Engine Modelling for Control Applications: A Critical Survey

In earlier work published by the author and co-authors, a dynamic enginemodel called a Mean Value Engine Model (MVEM) was developed. This model isphysically based and is intended mainly for control applications. In itsnewer form, it is easy to fit to many different engines and requires littleengine data for this purpose. It is especially well suited to embedded modelapplications in engine controllers, such as nonlinear observer basedair/fuel ratio and advanced idle speed control. After a brief review of thismodel, it will be compared with other similar models which can be found inthe literature. The attempt will be made to point out the differencesbetween the new modified MVEM and those developed elsewhere. In particular,the questions of fitting simplicity and general applicability are to betreated.     

Development and Validation of Hierarchical Models for the Design of Engine Control Strategies

The development of mathematical models for the design of controlstrategies for spark ignition automotive engines is described. The objectiveof the models, used for both simulation and optimization analysis, is theprediction of the effects of control strategies on fuel consumption andemissions of a vehicle over arbitrary driving cycles. In order to achievethe best compromise between precision, experimental costs, computationaltime and flexibility, a mixed modelling approach is used, withphenomenological and input-output models integrated within a hierarchicalsystem.Mean value models have been used to describe the most significant dynamiceffects: (i) air flow. (ii) two phases fuel flow in the intake manifold, and(iii) thermal flow in the cylinder walls. Stochastic effects due to sensorsand actuators can be also predicted.Two-zone and multizone thermodynamic models for the prediction ofpressure cycle sub-models for engine emissions (HC, CO, andNO _x and mechanical losses have been developed. Experimentaldesign techniques are also under development to optimize the interactionsbetween experimental analysis and models. Most of the models have beenintegrated in a computer code, used by a major automotive supplier.     

Di Diesel Engine with Variable Geometry Turbocharger (VGT): A Model Based Boost Pressure Control Strategy

The use of a variable geometry turbocharger (VGT) is a very efficienttechnology to improve performance of passenger car direct injection (DI)Diesel engines. In order to take full advantage of the potential of VGT,without penalty on fuel consumption and driveability, electronic control ofboost pressure is desirable. Standard control techniques (i.e.Proportional-Integral) do not assure a reliable VGT behaviour. This paperdescribes a new boost pressure control strategy developed to improve PItechnique, and based on a simplified modelling of VGT operation. Severaltests were carried out on engines to point out which physical quantitiesdirectly influence VGT dynamics, and to find out VGT operation modes (OMs).Simplified models for each OM were developed and model parameters wereidentified. Experimental tests provided a basis to design a control strategyable to meet general performance requirements. Since two different OMs wererecovered, two separate governors were developed, each active under certainconditions. A switching logic was created to manage transitions between thetwo governors. Finally, the new control strategy was validated on the enginewith good results.     

Engine Knock Detection from Vibration Signals Using Pattern Recognition

The paper deals with a diagnostic method that allows to detect engineknock. The developed algorithm differentiates three kinds of engine cycles:absence of knock, increasing knock and heavy knock. The decision is takenfrom a block vibration signal. The diagnostic method is based on patternrecognition. Three models of different data shapes provided from theaccelerometer are elaborated. This is done using a time-scale analysis toolcalled a wavelet network. It allows to extract relevant features from thesignal. The aim of the method is then to partition the feature space intoclasses representing the knock states. Experimental results are reported.     

基于RBF网络的有源噪声控制

提出一种神经自适应噪声有源控制（ANC）的方法。应用RBF（Radial Basis Function)网络对噪声进行有源控制。针对RBF的网络特点,使用递阶遗传算法确定网络参数（连接权、隐节点中心和宽度），同时解决了网络拓扑结构的优化训练。利用RBF网络的有源噪声控制系统用于三维空间传播的宽频带空调噪声的降噪获得了良好的效果。