•4 min read
In classical feedback control, like PI or PID (Proportional-Integral-Derivative) control, the controlled variables (CV) are directly measured, compared with a setpoint, and subsequently fed back to the process via a feedback control law.
In Fig. 8.10, the signals, without the time argument, are denoted by a small letter, where y is the controlled variable (CV) which is compared with the reference (setpoint) signal r. The tracking error ε (i.e. r - y) is fed into the controller, either in hardware or software, from which the control input u, also known as the manipulated variable (MV), is generated. The input u directly affects the process (P) from which an output (y) results. The sampled output is subsequently compared with r, which closes the loop. In practice, this loop continues until the controller is switch off. There exists extensive literature on feedback control (Doyle et al. 1992; Morris 2001; Ogata 2010), and this has been a subject of research for many years, starting with the works of Bode (1930) and Nyquist (1932).
Fig. 8.10 Feedback control with controller (C) and process (P). r reference signal, eps tracking error, u input signal, y output sigma
In RAS, typical CVs are temperature, pH, and dissolved oxygen (DO) concentration, for which reliable sensors exist. Consequently, feedback control of these water quality parameters can be easily realized. However, in practice, most often, the input and output signals are disturbed by noise processes, such as unknown random inputs and measurement noise. Moreover, the overall process (P) may change over time as a result of growth, maturation, senescence, etc. Fish feed is another input into the RAS and its effect on fish growth cannot be directly seen or measured. For these parameters, model-based controllers (e.g. feedforward, model predictive, and optimal control) are typically introduced to predict the response of a change in the control input. However, fish feed is commonly added on the basis of values found in tables or recipes, but this rule-based control may need some adjustment in real practice to act as a feedback controller. Fish behaviours in RAS are a classical feedback control measure as fish react physiologically to environmental changes with variations in movement, location, receptiveness to feed, etc.
Hydroponic production usually takes place in protected environments such as greenhouses or plant factories where both the root and aerial environment need to be controlled. On-off controllers that predictively model optimal aerial environments have been proven superior in experimental research, but commercialization has been slow, whereas feedback controllers are standard in most climate-controlled greenhouses. However, the actuator varies with the type of controller with heating valves and vents typically feedback-controlled but lighting usually having an ON-OFF mechanism and only a few being dimmable. Controllers that rely on sensor or data input can respond to fast growth in a protected environment and result in highquality produce with high market prices that improves its cost benefits. Many commercial greenhouses still have the classical centrally located sensor hanging 1—2 m above the crop and covering several hundred square meters is still in use, but multiple wireless sensors covering smaller areas are being introduced although much of the detailed data cannot be used because rather large climatic zones are controlled by the same actuators. Advances in sensor technology (e.g. microclimate temperature sensors, image processors, real-time gas-exchange or chlorophyll fluorescence measurements) connected to modelling software could use decision-support systems and become automated control systems.
In typical bioreactor systems, temperature, pH, dissolved oxygen in aerated systems, and gas fluxes in anaerobic systems are continuously measured and adjusted with available temperature, pH, and dissolved oxygen controllers. In addition to this, both hydraulic (HRT) and sludge retention (SRT) times are also frequently set by controlling (waste)water flows and biomass waste flows, respectively.