Font Size: a A A

Neural-Fuzzy Predictive Control And Its Applications On Recovery Boiler

Posted on:2002-07-27Degree:DoctorType:Dissertation
Country:ChinaCandidate:J LiFull Text:PDF
GTID:1118360032955082Subject:Control Science and Engineering
Abstract/Summary:PDF Full Text Request
Fuzzy system, Neural Network and Generalized Predictive Control theory are proposed in recent years. This paper integrates Fuzzy System, Neural Network and Generalized Predictive Control theory and introduces a kind of Neural-Fuzzy Predictive Controller. The application of this controller on the recovery system of a pulp factory shows the Neural-Fuzzy Predictive Controller's effectiveness.The dissertation includes following main contents:1. Review the evolution of fuzzy system and neural network, their respective advantages, disadvantages and the possibility of integration of fuzzy system and neural network.2. Introduce three kinds of integration of fuzzy system and neural network, that is fuzzy-neural system, neural-fuzzy system and hybrid fuzzy neural system.3. Introduce a new kind of neural-fuzzy system. This neural-fuzzy system can model a MIMO nonlinear system, its conclusion part of its fuzzy rules is ARX model. Using ARX model increases the ability of modeling and brings the advantage of integrated with GPC algorithm. The training method of this neural-fuzzy system is BP algorithm or BP integrated with LS algorithm. Using BP integrated with LS algorithm can increase the training speed.4. Introduce a neural-fuzzy predictive controller, which integrates GPC algorithm with the neural-fuzzy system. The rules and structure of the neural-fuzzy predictive controller are also introduced.5. Prove that the neural-fuzzy system can model any nonlinear function at any precision. A simulation result of a nonlinear function shows the good modeling ability of neural-fuzzy system. And the simulation results of green liquor density system show the good performance of the neural-fuzzy predictive controller.6. Introduce a hybrid fuzzy predictive controller. This hybrid fuzzy predictive controller integrated the advantage of predictive control and fuzzy control. The simulation results show its good performance.7. Introduce the normal recovery boiler of pulp factory. The main function of the recovery system is to recycle the energy and chemical drugs of black liquor. Usually recovery boiler firing control strategies use static method, such as constant black liquor volume flow, constant black liquor solids input, constant heat input, constant steaming rate, constant air consumption.8. Research a recovery system of a pulp factory. A static model based on materials and energy balance is established. An optimizing control strategy of recovery boiler is introduced. There are three layers in the control strategy, that is basic layer, advanced layer and monitoring layer. The basic layer is composed of single loops. The advanced layer is composed of three elements control, solids spray rate control, sootblowing control, black liquor droplet size optimization, firing control, where firing control uses the neural-fuzzy predictive control. The monitoring layer guarantees the safety and stabilization of firing, including blackout monitoring and excessive air control.9. Introduce the "ninth five year plan" national key project-"firing process modeling control and optimization commercial software". The project uses Suny TDCS9200 distributed control system. The use report of the factory shows the software brings 2.05% increase of steam yield per ton alkali, 0.79% increase of alkali recovery rate, 12% increase of operation periods. Finally, the author points out some problems still to be solved after summarizingthe work done in the dissertation.
Keywords/Search Tags:Neural-Fuzzy
PDF Full Text Request
Related items