最小资源分配网络及其在电站锅炉中的应用

MINIMAL RESOURCE ALLOCATION NETWORKS AND APPLICATION FOR A POWER STATION BOILER

燃煤电站锅炉内NOx 的生成规律非常复杂,与锅炉的燃煤、送风方式、燃烧器等许多运行参数和结构有关。人工神经网络具有联想、记忆、自适应、自学习、适于处理非线性问题等优点。该文采用基于RBF网络的最小资源分配网络(MRAN)对某电站锅炉NOx 的生成规律和效率进行建模,该模型不仅能自动调节隐节点数、学习速度快、学习精度高、适于在线运行,而且具有能同时预测NOx排放和锅炉效率等优点。该模型对电站锅炉的运行具有指导意义和参考价值。

The generation mechanism of NOx in boilers of a pulverized coal power station is very complex. It concerns with many operating parameters and structures such as coal, air and burners. Artificial neural networks (ANN) possess many advantages, such as association, memory, self-adption, self-learning, and the fitness to deal with non-linearity problems. A model for NOx emissions and efficiency of a pulverized coal power station boiler, established by minimal resource allocating networks (MRAN) on radial basis function (RBF) networks, possesses the excellence of self-tuning, hidden nodes, fast learning, high accuracy, fitness of operating on line, whats more, it can also predict the NOx emission and boiler efficiency at the same time. The model has the significant reference value for a pulverized coal power station boiler.