Rate-constrained wavelet multiresolution-model-based predictive control for hybrid combustion-gasification chemical looping process

Chemical looping process is a novel technology to separate oxygen from nitrogen using solid oxygen carrier to facilitate carbon dioxide capture in the design of next generation clean coal power plants. The chemical looping process involves multi-phase gas-solid flows that have a highly challenging uncertain nonlinear multi-scale dynamics with jumps, not captured well by traditional single-scale models and as a result not amenable to tight control through traditional approaches. As a consequence, there exists a need for developing model-based advanced control solutions to tightly and robustly operate and control the chemical looping process. In an effort to model and control such a complex system, single-input-single-output NARX models developed based on the temporal multi-resolution wavelet decompositions are presented. First, the NARX model, nonlinear in the wavelet basis, but linear in parameters, is obtained as a model identifier that well characterizes the nonlinear dynamics of single loop gas/solid flow behavior. Then, this model is approximated via Taylor expansion to obtain a linearized NARMA-L1 model. Based on the linear model, one-step ahead predictive control laws are designed. To reduce the excessive aggressiveness of the resulting control signals, ad hoc input constraints are designed and applied to limit the rate of control signal. Next, the one-step-head predictive control is extended to multistep adaptive generalized predictive control (GPC) scheme based on online identification of multi-resolution wavelet model structure. The control inputs and wavelet model parameters are calculated by optimizing the cost function using gradient descent methods. The convergence and stability of the proposed GPC scheme are shown using Lyapunov stability theorem. The real-time implementation results of the one-step-ahead predictive control law with rate constraint on the single loop configuration of Alstom's cold solid flow chemical loop test facility in Windsor, Connecticut, are presented. The tracking performance of the real-time controller is presented in this paper.