Study On Fuzzy Adaptive Neural Network In Updraft Drying Section Of The Pellets Grate Temperature Field Control

The iron and steel industry mainly uses pellets with high compressive strength and uniform mass distribution.Chain grate is an important thermal equipment for pellet drying,dehydration and preheating oxidation.The effect of pellet drying and preheating is mainly affected by the internal temperature field of the chain grate.Chain grate temperature field has the characteristics of large time delay,nonlinearity,multi-field coupling,and complex influencing factors,which makes it difficult to achieve real-time and precise control.This article took the temperature field of the updraft drying(UDD)section of the grate as the research object,and explored the balanced and stable control method of the temperature field of the updraft drying section.The main content and research results of this article include:1.Based on heat transfer related theories and neural network system identification theory,a gray-box model between the input and output of the chain grate updraft drying section system was established.Through heat transfer theory,computational fluid dynamics and related theories,the spatial distribution mechanism model of pellet bed temperature and gas temperature in the updraft drying section of the chain grate was established.Comprehensively considering the multi-factors in the modeling of the temperature field of the updraft drying section of the chain grate,a method for identifying the “gray-box” system based on BP neural network was proposed.A system identification model that provided input and output references.2.A fuzzy adaptive neural network control strategy whose structure and parameters can be adjusted online was studied.According to the input and output conditions of the established gray-box system identification model,a multi-input multioutput fuzzy adaptive neural network controller for the temperature field of the updraft drying section of the grate was designed.The structure of fuzzy adaptive neural network with variable parameter Gaussian membership function and fuzzy rule number can be adjusted online.A learning algorithm for fuzzy adaptive neural network to modify the membership function parameters and the number of fuzzy rules was given.3.The numerical simulation method was used to validate the input and output gray-box identification model of the temperature field of the updraft drying section of the chain grate and the tracking effect of fuzzy adaptive neural network control.Compiled the “gray-box” identification model of MATLAB,and considered its input conditions as the inlet air temperature,inlet gas flow velocity,sampling time,and output temperature value under the mechanism model of the chain grate updraft drying section.The output was the expected temperature value of multiple different temperature measurement points.The numerical simulation results showed that the accuracy of the model identification with the gray-box system can reach more than 95%.With reference to the input and output conditions of the gray-box model,a fuzzy adaptive neural network control program for the updraft drying section of the grate was compiled,and the centers of the gaussian membership function of the fuzzy adaptive neural network were-1 and 1,respectively,and the width was 0.8493,the number of fuzzy rules was 4.The numerical simulation results showed that the temperature control tracking curve around the lower space of the grate bed coincided with the theoretical temperature curve around 20 s,and the control steady-state error was less than 3%.The temperature control tracking curve of the temperature field above the pellet bed coincided with the theoretical curve around 300 s,and the control error gradually decreased and remained within 8%.4.The fuzzy adaptive neural network control method was tested and verified by using multi-physics coupling process test device.A genetic algorithm was used to optimize the placement and number of temperature sensors,and 21 temperature field measurement sensors and their optimal locations were determined.A software program for fuzzy adaptive neural network control was compiled using MATLAB and LabVIEW.The results showed that fuzzy adaptive neural network control temperature field test point was stable around 270 s,the error was within 10%,and the local maximum temperature error after stabilization was is 3? The relative error of the expected output of the position was 5.45%,which satisfied the control accuracy requirements.The research results of this paper provided a theoretical reference for the temperature field control of the updraft drying section of the chain grate,and gave a reference value for improving the drying quality and the energy efficiency of pellets.