Neuro-Fuzzy Control for Semi-active Dampers - Fuzzy Modeling and Control: Theory and Applications - page 178

Fig. 9.8 Simulink model with the proposed neuro-fuzzy control
The first phase is focused on obtaining the initial values for the system parameters.
Moreover, an optimization process of the nodes at the hidden layer is carried out,
achieving a reduction of the number of Fuzzy rules. In the last phase, a re-adjusting
of the different parameters of the Neuro-Fuzzy System is performed. In fact, an
algorithm based on the Kohonen network has been applied in the first phase and an
algorithm based on the least squared method has been applied in the second one.
Similarly to the previous section, a reference table should be built with the most
adequate voltages for each combination of amplitude and frequency. Once the ref-
erence table is built and the training phase is concluded, the already trained Neuro-
Fuzzy system is combined by simulation with the TMD model as Fig. 9.8 shows.
In this way, when an external disturbance is introduced in the system, an
accelerometer takes measures over the footbridge obtaining amplitudes directly and
the dominant frequencies are obtained calculating FFT. These values are introduced
into the Neuro-Fuzzy system as inputs, then the device is in charged of providing the
most adequate voltage values for each case. The insertion of this intelligent control
methodology, allows providing to the structure a better damping since its value is
updated each 0.5 s. It is useful to note at this point that this kind of isolation system
is very efficient in order to reduce the usual external perturbations existing on foot-
bridges. Furthermore, it slows down degradation issues in structural integrity as the
vibration disturbances are now far away from generating resonance problems.
9.4 Conclusions
In this chapter, two applications of the semi-active dampers along with neuro-fuzzy
techniques have been presented. In fact, two types of vibration reduction strategies
have been shown. First one based on incorporating an isolator device and the other
one based on Tuned Mass Dampers (TMD). Both strategies have been illustrated
with an application. First application is focused on reducing the vibration generated