Pasar al contenido principal

NUEVO METODO DE APRENDIZAJE POR REFUERZO MULTIOBJETIVO BASADO EN MODELOS. APLICACION A CONTROL PREDICTIVO.

PID2020-120087GB-C21

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal de I+D+i Orientado a los Retos de la Sociedad
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2020
Plan Estatal 2017-2020 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSITAT POLITÈCNICA DE VALÈNCIA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 4
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Modified Active Disturbance Rejection Predictive Control: A fixed-order state-space formulation for SISO systems

  • Martínez-Carvajal, Blanca Viviana
  • Sanchís Saez, Javier|||0000-0001-9697-2696
  • Garcia-Nieto, Sergio|||0000-0002-2722-742X
  • Martínez Iranzo, Miguel Andrés|||0000-0002-1444-0651
[EN] This paper presents a novel control strategy that provides active disturbance rejection predictive control on constrained systems with no nominal identified model. The proposed loop relaxes the modelling requirement to a fixed discrete-time state¿space realisation of a first-order plus integrator plant despite the nature of the controlled process. A third-order discrete Extended State Observer (ESO) estimates the model mismatch and assumed plant states. Moreover, the constraints handling is tackled by incorporating the compensation term related to the total perturbation in the definition of the optimisation problem constraints. The proposal merges in a new way state¿space Model Predictive Control (MPC) and Active Disturbance Rejection Control (ADRC) into an architecture suitable for the servo-regulatory operation of linear and non-linear systems, as shown through validation examples., This work has been supported by MCIN/AEI/10.13039/501100011033 [Project PID2020-120087GB-C21] , MCIN/AEI/10.13039/501100011033 [Project PID2020-119468O-I00] , the Generalitat Valenciana regional government, Spain [Project CIAICO/2021/064] , and the Ministry of Science, Technology and Innovation of Colombia [scholarship programme 885] .
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Conditional scenario-based energy management algorithm with uncertain correlated forecasts

  • Gonzalez-Querubin, Edwin
  • Sanchís Saez, Javier|||0000-0001-9697-2696
  • Salcedo-Romero-de-Ávila, José-Vicente|||0000-0003-1577-5039
  • Martínez Iranzo, Miguel Andrés|||0000-0002-1444-0651
[EN] This paper introduces the application of the novel approach known as Conditional Scenario-Based Model Predictive Control (CSB-MPC) into energy management in a low voltage distribution system (LVDS). The LVDS comprises renewable energy sources (RES) and an energy storage system, with correlated power generation and demand under uncertainty. This correlation is leveraged to derive a reduced set of conditional scenarios and their associated probabilities. This reduced set retains the essential characteristics of a larger set of equiprobable scenarios generated based on information about uncertainties in power generation and demand forecasts. Instead of computationally demanding methods using the larger set, a feasible Scenario-Based Mixed-Integer Linear Program (MILP) is solved by employing the reduced set. This optimisation minimises costs associated with energy consumption from the main grid and the potential waste of generated power. In the objective function, the scenarios are weighted by their probabilities to mitigate the impact of less likely forecasts, leading to efficient LVDS component utilisation and reduced operating costs. Simulation results comparing the performance of the LVDS with CSB-MPC against other MPC approaches, including classic scenario-based, chance-constrained, and deterministic methods, reveal that CSB-MPC consistently exhibits higher probabilities of operational constraint satisfaction, with differences of up to 10% or even 30%. Notably, CSB-MPC achieves these results with similar solution times to classic scenario-based MPC but with lower LVDS operating costs, between 6.62% to 11.31% less., This work was supported in part by the GVA Regional Government, Spain through project CIAICO/2021/064, in part by the MCIN/ AEI/10.13039/501100011033, Spain under Grant PID2020-120087GB-C21, and in part by the Ministry of Science, Technology and Innovation of Colombia under scholarship programme 885.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Conditional scenario-based model predictive control

  • González, Edwin
  • Sanchís Saez, Javier|||0000-0001-9697-2696
  • Salcedo-Romero-de-Ávila, José-Vicente|||0000-0003-1577-5039
  • Martínez Iranzo, Miguel Andrés|||0000-0002-1444-0651
[EN] This paper proposes a novel MPC approach called conditional scenario-based model predictive control (CSB-MPC), developed for discrete-time linear systems affected by parametric uncertainties and/or additive disturbances, which are correlated and with bounded support. At each control period, a primary set of equiprobable scenarios is generated and subsequently approximated to a new reduced set of conditional scenarios in which each has its respective probabilities of occurrence. This new set is considered for solving an optimal control problem in whose cost function the predicted states and inputs are penalised according to the probabilities associated with the uncertainties on which they depend in order to give more importance to predictions that involve realisations with a higher probability of occurrence. The performances of this new approach and those of a standard scenario-based MPC are compared through two numerical examples, and the results show greater probabilities of not transgressing the state constraints by the former, even when considering a smaller number of scenarios than the scenario-based MPC., This work was supported in part by the MCIN/AEI/10.13039/501100011033 under Grant PID2020-120087GB-C21, and in part by the Ministry of Science, Technology and Innovation of Colombia under scholarship programme 885.
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

A model independent constrained predictive control for the Furuta Pendulum, Control predictivo con restricciones e independiente de modelo para el p´endulo de Furuta

  • Martínez, Blanca
  • Sanchis, Javier
  • García-Nieto, Sergio
[Resumen] Este trabajo presenta el control de seguimiento de un péndulo Furuta basado en el diseño de dos lazos paralelos con restricciones, rechazo activo de perturbaciones y predicción de salidas. Para cada subsistema, se asume que la planta gobernada tiene una dinámica de primer orden más integrador (FOPI). Así, se calcula una ley de control predictivo aplicando una estrategia de horizonte deslizante en dicha planta (FOPI). La diferencia entre la dinámica real y el modelo de predicción asumido se compensa mediante el mecanismo de rechazo de perturbaciones heredado del control activo de rechazo de perturbaciones (ADRC) e incorporado en los lazos. Para el diseño del control no se realiza ninguna identificación de modelos ni linealización matemática. Además, la estrategia permite incorporar las restricciones reales del sistema. Este trabajo valida con resultados prometedores la arquitectura denominada Modified Active Disturbance Rejection Predictive Control (MADRPC) mediante la estabilización de un sistema mecánico subactuado considerando las variables restringidas y en ausencia de un modelo nominal, en contraste con el enfoque estándar en el Control Predictivo de Modelos (MPC) en espacio de estados., [Abstract] This paper presents the tracking control of a Furuta pendulum based on the design of two parallel-constrained loops with active disturbance rejection and outputs predictions. For each subsystem, it is assumed that the governed plant resembles firstorder plus integrator (FOPI) dynamics. So, a predictive control law is computed by applying a receding horizon strategy in such FOPI plant. The mismatch between the actual dynamics and the assumed prediction model is compensated through the disturbance rejection mechanism inherited from the Active Disturbance Rejection Control (ADRC) incorporated in the loops. No modelling identification or mathematical linearisation is performed for the control design. Moreover, the strategy allows the incorporation of the actual system constraints. This work validates with promising results the architecture named Modified Active Disturbance Rejection Predictive Control (MADRPC) by stabilising a sub-actuated mechanical system considering the constrained variables and in the absence of a nominal model, in contrast to the standard approach in the state-space Model Predictive Control (MPC)., Colombia. Ministerio de Ciencia, Tecnología e Innovación; Scholarship 885, Generalitat Valenciana; CIAICO/2021/064