Modeling, Identification and Control of Irrigation Station with Sprinkling: Takagi-Sugeno Approach - Complex System Modelling and Control Through Intelligent Soft Computations - page 462

Indeed, the T-S fuzzy model can approximate highly nonlinear system into several
locally linear subsystems interconnected. The identi cation problem in the T-S

fuzzy model can be summarized in two steps: structure identi

cation and parameter
estimation. Several techniques were developed to conclude the modeling of these
systems: we quote primarily the neuro-fuzzy technique (Daneshwar and Noh 2013 ;
Azar 2010a ) and clustering technique (Daneshwar and Noh 2013 ; Azar 2010a ) and
clustering technique (Troudi et al. 2011 ; Li et al. 2013 ; Jang et al. 2007 ; Pingli et al.
2006 ; Xu and Zhang 2009 ; Zahid et al. 2001 ; Chakchouk et al. 2014 ). Indeed
Several researchers have noticed that a nonlinear system can be approximated by
the sum of several linear sub-systems. Method of clustering proves to be an
interesting technique for identi cation and the modelisation of the nonlinear sys-

tems. Indeed, this technique consists in approximating the total nonlinear system by
a vague model of Takagi-Sugeno type. In this case, each cluster represents one
fuzzy rule of Takagi-Sugeno. The number of clusters is

xed by an expert
according to the type and the performances of application considered. By conse-
quent to each cluster one correspond homogeneous zone of operation such that is
de ned in the form of a linear local model. We are interested to model and identify

a nonlinear system by the fuzzy logic approach such as Takagi-Sugeno (T-S)
approach. The latter, uses modeling containing linguistic rules to obtain the model
of system outputs. Initially, we present the fuzzy logic approach design, we gives an
outline on the first two models. Then, we detail (T-S) model, uses the method of
fuzzy coalescence for the identi

cation of the nonlinear systems by the fuzzy C-
means (FCM) algorithm. We will in addition present tests of validation of (T-S)
model. Then, we will give the results of identi cation and modeling of the station of

irrigation by sprinkling.
The remainder of this chapter is described such as the following section. In the

first section we have describe the station of irrigation by sprinkling, in which we
de ow and other

components of our station, secondly, in this section we detail the operation mode
and the

ne the practical constraints existing on the outputs pressure and fl

flowcharts of the closed loop mode with any controller and how select the
operation mode. In the second section we have describe the Fuzzy coalescence
algorithms. Thirdly, we spend to detail the FCM algorithm step by step. Finally, we

fl

finished by application of FCM algorithm to the irrigation station by sprinkling
located in the laboratory shown in the Fig. 1 . After identi cation and modelisation

with FCM algorithm it is necessary to validate our simulation results (model
mathematic of our pumping station) with Root Mean Square Error test (RMSE) and
the Variance accounting for test (VAF) and many other validation tests we have test
the stability of our open loop model, after modelisation and identi

cation we
control our T-S obtained model by two types of controllers, PI controller of the
station of irrigation and Fuzzy logic regulator, and we finish our chapter with a

comparative study between these controllers.