Modeling And Optimization Of Pulverized Coal Boiler Combustion System Based On RVM Cluster

As the main form of modern coal-fired boiler in power market, pulverized coalfurnace has a wide range of implementation, it not only consumes coal resources togenerate electricity and power for our work and life, but also brings NOx pollutionemissions and other pollutants to the environment. The pollution is huge and energy crisisare becoming increasingly serious, so the study of pulverized coal furnace has importantmeaning to combustion optimization. Pulverized coal furnace combustion system is astrong nonlinear system with high coupling, it has extremely complex characteristics ofcombustion process. To optimize the system, firstly we must establish the boiler model inwhich pulverized coal furnace parameters as input and the system model of NOxemissions and boiler combustion as output, then we should optimize the input values ofthe system.In this paper, i will use cluster (Boosting) algorithm and the relevance vector machine(Relevance Vector Machine, RVM)which is a hybrid modeling method introduced topulverized coal boiler combustion optimization, we select multiple relevance vectormachine as weak learning cluster algorithm on some subset of the training data points tolearn to get a different model, and combined with the PSOGSA (Particle SwarmGravitational Search Algorithm, PSOGSA) which we use as model parameter optimizealgorithm to optimize the model, according to the prediction error we calculate themodel’s corresponding weights, then we sum all the weak learning machines’ weights toestablish a strong system as the final prediction model, which indicates the optimizationprocess in pulverized coal boiler,we choose low NOx emission and high combustionefficiency as the goal, aiming at finding adjustable operation parameters of boiler by usingPSOGSA optimization algorithm.