An optimization-based control strategy for energy efficiency of discrete manufacturing systems

Abstract

In order to reduce the global energy consumption and avoid highest power peaks during operation of manufacturing systems, an optimization-based controller for selective switching on/off of peripheral devices in a test bench that emulates the energy consumption of a periodic system is proposed. First, energy consumption models for the test-bench devices are obtained based on data and subspace identification methods. Next, a control strategy is designed based on both optimization and receding horizon approach, considering the energy consumption models, operating constraints, and the real processes performed by peripheral devices. Thus, a control policy based on dynamical models of peripheral devices is proposed to reduce the energy consumption of the manufacturing systems without sacrificing the productivity. Afterward, the proposed strategy is validated in the test bench and comparing to a typical rule-based control scheme commonly used for these manufacturing systems. Based on the obtained results, reductions near 7% could be achieved allowing improvements in energy efficiency via minimization of the energy costs related to nominal power purchased.

Energy Consumption Dynamical Models for Smart Factories Based on Subspace Identification Methods

“…being u, u ∈ Z ≥1 the lower and upper bounds to activate the device, respectively, for a specific input. Following [4], [5] to managing peripheral devices towards improving the energy efficiency of the industrial machines without sacrificing either production or quality, this paper has focused on showing and selecting an appropriate algorithm to identify energy models.…”

“…In this way, the opportunity to develop a methodology that eases the construction of the models based on the input-output data can be designed and implemented. This methodology should be oriented to control, with the aim of being used in the improvement of energy efficiency through the prediction of energy consumption [4], [5].…”

“…Several approaches have been proposed to create datadriven energy consumption models for industrial machines, which focus mainly on cutting/milling/machining methods, parameters and tool geometries [4], [6], [7]. Using strategies such as the finite element method (FEM), which manage to characterize the energy footprint with respect to the machine state and the material to be cut/milled/machined [7].…”

“…Thus, the main contribution of this paper is the description and applicability of a subspace identification (Sub-ID) algorithm based on RQ decomposition, which will allow identifying energy consumption models using input-output data. Therefore, the obtained models follow, with a suitable accuracy, the dynamics evolution of a validated energy consumption profile, fact that allows to use such models for the design of different model-based control strategies such as optimal, predictive or robust controllers, e.g, the design and implementation of a real-time Model Predictive Control (MPC), it has been achieved using energy models computed through Sub-ID based on CVA weighing matrices [4], [5].…”

A comparison of modelling approaches for closed-loop decision making over discrete domains in manufacturing systems

“…Then, to test the performance of the proposed approaches, an optimisation-based controller proposed in [17] to min-imise the energy consumption of a manufacturing process line will be considered. In this case, the optimal activation instants of binary and discrete actuators should be determined considering the operating constraints on both the process line and the system actuators.…”