CAJ | 학술논문

针对瞬态工况下油膜参数难于准确确定，提出了基于混沌径向基神经网络的汽油机瞬态工况油膜参数预测模型。首先证明了汽油机油路系统时间序列具有非线性混沌特性，对试验测定的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。然后，利用混沌算法确定隐含层高斯函数径向基中心和输出层连接权值，使其达到全局最优，加快了RBF神经网络的收敛速度。最后，将预测结果与采用BP神经网络模型和最小二乘法辨识的结果进行比较，验证了混沌RBF神经网络模型具有较强的非线性预测能力，能有效地提高油膜动态参数的预测精度，进而得出不同工况下的油膜参数动态特征。
침대순태공황하유막삼수난우준학학정，제출료기우혼돈경향기신경망락적기유궤순태공황유막삼수예측모형。수선증명료기유궤유로계통시간서렬구유비선성혼돈특성，대시험측정적수거진행상공간중구，이용RBF신경망락대중구후적수거진행훈련화예측。연후，이용혼돈산법학정은함층고사함수경향기중심화수출층련접권치，사기체도전국최우，가쾌료RBF신경망락적수렴속도。최후，장예측결과여채용BP신경망락모형화최소이승법변식적결과진행비교，험증료혼돈RBF신경망락모형구유교강적비선성예측능력，능유효지제고유막동태삼수적예측정도，진이득출불동공황하적유막삼수동태특정。
Aiming at the difficulty in accurately determining fuel film parameters in transient conditions, a model for predicting the fuel-film parameters in gasoline engine is proposed based on chaos radial basis function ( RBF) neural network. Firstly, it is proved that the time series of gasoline engine fuel circuit system exhibit a non-linear chaotic characteristic, a phase space reconstruction is conducted on test data, and the data reconstructed are trained and predicted by RBF neural network. Then chaos algorithm is used to determine and optimize the Gaussian radial basis function center of hidden layer and the connection weights of output layer, accelerating the convergence rate of RBF neural network. Finally the predicted results are compared with those using BP neural network model and least square identification. It is shown that chaotic RBF neural network model has stronger nonlinear prediction capability and can effectively improve the prediction accuracy of dynamic fuel film parameters, and hence the dy-namic features of fuel film parameters in different conditions can be obtained.