Soft-sensor Method Of Circulating Ash Utilization In Cfb-fgd Process Based On Rbf Neural Network

Circulating Fluidized Bed for Flue Gas Desulfurization (CFB-FGD) is quite a novel technology that shows significant advantages with a low Ca/S ratio in comparison with other semi-dry technologies and that could also be competitive with the widely-used wet FGD technic. The advantages of CFB over other technologies include:product is easy to deal with desulfurization; smaller space needed and possibility of easily retrofitting existing plants; flexibility in operation with regard to varying boiler load; low energy consumption; reduced installation and operating costs.Firstly, the mechanism and technics of CFB-FGD Processes were introduced. According to analyse the mechanism and technics of CFB-FGD, shows that the repeated recycle use of circulation ash improves the efficiency of CFB-FGD greatly. At present, cycling materials on the specific role of the research process is not very clear, therefore the measurement of Circulating Ash is conducive to further study CFB flue gas desulphurization technology, and achieves CFB-FGD system optimization control.At present there is very little literature about the effect of circulating ash in the desulfurization. For further reaserching circulating fluidized bed desulfurization process of recycling by the ash, a soft sensor circulating ash utilization measurement was studied in this paper.The selection of auxiliary variables plays an important role in establishing soft-sensing model. Therefore, influencing factors about circulation ash utilization was analyzed in detail in the paper. The auxiliary variables selected included:the quantity of water; the quantity of fresh desulfurizer; the quantity of circulating ash and the gas concentration, temperature and flowrate in the entry of the reactor.Particle swarm optimization was studied, and modified particle swarm optimization (MPSO) was porposed by introducing inertia weights and constriction factor. This method combined MPSO and gradient-descent, took full advantage of the global searching performance of MPSO and the local optimized effectiveness of RBF neuralnetworks; overcomed general PSO algorithm convergent instability and the disadvantage of RBF networks with falling into local minimum, a hybrid optimization algorithm for radial basis function (RBF) neural networks based on modified particle swarm optimization (MPSO) was proposed. Soft sensor was realized by using the after-trained RBFNN in circulating ash utilization model. Simulation results showed that, based on the RBF neural network soft sensor model has high accuracy, good performance and good prospects.