A Study On L₁ Adaptive Control And Its Application In Thermal Processes

Under the background that lager amount of renewable power sources are connected to the utility grid,low load operation,frequent load varying and deep peaking has been common for the traditional thermal power plants,bringing new demands to the controller design for the thermal processes.The thermal processes in power plants are substantially with high uncertainty.These uncertainty becomes especially critical under varying working conditions in forms of nonlinearity,time-delay,varying parameters,etc.,which causes unsatisfactory performance to the conventionally designed controller.As an advanced robust adaptive control approach,L₁ Adaptive Control is with advantages of improved performance and robustness over the systems with uncertainty.However,L₁ Adaptive Control is still inapplicable to thermal process either in theory or method for the unique properties with thermal process.This dissertation devoted to the extension of L₁ adaptive control algorithm and its applications to power plants thermal process,and mainly focuses on:1.The L₁ Adaptive Control algorithm is extended to multivariable systems with output feedback.The extension is discussed to the systems with diffeomorphic output mapping and normal strict real systems,respectively.For the multivariable systems with diffeomorphic output mapping,the output feedback problem is transferred into an equivalent state feedback case by the use of differential geometry.For normal strictly real multivariable systems,an output predictor and corresponding adaptive law are designed to make estimation of the uncertainty in the system while a compensation part is contained in the control law.Then the closed-loop dynamic of the system is designed with pole-placement.The simulation experiments with Bell-?str?m boiler-turbine unit and 300MW combustion control system respectively proves the effect and superiority of the proposed algorithm.2.Aiming to the uncertain parameters mixed with time-delay associated to the thermal processes,L₁ Adaptive Control for time-delay system is proposed on the basis of internal model principle.The delayed time is integrated to the state predictor of L₁ Adaptive Control,in order to counteract the adverse effect on the system caused by time-delay.The approaches for low order filters designing are discussed severally for one order plus time-delay systems and high order plus time-delay systems.An analytical approach to determine the bandwidth of the low pass filter is put forward for one order plus time-delay systems while a numerical approach for searching the lowest feasible bandwidth of the low pass filter is given for high order plus time-delay systems.3.For incorporating system constraints into L₁ Adaptive Control,a control design with integration of L₁ Adaptive Control and receding horizon optimization is put forward.On the basis of the fact that the uncertainty compensation part and set-point tracking part are decoupled,the set-point related feed-forward term and output feedback term are replaced by a real-time updated optimal control term,which is based on receding horizon optimization.The performance and robustness can be adjusted by choosing proper weights matrices.The effect of the approaches is verified by the simulation on a super-heated steam attemperater with large rage of parameter perturbation.4.The L₁ Adaptive Control is further integrated with economic model predictive control,thus a control approach is proposed for economic load tracking of super-critical boiler-turbine units.The L₁ Adaptive Controller is employed to compensate the uncertainty associated with the system which makes the plant approximate a given linear reference model in dynamic.Then the reference model is employed in the economic predictive controller,which takes the system economy of some horizon in the future as its optimized index.The algorithm is economy-oriented but the load tracking performance and the economy of the boiler-turbine units are taken into account at the same time.