基于GA优化RBF网络的焦炭质量模型

The coke quality model based on GA optimization RBF network

文章提出了一种基于遗传算法(GA)优化径向基函数(RBF)神经网络的焦炭质量预测模型。RBF网络存在两个关键问题:一是如何确定隐含层中心,而是如何调整网络权值。本文通过减聚类算法确定RBF网络基函数的中心数目,应用遗传算法对RBF网络权值进行优化。主要对焦炭的抗碎强度、耐磨强度、反应性指数和反应后强度使用GA优化RBF神经网络预测。结果表明该模型有较强适应性,同时能保证较高的预测精度,具有一定的实用价值。

An optimization of coke quality prediction model of radial basis function（RBF） neural networks based on genetic algorithms（GA） ispresented. RBF networks exists two key questions： First, how to determine the center of the hidden layer, and how to adjust the network weights. The number of center-based RBF network function is determined by reducing clustering algorithm, and genetic algorithm optimizes the RBF network weights. Coke crushing strength, the strength of GA by optimize RBF neural network prediction after abrasion resistance, reactivity index and response are studied. Simulation results show that the proposed model has strong adaptability and ensurs a high prediction accuracy, which has some practical value.