Genetic Algorithm Improvement And Modified Model Research Of Cement Grate Cooler Heat Transfer Parameters

The MI&HC algorithm is time-consuming,SHC algorithm produces lots of redundant edges,and MMHC algorithm has low efficiency of global optimizing capacity,aiming at these problems,we construct two improved batch Bayesian learning algorithms.These algorithms are IHC algorithm and IGHC algorithm.Genetic algorithm has low efficiency of global optimizing capacity,aiming at this problem,we propose one improved algorithm,IMGA algorithm.Then,proposing one incremental learning algorithm,IBN algorithm based on IMGA algorithm.Utilizing the improved batch learning algorithm and incremental learning algorithm construct the static model and incremental model of heat exchange parameters in the cement grate cooler.And conduct the fault diagnosis of the secondary temperature.The specific research works are as follows:Firstly,we proposed IHC(Improved Hill Climbing)algorithm.IHC algorithm utilizes mutual information and conditional independence test to construct initial structure,deals with isolated nodes,and searches for the best Bayesian structure based on improved hill climbing operators.Then we proposed IGHC(Improved Genetic Hill Climbing)Algorithm.IGHC algorithm adjusts the crossover probability and mutation probability automatically according to the difference value of score,adjusts the crossover manners and mutation manners automatically according to the difference value of score.Secondary,we proposed IMGA(Immune Genetic Algorithm).IMGA introduces improved immune strategy,crossover and mutation automatic strategy in population,introduces improved marriage strategy and teacher-student communication mechanism between two populations,the searches and constructs for the Best Bayesian structure.In order to deal with the new incremental data,we construct one incremental Bayesian algorithm IBN(Incremental Bayesian Network)based on IMGA algorithm.IBN algorithm constructs one function to monitor the incremental data,utilizes match degree function to calculate the match degree of current structure and new data,theutilizes automatic operators to modify the current structure.Lastly,we analysis of cement clinker heat transfer principle,then chose seven deep related heat transfer parameters,utilize the constructed Batch learning algorithm and Incremental leaning algorithm to construct the static model and incremental modified model to amend the current structure and utilize MLE(Maximum Likelihood Estimation)algorithm conduct parametric learning to the static data regression model and incremental modified data regression model.Utilize VE(Variable Elimination)algorithm conduct fault diagnosis and reasoning of secondary air temperature of heat transfer.