Fuzzy-SVM Prediction For Slagging Characteristics Of Coal-Fired Boilers

Coal-fired boiler slagging can reduce boiler efficiency and threaten the safety of power plant.Therefore,it is necessary and meaningful to establish an accurate and effective prediction model of boiler slagging by studying the combustion characteristics of coal-fired boiler and understanding the characteristics of boiler slagging so as to ensure the efficient and safe operation of power plant boilers.This paper introduces the research status and evaluation index of slagging in coal-fired boilers.Through the analysis of coal type and boiler structure,seven indexes of softening temperature t,silicon ratio G,silicon-aluminium ratio SiO₂/Al₂O₃,ratio of base to acid B/A,comprehensive index R,dimensionless furnace tangential diameter d_si/D and dimensionless furnace maximum temperature t _W were selected as input of the model.Five membership functions are used to fuzzify the input indices,including triangular membership function,trapezoidal membership function,semi-circular membership function,Cauchy membership function and oblique membership function.The fuzzified results are used as input of support vector machine（SVM）to establish the membership functions respectively.Fuzzy-Support Vector Machine（SVM）model for predicting boiler slagging.In order to optimize the parameters of each membership function,an improved quantum particle swarm optimization（QPSO）algorithm is proposed,which integrates multi-neighborhood local search strategy into the traditional QPSO algorithm,enhances the search accuracy of the original algorithm,and establishes a fuzzy support vector machine boiler slagging prediction model based on improved QPSO to optimize the membership function.In this paper,40 groups of boiler samples are used,of which 30are training samples and 10 are testing samples.The experimental results verify the validity of the method of combining membership function with support vector machine.The experimental comparison also shows that the improved quantum particle swarm optimization algorithm optimizes the parameters of the membership function.Fuzzy Support Vector Machine（FSVM）model has better prediction accuracy and can be used as slagging prediction method for coal-fired boilers in power plants.