基于混沌RBF神经网络的汽油机进气流量预测研究

针对汽油机进气流量的多维非线性特性,提出了一种混沌径向基(RBF)神经网络的汽油机进气流量预测模型。证明了汽油机进气流量时间序列具有混沌特性,对采集的原始数据进行相空间重构,利用RBF神经网络对重构后的数据进行训练和预测,并利用混沌算法确定输出层连接权值和隐含层高斯函数径向基中心,使其达到全局最优,加快了RBF神经网络的收敛速度。仿真结果表明,与空气进气流量平均值模型、RBF神经网络预测模型比较,该模型具有更高的预测精度,为精确及时测试汽油机进气流量提供了一种全新的软件测量方法。

Gasoline engine intake flow forcast study of chaotic RBF neural network

A soft predictive model based on Chaos-RBF neural network is proposed for the intake air flow of gasoline engine as its multidimensional nonlinear characteristics. The engine air intake flow time series with chaotic characteristics have been proved;the phase space of the original data has also been reconstructed before using RBF neural network to train and predict. And then, the result has been compared with the air inlet flow average model and RBF neural network forecasting model. Chaos algorithm is used to determine and optimal the implied Gaussian radial basis function center and the out put layer connection weights, in order to accelerate the convergence rate of RBF neural network. The simulation results show that this model is a new method to measure the intake air flow of the engine with more accuracy and timeless, which is superior to the intake air flow average model and RBF neural network prediction model.