Automatic Control of Bioprocesses - Measurement, Monitoring, Modelling, and Control of Bioprocesses - page 51

Fig. 4 Basic fuzzy logic scheme for two input values, two rules, and one output value

applied, which is equivalent to the maximum and minimum determination of the
membership values of a rule, respectively. For each rule the so-obtained value is
used as the upper limit value for the conclusion (''then'' part) of the rule. During
the defuzzification, the output fuzzy sets of all rules are aggregated to one output
fuzzy set as shown in Fig. 4 . The centroid of the output fuzzy set is calculated as
the value for the actuating variable of the controller.
A fuzzy controller using nine rules was implemented by Ruano et al. [ 47 ] for a
biological nitrification process in a pilot plant with wastewater from a full-scale
plant. Instead of using an expensive nitrogen sensor, they employed several pH,
oxidation-reduction potential (ORP), and DO sensors. Their fuzzy controller
comprises two independent controllers: the nitrification as well as the denitrifi-
cation process controller. The former works as a supervisory control of the aera-
tion control system, whereas the latter modifies the internal recycle flow rate from
the aerobic to the anoxic reactor. The authors demonstrated that using low-cost
sensors in combination with their fuzzy controller leads to a minimized energy
consumption of the process.
For the temperature control of a batch reactor, Causa et al. [ 48 ] compared
different versions of a hybrid fuzzy predictive controller. Two on/off input valves
and a discrete-position mixing valve were used as controlled variables. The
authors concluded that the hybrid fuzzy predictive control in combination with an
optimization algorithm based on a genetic algorithm gives similar performance to
that of a typical hybrid predictive control strategy but a significant saving with