The Research Of Boiler Measurement And Control System Based On Generalized Predictive Control

Coal boiler is a kind of important dynamic device, and has significant state in thedevelopment of national economy. In recent years, as people's consciousness of energyand environment protection is building up, on-line measurement and optimization controlof boiler production process has become one of the hot research aspects in the area ofenergy, environment protection and control. This paper analyzes boiler energy-exchangeprocess's multi-input, multi-output characters, and the characters of big time lag and tightcoupling. On the basis of full research of basic theory of traditional generalizedpredictive control (GPC), a kind of new GPC algorithm is put forward and a boilercontrol system based on the new algorithm is exploited.Parameters' diversity and control's complexity of the boiler combustion process isanalyzed and the advantage of GPC mechanism in complex industrial process isresearched. According to the problem of comparatively slow calculation caused by heavyand complicated matrix inversion in GPC traditional algoritm and using the researchresults of predictive control direct algorithms for reference, a kind of new algorithmusing matrix iteration algorithm to solve optimal solution is put forward. For a systemwith the rank of M, the new algorithm can get the optimal solution through iteration foronly M times, which reduces calculation amount. The bottle-neck problems that matrixinversion or part matrix inversion in GPC traditional algoritms and direct algorithmsmakes the time of data processing too long and takes up too large EMS memory aresolved.The speed of the algorithm and real-time performance of the system are improved.The combustion process's character of time lag and measurement method of keyparameters are researched, and soft-sensor mathematic model based on GPC algorithm isput forward. Real-time on-line measurement of key parameters in combustion process isrealized and the problem of time domain matching for measurement and control in boilercontrol is solved.According to the problems of coupling of model parameters and noises, and timevarying of systematic time lag and equipment capabilities, a kind of on-line identificationmodel correction method of recursive extended least square time-varying forgettingfactor is given. This method can adjust forgetting factor with the systematic dynamiccharacters automatically, modify model error in time to improve predictive precision andguarantee control effect and solve the problem that parameters estimation becomes slowin complex industrial conditions.The problems of nonlinearity of energy-exchange process of boiler and constraintcondition of input, output and process parameters (such as temperatuer in hearth) areresearched, and a kind of fast convergent recursive new algorithm of nonlinearconstrained GPC is constructed. The recursive algorithm avoids traditional algorithms'disadvantages, for example, traditional algorithms need the optimization at every point,which takes a long time, can't fully use available information and too rely on theprecision of nonlinear predigested linear mode. The new algorithm increases the selectionof generalized predictive control time domain, and solves the problem that controlcharacters is too bad and stable control even can't be realized because of impropernonlinear programming,. This new algorithm can be applied in the complex controlproblems of arbitrary function constraint conditions.The mechanism of the new algorithm is deeply researched, and astringency proof ofthe new algorithm is given. Combining electromagnetism disturbance of boilercombustion process, time varying of systematic time lag and equipment capability andnonlinearity of energy exchange, a lot of simulation experiments are done and industrialappplication is realized. The results show that the new GPC is feasible for industrialboiler which is a kind of complicated industrial multi-input, multi-output and nonlinearprocess with tight coupling and big time lag. At last, the paper further discusses andassumes the application foreground of GPC in complex industrial process control, andshows its board development space and application foreground.