Modeling, Estimation and Control of Indoor Climate in Livestock Buildings

The main objective of this research is to design an efficient control system for the indoor climate of a large-scale partition-less livestock building, in order to maintain a healthy, comfortable and economically energy consuming indoor environment for the agricultural animals and farmers.

In this thesis, a conceptual multi-zone climate model is proposed according to the knowledge about the hybrid ventilation theory. The method is to compartmentalize the building into some well-mixed macroscopic homogeneous zones, with the major emphasizes on the occupied spaces where the animals confined in. With necessary assumptions and simplifications, the dominant air flow distributions are investigated and the phenomenon of horizontal variations is well depicted.

The designed entire control system consists of an outer feedback closed-loop dynamic controller and an inner feed-forward redundancy optimization. The dynamic control is implemented through Model Predictive Control (MPC) algorithm based on linearized time invariant state space representations. The application of moving horizon approach for estimation and regulation, accompanied with target calculation and disturbance modeling, demonstrate the significant advantages of MPC on performance improvement, tracking reference, rejecting disturbance, compensating model/plant mismatch, and naturally dealing with constraints. An auto-covariance least-square method is applied to increase the estimation quality through recovering the unknown covariance entering the system.

The redundancy optimization is applied through exploiting the nonlinearities of the actuators to passively attenuate the high frequency disturbance (wind gust), to accommodate the limitation of the bandwidth of the closed-loop system as well as pursuit of an optimum energy solution. By assigning different weights in the objective function which is based on energy consumption considerations and the covariance of the high frequency component disturbances, the optimal control command generated from the dynamic controller are reallocated to the end effectors. This strategy enhances the resilience of the control system to disturbances beyond its bandwidth, increases the manipulators utilization efficiency, and reduces energy consumption by solving a constrained convex optimization.

Through comparative simulation results analysis, the proposed modeling and control technique is proved to be able to better determine the optimal ventilation operation, increase the actuator's utilization, guarantee the pleasant indoor thermal comfort and the reasonable indoor air quality, lower cost (energy) and improve the quality (productivity). This technique is expected to be feasible in the similar real scale livestock buildings, and could be considered as an alternative solution to the current used decentralized PID controller.