Application of Gaussian processes for black-box modelling of biosystems.

Different models can be used for nonlinear dynamic system identification and the Gaussian process model is a relatively new option with several interesting features: model predictions contain the measure of confidence, the model has a small number of training parameters and facilitated structure determination, and different possibilities of including prior knowledge exist. In this paper the framework for the identification of a dynamic system model based on Gaussian processes is shown, illustrated on a simulated bioreactor example and then applied to two case studies. The first one addresses modelling of the nitrification process in a wastewater treatment plant and the second models biomass growth in the Lagoon of Venice. Special emphasis is placed on model validation, an often underemphasised part of the identification procedure, where the Gaussian model prediction variance can be utilised.

Wastewater treatment benchmark: what can be achieved with simple control?

In this paper a simple control strategy is applied to and assessed on the wastewater treatment benchmark. The controllers used in the presented control strategy are PI controllers, feedforward control and a step-feed procedure. The controlled variables are not directly the effluent concentrations but other process variables which have an effect on the effluent. The setting of set-points is also analyzed to select the values with the best performance. Set-point analysis has shown that with an optimal setting of set-points under stormy influent conditions, the achieved plant performance is also retained for rainy and dry influent conditions. The evaluation of plant performance indicates that with the proposed control strategy, a lower number of effluent violations has been achieved, as well as lower energy consumption and lower sludge production, when compared to results published up to now. Only the effluent quality criterion deteriorated.

System analysis for optimal control of a wastewater treatment benchmark.

The paper presents an analysis and optimisation of a wastewater treatment benchmark. The benchmark is a simulation environment defining a plant layout, simulation model, influent data, test procedures and evaluating criteria that should be used for comparing different control strategies. In this paper an analysis of the benchmark which addresses the influences of potential manipulated variables on control performance under different operating conditions is presented. In the study optimisation is used to define the optimal values of the manipulated variables under constant as well as dynamic influent conditions. The results indicate that such an analysis and optimisation give important information about the manipulated variables under varying influent conditions and consequently about possible control strategies.

Experimental design of an optimal phase duration control strategy used in batch biological wastewater treatment.

The paper presents the design of an algorithm used in control of a sequencing batch reactor (SBR) for wastewater treatment. The algorithm is used for the on-line optimization of the batch phases duration which should be applied due to the variable input wastewater. Compared to an operation with fixed times of batch phases, this kind of a control strategy improves the treatment quality and reduces energy consumption. The designed control algorithm is based on following the course of some simple indirect process variables (i.e. redox potential, dissolved oxygen concentration and pH), and automatic recognition of the characteristic patterns in their time profile. The algorithm acts on filtered on-line signals and is based on heuristic rules. The control strategy was developed and tested on a laboratory pilot plant. To facilitate the experimentation, the pilot plant was superimposed by a computer-supported experimental environment that enabled: (i) easy access to all data (on-line signals, laboratory measurements, batch parameters) needed for the design of the algorithm, (ii) the immediate application of the algorithm designed off-line in the Matlab package also in real-time control. When testing on the pilot plant, the control strategy demonstrated good agreement between the proposed completion times and actual terminations of the desired biodegradation processes.