Reinforced methods for dynamical system identification and adaptive control

In designing adaptive control systems, one ought to determine the type of uncertainties appearing in the plant to be controlled. Some important uncertainties can occur at the output of the control system, inducing negative effects that might sometimes lead to unstable control. One of the main ones is the uncertainty arising in the output disturbance description of a plant model. Nonnormal noise, and, particularly, the presence of outliers due to occasional failures of a signal acquisition device degrade the performance and control of the stochastic dynamical system. None the less important is the uncertainty that arises due to nonlinearities of measurement devices and/or control actuators. Nonlinearities, such as friction, deadzone, saturation, preload, backlash, and hysteresis, are called (hard-nonlinearities), and are common in most control systems, especially in electro-mechanical ones. Usually, such nonlinearities can occur at the output of the system, inducing negative effects. They significantly limit the performance of control systems.The influence of both uncertainties in respect of the proximity of stochastic control system output to the reference or set-up signal is analysed here theoretically and using numerical simulation by PC. The problem of joint time-varying parameters and the time delay tracking by processing observations, is solved here as well. The monograph is a result of more than twenty-year experience of the author in research and teaching of digital signal processing (DSP) and system identification for adaptive control. It is aimed at three major groups of readers: senior undergraduate students, graduate students, and scientific research workers in electrical engineering, computer engineering, computer science, and digital control, too.