Adaptive PI-control for an air sterilization module

Abstract: The goal of this master thesis has been to implement adaptive PI-control of two air flow heaters in an air sterilization module. This is done by modeling the module and using the information from the model to estimate the dynamics of the heaters. The idea of this thesis is to implement adaptive control with as simple control structures and tuning algorithms as possible, to keep it as generalized and accessible as possible. This way it will be easier to apply the same framework to other components. The implementation uses a range of models to express the thermodynamic behaviour of two heaters interconnected by a heat exchanger and some piping with heat losses. Using the models, a recursive least squares algorithm estimates the parameters of each heater, and finally the heaters are tuned using the lambda-method and the AMIGO-method. The models are created both from physical models and from non-physical models based purely on measurement data using the best available order. The implementation was done using MATLAB Simulink from which the code was then ported over to BECKHOFF TwinCat and integrated into the machine code. Due to time constraints the implementation only got to the simulation stage in Twin- Cat, but it showed promise and worked as expected. Simulations were also done in Simulink using measurement data which also showed promise for a functional framework.