Circulating fluidized beds (CFB) have been widely applied to many areas of industry such as chemical processing, petroleum refining, catalytic cracker processing, power generation, and waste treatment.; Recently, a mathematical model of the CFCFB standpipe was successfully developed and tested using an extended Kalman filter (EKF) and an Hinfinity estimator algorithm. Using this standpipe mathematical model requires a solids circulation rate (SCR) to be a measurable variable offered from a spiral installed in the standpipe of the CFCFB. In this research, a linear state space system model is developed in order to estimate the solids circulation rate, using a least squares estimator and a subspace algorithm.; In this research, a sliding mode estimator is applied in order to estimate the states and the bed-height using a mathematical model with the estimated SCR from the pressure drop profiles. The sliding mode estimator uses the Lyapunov stability criteria to obtain a gain that drives the estimator dynamic to a defined sliding surface, which usually is the first or second order differential equation of the error dynamic defined as the difference between the mathematical model equation and the estimator dynamic.; Although an entire CFCFB dynamic system has not been built, extensive experimental data sets are available, so a neural network is a strong candidate in building the entire CFCFB system dynamic model. In this research, the neural network is used to obtain the entire system model of the CFCFB for simulation purposes. In the neural network, back-propagation algorithms are adapted and tansig functions are used for the neurons. (Abstract shortened by UMI.)
展开▼
