Dynamic Modeling And Advanced Control Of Parabolic Trough Based Solar Power System

Traditional fossil energy faces problems of depletion and environmental pollution.To address these issues,most countries in the world have considered renewable energy as their future energy development strategy.Solar energy is one of the most abundant renewable energy sources in the world,therefore,solar-based power generation has been widely used in industry.At present,solar-based power generation technology can be divided into two categories: photovoltaic power generation and solar thermal power generation.Different from photovoltaic power generation technology,solar thermal power generation technology utilizes photothermal conversion to generate hightemperature steam to drive steam turbines to generate electricity.Therefore,the solar thermal power plant has similar power output characteristics as the conventional fossil-fired power plant,and it can be more easily connecected to the grid.Another advantage of solar thermal power generation technology is its less cost and difficulty of storing solar energy,because it uses thermal energy storage rather than chemical energy storage.At present,parabolic trough concentrating solar thermal power(PTC-CSP)generation is one of the mainstream commercial solar thermal power generation technologies.Although related technology has received extensive attention,its control problem has still not been sufficiently studied,especially the most important execution-level power generation control,which is even not mentioned in existing research works.To address these problems,this paper first studies simplified dynamic modeling approach of parabolic trough solar thermal power plant.The simplified model is employed as the simulation platform to study the decision-level and execution-level control strategy.Then advanced control technics is applied to the solar thermal power plant to improve the control performance.The main content of the paper is organized as follows:(1)Existing models of PTC-CSP plants cannot be conveniently used for control system simulation,thus we develop a simplified model of PTC-CSP plants with two-tank direct energy storage and two-tank indirect energy storage for control system simulation.The proposed model considers the dynamics of main heat transfer process and power generation process,including the dynamics of the solar collection system,the energy storage system,the steam generation system and the power generation system.Meanwhile,less important factors that influence the modeling of the PTC-CSP plant are ignored,such as the dynamics of the pumps,pipes and valves.Therefore,the proposed model avoids complex modeling procedures and captures key dynamics of the power plant,which makes it a well-suited platform for control system simultion.Finally,the steady-state and dynamic characteristics of the proposed model are validated.(2)Assuming the electricity is saled in a free market,the optimal generation scheduling of the PTC-CSP plant considering the prediction uncertainty of solar radiation and electricity pric is studied based on the mixed integer programming(MINLP)approach and maximum principle approach.MINLP approach establishes the mathematical model of the scheduling problem and solves it directly.In the MINLP scheme,this paper studies a linear equivalent robust approach and an improved stochastic scheduling approach to deal with the uncertainty.The equivalent robust approach is a convervative approach,which can consider the influence of prediction uncertainty on the performance index and constraints.The improved stochastic scheduling approach can balance the profits and the conservatism.The MINLP approach can accurately describe the scheduling problem and theoretically has a better performance.However,solving the MINLP problem requires a long computation time and the solution is unreliable.Hence this paper studies the maximal principle based approach,which avoids solving MINLP problem and has a fast and reliable solution.To further consider the influence of uncertainty,a Monte Carlo approach is used to simulate the uncertainty in electricity price and solar intensity,and the scheduling results are weighted to reduce the influence of uncertainty.Finally,this paper compares the MINLP approach and the maximum principle approach and discuss how to select the best approach.(3)The execution-level control of the PTC-CSP plant is studied.To deal with the large inertia,the strong coupling effect,the difference in time delay of multiple input channels and direct feedthrough between inputs and outputs,we propose a model predictive control(MPC)strategy based on extended nominimal state space model(NMSS).The proposed method overcomes the deficiencies of conventional state space model based MPC and avoids the design of state observers.To reduce the influence of solar radiation disturbance,moving horizon estimation(MHE)is employed to estimate the solar radiation,and we derive the standard quadratic programming form of MHE and its analytic solution under unconstrained case.Based on MHE,a feedforward MPC(FF-MPC)is proposed as the disturbance rejection controller,which can separate the design of tracking controller and disturbance rejection controller.To further reduce the fluctuation of power output caused by disturbance,an energy-balance based disturbance rejection method is proposed.The proposed method can cooperate with the FF-MPC to reduce the influence of solar radiation.