CAJ | 학술논문

针对由氧传感器构成的瞬态空燃比反馈控制系统无法满足实时性要求的问题，提出了基于混沌时序最小二乘支持向量机（LS‐SVM ）的瞬态空燃比预测模型。对试验采集到的一维空燃比数据利用相空间重构技术构造多维空间数据，恢复空燃比时间序列的多维非线性特性，然后采用LS‐SVM对重构后的数据进行训练及预测，得出预测结果。仿真结果表明：与Elman神经网络预测模型及前馈BP神经网络预测模型相比较，混沌时序LS‐SVM预测模型具有更强的非线性预测能力，能够有效地提高瞬态空燃比的预测精度。
침대유양전감기구성적순태공연비반궤공제계통무법만족실시성요구적문제，제출료기우혼돈시서최소이승지지향량궤（LS‐SVM ）적순태공연비예측모형。대시험채집도적일유공연비수거이용상공간중구기술구조다유공간수거，회복공연비시간서렬적다유비선성특성，연후채용LS‐SVM대중구후적수거진행훈련급예측，득출예측결과。방진결과표명：여Elman신경망락예측모형급전궤BP신경망락예측모형상비교，혼돈시서LS‐SVM예측모형구유경강적비선성예측능력，능구유효지제고순태공연비적예측정도。
For the problem that the feedback control system of transient air‐fuel ratio with oxygen sensor could not realize the real‐time demand ,the prediction model of chaos least square support vector machine was put forward .The multi‐dimensional space data were constructed with the collected test data ,the multi‐dimensional non‐linear characteristics of air‐fuel ratio time series were restored ,the reconstructed data were trained with LS‐SVM and the prediction results were acquired .The results show that the chaos LS‐SVM prediction model has the non‐linear prediction ability and can improve the prediction accuracy of air‐fuel ratio effectively compared with the Elman and BP network model .