respectively, and train a neural network classification model, while explicitly considering sensor cost and data redundancy. Several burst (one at a time, at any WDN node) and water demand scenarios are simulated, for a time window of 30 days, for accounting spatial/temporal uncertainties of WDN hydraulic behaviour. The methodology is applied to a real-world WDN in Italy, pre-clustered in four detection areas, showing outstanding detection/localization test accuracy for small (80 % / on average 86.55 %), and especially moderate (100 % / on average 98.43 %) to large (100 % / on average 96.18 %) burst entities, even when relying on a very reduced number of sensors (just five flow meters located on the boundary pipes between detection areas).

In this paper the design of a nonlinear Model Predictive Control algorithm based on Recurrent Equilibrium Network models is addressed. Firstly, a tailored observer for the Recurrent Equilibrium Network model is proposed, in order to provide to the Model Predictive Control optimization an initialization that takes into account the past history of the system. Then, the Model Predictive Control optimization is designed including a proper terminal cost to guarantee closed loop stability for any choice of the prediction horizon. Copyright (C) 2024 The Authors.

This paper proposes a stabilizing Model Predictive Control algorithm, specifically designed to handle systems learned by Incrementally Input-to-State Stable Recurrent Neural Networks, in presence of input and incremental input constraints. Closed-loop stability is proven by relying on the Incremental Input-to-State Stability property of the model, and on a terminal equality constraint involving the control sequence only. The Incremental Input-to-State Stability is also used to derive a suitable formulation of the Model Predictive Control terminal cost. The proposed control algorithm can be readily applied to a wide range of Recurrent Neural Networks, including Gated Recurrent Units, Echo State Networks, and Neural Nonlinear Autoregressive eXogenous models. Furthermore, this work specializes the approach to handle the particular case of Long Short-Term Memory Networks, and showcases its effectiveness on a four tanks process benchmark.

In this paper a data-driven fault detection technique based on parity space is applied to the problem of detecting unannounced meals for type 1 diabetes patients. This method involves the generation and evaluation of a residual signal to detect faults (unannounced meals) acting on the system (patient). Insulin on Board (IOB), meal intake, glucose and its second derivative have been selected as input signals to generate the residuals, and the parity space matrices are estimated using 8-hour training in silico data generated by the UVA/Padova simulator. The residual evaluation module compares the residual with a threshold that is optimized for each patient using 2-day tuning data. The complete system is evaluated on 1-week scenario obtaining promising results (TPR=95%, PPV=72%) with a detection delay of 42 minutes for 80% of patients. For the 20 outlier patients, a fine-tuning and patient-tailored input signal processing are proposed as a future development.

This paper proposes a novel system identification schema to obtain a model of a ship manouvering process using artificial Neural Networks (ANNs), computational models that have changed the research paradigm, bringing remarkable advantages in several fields. The ANNs capabilities in modelling nonlinear dynamical systems are undeniable and several approaches have been proposed in recent years. In our work, an ordinary black-box approach used to determine the input-output relationship of the system under investigation is initially outlined. Then, a physics-oriented approach to train a recurrent neural network structure is provided and thoroughly explained, investigating also the possible adoption of a simpler network structure. The results obtained with the physics-oriented approach in modelling the ship manouvering process are significantly better than the ones achieved with the pure black-box approach. Consequently, the physics-oriented approach resulted to be an exceptional tool for inferring the physical laws behind a nonlinear system accounting for a limited amount of data. The effectiveness of this method motivates further studies to evaluate its possible implementation in model-based control algorithms.