Control of chaotic systems using targeting by extended control regions method (Q5940267)

The authors use the extended control regions (ECR) algorithm [see \textit{E. Ott, C. Grebogi} and \textit{J. A. Yorke}, Phys. Rev. Lett. 64, 1196--1199 (1990; Zbl 0964.37501)] for targeting, which is improved by using individual neural networks for each activation region. As examples, the authors study:NEWLINENEWLINE-- the logistic map represented by the equation NEWLINE\[NEWLINE x_{n+1} = px_n(1-x_n), NEWLINE\]NEWLINE where \(p\) is the control parameter;NEWLINENEWLINE-- the Hénon map, which is represented by the relations NEWLINE\[NEWLINE x_{n+1} = p+0.3y_n -x_n^2\quad\text{and}\quad y_{n+1}=x_n; NEWLINE\]NEWLINENEWLINENEWLINE-- the Lorenz system governed by the equations NEWLINE\[NEWLINE\dot x = \sigma(y-x),\qquad \dot y = (\rho x-y-xz),\qquad \dot z = (xy-\beta z), NEWLINE\]NEWLINE where \(\sigma\), \(\rho\) and \(\beta\) are the system parameters;NEWLINENEWLINE-- the Duffing oscillator governed by the equations NEWLINE\[NEWLINE\dot x = y,\qquad \dot y = -x-x^3-dy+f\cos(wz),\qquad \dot z = 1, NEWLINE\]NEWLINE where \(d\), \(f\) and \(w\) are the system parameters.