【正文】 畢業(yè)論文不能順利完成，感謝他對(duì)我的悉心關(guān)照。感謝樊聰同學(xué)，他為我提供了大量論文專著。也要感謝謝感謝南昌凱馬柴油機(jī)有限公司提供了試驗(yàn)設(shè)備，凱馬公司優(yōu)越的試驗(yàn)條件是論文寫作完成的重要前提。它的研發(fā)實(shí)力和員工的愛崗敬業(yè)給我很大的啟示。感謝南昌凱馬柴油機(jī)有限公司鐘小琛工程師和鄒揚(yáng)寧工程師指導(dǎo)我完成試驗(yàn)，特別感謝鐘小琛工程師給予的詳細(xì)講解和幫助。感謝華東交通大學(xué)和凱馬南昌柴油機(jī)廠給我的學(xué)習(xí)機(jī)會(huì)。附錄A 外文翻譯原文部分Turbocharger Modeling for Automotive Control ApplicationsPaul MoraalFord Forschungszentrum AachenIlya KolmanovskyFord Research LaboratoriesABSTRACTDynamic simulation models of turbocharged Diesel and gasoline engines are increasingly being used for design and initial testing of engine control strategies. The turbocharger submodel is a critical part of the overall model,but its controloriented modeling has received limited attention thus far. Turbocharger performance maps are typically supplied in table form, however, for inclusion into engine simulation models this form is not well suited. Standard table interpolation routines are not continuously differentiable, extrapolation is unreliable and the table representation is not pact. This paper presents an overview of curve fitting methods for pressor and turbine characteristics overing these problems. We include some background on pressor and turbine modeling, limitations to experimental mapping of turbochargers, as well as the implications of the pressor model choice on the overall engine model stiffness and simulation times. The emphasis in this paper is on pressor flow rate modeling, since this is both a very challenging problem as well as a crucial part of the overall engine model. For the pressor, four different methods, including neural networks, are presented and tested on three different pressors in terms of curve fitting accuracy, model plexity, genericity and extrapolation capabilities. Curve fitting methods for turbine characteristics are presented for both a wastegated and a variable geometry turbine.INTRODUCTION AND BACKGROUNDDynamic simulation models of turbocharged Diesel and gasoline engines are increasingly being used for design and initial testing of engine control strategies. The turbocharger submodel is a critical part of the overall model, but its controloriented modeling has received limited attention thus far. A standard approach still appears to be to include the turbocharger performance data in the form of lookup tables directly into the model. However, this form is not ideally suited for use in controloriented engine models because the standard linear interpolation routine is not continuously differentiable, sometimes leading to apparent discontinuities in simulations. Furthermore, and more seriously, this type of model does not adequately handle operating conditions outside of the mapped data range, for example at very low turbocharger rotational speeds. While engine mapping usually covers the entire operatingrange, the situation for the turbocharger unit is different. Generally, it is possible to obtain the performance characteristics from the supplier. However, the turbocharger characteristics are typically only mapped for higher turbo speeds (typically 90000 RPM and up) and pressure ratios, whereas the operating range on the engine extends down to 10000 RPM shaft speed and pressure ratios close to unity. This is illustrated in Figure 1. As a result, standard interpolation methods (polynomial regression, lookup tables) generally fail to produce reasonable results outside of the region where experimental data is available. In fact, because of the nonlinear nature of the pressor and turbine characteristics even interpolation through lookup tables has been found to cause unacceptable performance in simulations. In this paper we will present a number of different curve fitting methods for turbochargers which allow a more pact implementation of the turbocharger submodel. Special emphasis will be on the abilities of the different methods to extrapolate the performance characteristics into regions which are usually not mapped, but which are encountered during normal operation of the engine. First we39。ll provide some insight into the limitations of experimental turbocharger mapping.LIMITATIONS OF EXPERIMENTALTURBOCHARGER MAPPINGWhenever the choice is given, one should obviously aim for extending the range of available experimental data, rather than trying to predict/extrapolate the behavior outside of the given range (unless good models of the underlying physics are available, in which case a few experimentally measured data points may suffice to characterize a large operating region and experimental development time can thus be reduced). With regard to extending the range of experimental mapping of turbocharger units on a flow bench (in particular, lower turbine speeds and lower mass flow rates), two problems arise, which would require significant effort to work around. The following is an explanation provided by David Flaxington , at the time working at Allied Signal Garrett.FLOW SENSOR ACCURACYThe flow through the pressor (the same arguments hold for the turbine) is usually measured by determining the differential pressure across a properly sized orifice placed in the flow path, and using Bernoulli39。s law, assuming inpressible flow (constant density), to determine the flow rate . This gives a smooth variation of pressor characteristics as the flow rate is varied. One can map at lower speeds if smaller orifice plates and nozzles are used to measure the lower flows, but this produces a problem. As the flow rate is reduced, the accuracy of the readings with any one measuring device reduces. Changing at some point to another measuring device sized for lower flow rates then causes a step change in the mass flow readings as the accuracy of the measurement is again improved. With a vast set of orifice plates and nozzles this problem could be circumvented. However, this is not standard practice and improvements are largely dependant uponthe goodwill of the supplier. HEAT TRANSFER EFFECTSA second problem associated with the low speed region of the maps is caused by the heat conduction f