U. Kortela

Neuro-fuzzy modelling of power plant flue-gas emissions

This paper concerns process modelling using fuzzy neural networks. In distributed logic processors (DLP) the rule base is parameterised. The DLP derivatives required by gradient-based training methods are given, and the recursive prediction error method is used to adjust the model parameters. The power of the approach is illustrated with a modelling example where NOx-emission data from a full-scale fluidised-bed combustion district heating plant are used. The method presented in this paper is general, and can be applied to other complex processes as well.

Dynamic model for a bubbling fluidized bed coal combustor

Abstract The bed fuel inventory is one of the key variables in the fluidized bed combustion process when describing the dynamic behaviour of combustors burning fuels like coal or graphite. Many of the essential features of fluidized bed combustor responses can be simulated with the dynamic model presented. The model consists of differential equations describing the dynamic behaviour of the furnace and steady-state calculations for the burning of char particles. The simulations show good agreement when compared to process measurements from a commercial-sized coal combustor.

Learning Control of Thermal Systems

Thermal processes are common in industry. The systems are often quite complex because of the phenomena involved and the sizes of the plants. This Chapter shows how learning techniques can be used for such complex systems.

DIAGNOSIS SYSTEM FOR CONTINUOUS COOKING PROCESS

Abstract Industrial systems generate a lot of information for operators. Increased measurement information flow might cause difficulties for the process operators to observe the process states and faulty process operation. In this study, measurements and statistical variables are combined by fuzzy logic to generate key factors for several points in the continuous cooking digester. The overall diagnosis system combines the key factors into one system which is used for the operational purposes and as a helping tool for process condition monitoring.

Cascade generalized predictive controller: two in one

The paper presents a cascade generalized predictive controller. The cascade control task is performed by one predictive controller and the cascade feature is incorporated in a special predictor. Simulation results are presented comparing the performances of the proposed control algorithm to traditional cascade loops including two PI or two GPC controllers. The paper investigates the effects of noise filter on the robustness of the control loops in the cascade control structure. It shows, that with the proposed predictor it is possible to adjust independently the robustness of the inner and outer loops, meanwhile in the traditional cascade loop there are cross effects in this sense. Finally a real time application of the proposed algorithm is presented: the cascade GPC was tested in the oxygen control loop of an experimental fluidized bed boiler.

FUZZY CONTROL OF COMBUSTION WITH GENETIC LEARNING AUTOMATA

Abstract It is difficult to achieve effective control of time variable and nonlinear plants such a fluidized bed boiler. A method of designing a nonlinear fuzzy controller is presented. However, its early application relied on trial and error in selecting either the fuzzy membership functions or the fuzzy rules. This made it heavily dependent on expert knowledge, which may not always available. Hence, an adaptive fuzzy logic controller such as Adaptive Neuro-Fuzzy Inference System (ANFIS) removes this stringent requirement. This paper demonstrates the application of ANFIS a nonlinear Multi Input Single Output fuel feeding and combustion system and a fuzzy controller design for the system with optimization with Genetic Learning Automata (GLA). An ANFIS model has been developed to determine the exact amount of fuel fed to a combustion chamber. This property is impossible to measure directly, but it is required for improving combustion control. The control of the combustion base on two Takagi-Sugeno type controllers, which were optimized by GLA. The control system has been validated on experiment data obtained in a case-study power plant. The results have shown that the system is able to capture the nonlinear feature of the fuel feeding system.

Integrated Optimisation and Control System to Reduce Flue Gas Emissions

Abstract The control of the fluidised bed boilers can be studied as a layered system, which contains two main layers. The long-term operation of the process is optimised in the upper layer and the stabilising control loops are in the lower layer. These two main layers can be connected in a direct or indirect way. At the optimisation level, the flue gas emissions have to be minimised and the thermal efficiency of a boiler has to be maximised. The optirnisation is impossible, if stabilising controllers do not work properly. Combustion power control (CPC) is used to stabilise burning.

Adaptive Process Control Using Controlled Finite Markov Chains Based on Multiple Models

Abstract Controlled finite Markov chain (CFMC) approach can deal with a large variety of signals and systems with multivariable, non-linear and stochastic nature. In this paper, adaptive control based on multiple models is considered. For a set of candidate plant models, CFMC models (and controllers) are constructed off-line. The state transitions predicted by the CFMC models are compared with frequentist information obtained from on-line data. The best model (and controller) is chosen based on the Kullback–Leibler distance. This approach to adaptive control emphasizes the use of physical models as the basis of reliable plant identification. Three series of simulations are conducted: to examine the performance of the developed Matlab-tools; to illustrate the approach in the control of a non-linear non-minimum phase van der Vusse CSTR plant; and to examine the suggested model selection method for the adaptive control.

PREDICTION AND CONTROL OF BLOW-LINE KAPPA NUMBER

Abstract Two industrial continuous cooking processes producing both softwood and hardwood pulp were studied. Real time kappa number profiles of conventional and Downflow Lo-Solids cooking were modelled. Gustafsons kappa number model with optimised parameters was used. Blow-line kappa numbers were successfully predicted before cooking zone. New kappa number control strategy is presented. In control strategy temperature set point is determined iteratively using prediction model of kappa number.

Model-Based Multivariable Control of a Secondary Air System Using Controlled Finite Markov Chains

Abstract Controlled finite Markov chains are used for solving a multivariable control problem. The controlled system was the secondary air flow system in a fluidized-bed combustion power plant. A four-input four-output system control problem was formulated as a CFMC, and solved using dynamic programming. A non-linear system model, developed in earlier studies, was used in this model-based design. Three filters were applied to smoothen the (finite) control actions, for feasible plant control. These were based on averaging over finite state-space, references, and averaging over time. Simulations were conducted to illustrate the approach, and compare it with respect to SISO PID design. The simulations indicated that system control can be succesfully designed based on CFMC techniques, even for a 4×4 MIMO system.