LMI-based MPC of Temperature and Humidity in Greenhouses using different learning models
Greenhouses are used to create an indoor microclimate for crop development, which is controlled by heating and ventilation to provide the suitable environment for the crops.
One approach is to use a deep neural network (DNN) to learn the dynamical model of the greenhouse climate to analyze and design a controller. It has been shown that DNN dynamic models are fully equivalent to Piecewise Afffine (PWA) systems, which can be used to design a controller by formulating the problem as LMIs.
Another approach is to use a lazy learning approach called Locally Weighted Learning (LWL), which share similarities with the approach of using Takagi-Sugeno (T-S) fuzzy model. In this approach, the dynamic model is described in the form of a sets of linear systems with different weights.
- Formulate the problems in form of LMIs
- Show that the proposed method is feasible and can guarantee the stability of the system
- Do simulations to illustrate the method.
Topic Areas:
Control Systems Theory, MPC, Nonlinear Systems, LMI, Learning.
Helpful/Required Prerequisites:
Background knowledge: Optimal Control, Nonlinear Control and Linear Algebra.
Simulation: Matlab.
Language: English.
Project start:
Currently available.
Estimated time requirements:
Theory: 50%
Simulations: 50%
Contact:
Hoang Hai Nguyen