When an epidemic spreads into a population, it is often impractical or impossible to continuously monitor all subjects involved. As an alternative, we propose using algorithmic solutions that can infer the state of the whole population from a limited number of measures. We analyze the capability of deep neural networks to solve this challenging task. We base our proposed architecture on Graph Convolutional Neural Networks. As such, it can reason on the effect of the Abhishek Tomy and Matteo Razzanelli contributed equally to this work.
Summary
This paper deals with the application of model predictive control (MPC) to optimize power flows in a network of interconnected microgrids (MGs). More specifically, a distributed MPC (DMPC) approach is used to compute for each MG how much active power should be exchanged with other MGs and with the outer power grid. Due to the presence of coupled variables, the DMPC approach must be used in a suitable way to guarantee the feasibility of the consensus procedure among the MGs. For this purpose, we adopt a tailored dual decomposition method that allows us to reach a feasible solution while guaranteeing the privacy of single MGs (ie, without having to share private information like the amount of generated energy or locally consumed energy). Simulation results demonstrate the features of the proposed cooperative control strategy and the obtained benefits with respect to other classical centralized control methods.
Distributed Model Predictive Control refers to a class of predictive control architectures in which a number of local controllers manipulate a subset of inputs to regulate a subset of outputs composing the overall system. These controllers may cooperate to find an optimal control sequence that minimizes a global cost function, as in the case of Cooperative Distributed Model Predictive Control (CD-MPC). In this paper two linear CD-MPC algorithms for tracking are proposed. The aim of these controllers is to drive the outputs of the overall system to any admissible piece-wise constant set-point, satisfying input and state constraints. However, in the available literature this result is achieved by using a set of centralized variables that keep track of the global state of the system. In contrast, we develop novel CD-MPC approaches for tracking that rely on “as local as possible” information instead of the plant-wide information flow. These new control strategies reduce the required communication overhead, local computational demands, and are more scalable than CD-MPC algorithms available in the literature. We illustrate the main characteristics and benefits of the proposed approaches by means of a multiple evaporator process example
<p class="Abstract"><span lang="EN-US">The paper presents an overview of the recent and ongoing research activities of the Italian Interuniversity Center on Integrated Systems for the Marine Environment (ISME) in the field of geotechnical seismic surveying. Such activities, performed in the framework of the H2020 European project WiMUST, include the development of technologies and algorithms for Autonomous Surface Crafts and Autonomous Underwater Vehicles to perform geotechnical seismic surveying by means of a team of robots towing streamers equipped with acoustic sensors.</span></p>
We propose in this paper novel cooperative distributed MPC algorithms for tracking of piecewise constant setpoints in linear discrete-time systems. The available literature for cooperative tracking requires that each local controller uses the centralized state dynamics while optimizing over its local input sequence. Furthermore, each local controller must consider a centralized target model. The proposed algorithms instead use a suitably augmented local system, which in general has lower dimension compared to the centralized system. The same parsimonious parameterization is exploited to define a target model in which only a subset of the overall steady-state input is the decision variable. Consequently the optimization problems to be solved by each local controller are made simpler. We also present a distributed offset-free MPC algorithm for tracking in the presence of modeling errors and disturbances, and we illustrate the main features and advantages of the proposed methods by means of a multiple evaporator process case study
This article describes a hybrid simulation approach meant to facilitate the realization of a simulator for underwater vehicles with one or more manipulators capable of simulating the interaction of the vehicle with objects and structures of the environment. The hybrid simulation approach is first described and motivated analytically, then an analysis of simulation accuracy is proposed, where, in particular, the implications of added mass simulation are discussed. Then, a possible implementation of the proposed architecture is shown, where a robotic simulator of articulated bodies, capable of stable and accurate simulation of contact forces, although unfit to simulate any serious hydrodynamic model, is tightly interfaced with a general purpose dynamic systems simulator that is used to simulate the hydrodynamic forces, the vehicle guidance, navigation, and control system, and also a manmachine interface. Software details and the technicalities needed to interface the two simulators are also briefly presented. Finally, the results of the simulation of three operational scenarios are proposed as qualitative assessment of the simulator capabilities.
When an epidemic spreads into a population, it is often unpractical or impossible to have a continuous monitoring of all subjects involved. As an alternative, algorithmic solutions can be used to infer the state of the whole population from a limited amount of measures. We analyze the capability of deep neutral networks to solve this challenging task. Our proposed architecture is based on Graph Convolutional Neural Networks. As such it can reason on the effect of the underlying social network structure, which is recognized as the main component in the spreading of an epidemic. We test the proposed architecture with two scenarios modeled on the CoVid-19 pandemic: a generic homogeneous population, and a toy model of Boston metropolitan area.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.