Active response control of an offshore structure under wave loads using a modified probabilistic neural network

Abstract

Offshore structures are subjected to wave, wind, and earthquake loads. The failure of offshore structures can cause sea pollution
as well as loss of property and life. Therefore, the safety of such structures is an important issue. The reduction of the
dynamic response of an offshore structure subjected to wind-generated random ocean waves is another critical problem with
regard to the serviceability, fatigue life, and safety of the structure. In this article, the responses of offshore structures
under random ocean waves are controlled using a modified probabilistic neural network (MPNN). As a more advanced method, it
uses the global probability density function (PDF) produced by summing the heterogeneous local PDFs automatically determined
from the individual standard deviation of each variable. The state vectors in a state-space model of a structure and the resulting
control forces made by a linear quadratic regulator algorithm were used to generate the training patterns for the MPNN and
a conventional multilayer perceptron (MLP). The results were compared with those produced by back-propagation based on the
MLP. The proposed MPNN method shows good results not only in controlling the responses but also in terms of the computation
time.