Search Results

Today’s automobiles are among the most sophisticated machines on the planet. Much of the functionality of modern automobiles emanates from embedded software features that control electronic, mechanical or pneumatic devices. Over the past few decades the number of software features and the associated code has grown exponentially and the respective embedded software systems have reached a level of complexity which is increasingly difficult to manage. As a consequence, recalls due to software defects have become a major concern and today constitute about 50% of the overall warranty cost [1]. Since the operation of automobiles has severe public safety implications, the development of embedded automotive software has become subject to stringent functional safety standards (ISO 26262) and compliance with these standards has become a major hurdle in the development of automotive software.

This paper highlights and discusses the development and deployment of an enterprise-level tool infrastructure that fits into the production build environment of Ford Motor Company. A particular focus is on navigating bottlenecks and pitfalls that arise with the adoption of a model-based software development process. This includes provisions to support centralized data and architecture artifact management (including version control across the lifecycle of the software), support to integrate and manage legacy software artifacts, support to archive and bookshelf development milestones, and last but not least, built-in intelligence to spot potential sources of software defects early in the development stage.

The focus of this paper is an air charge estimator for engine control system applications which do not feature a mass air flow (MAF) sensor. The proposed approach, beyond its independency of a MAF sensor, is designed to be compatible with the confines of a typical production control system configuration. The air charge estimation algorithm is based on mean-value models for the manifold pressure dynamics and the gas flows through the throttle and valve orifices. It involves nominal static models for the volumetric efficiency of the engine and for the throttle discharge coefficient. The static models for those parameters are complemented with correction factors that are adjusted on-line. The update of the volumetric efficiency correction is implemented in the form of a Kalman-filter which uses the difference between the measured and the modeled manifold pressure as an error metric.

The present paper describes the development of a comprehensive engine control strategy for a 5.7 Liter eight cylinder engine with dual exhaust system. The control strategy is torque-based and distinguishes two essential control modes, engine speed control under idle engine operation and torque (load) control under normal engine operation. The controller synthesis is carried out within the context of a thoroughly model-based design paradigm. It involves the synthesis of one distinct multi-variable feedback controller for each of the two control modes. A particular focus is on synthesizing robust feedback-controllers that require very little experimental calibration. To this end, multi-variable H∞ design principles are applied. The linear models underlying the controller synthesis are derived from an accurate nonlinear engine model. The fully integrated controller is experimentally validated on a test vehicle equipped with a custom-made, full-authority controller prototyping system.

This paper outlines the derivation of an analytical volumetric efficiency model that can be used in mean-value engine models or in air estimation algorithms for variable cam phasing and two step valve lift applications. The model is the product of a physics-based modeling approach. It accounts for the most prominent effects that occur during the gas exchange phase of a four cycle combustion process. Variable valve lift and valve timing are intrinsically modeled in terms of their geometric nature. The gas pressure trajectory, which has a crucial impact on the volumetric efficiency, is modeled via piece-wise linear approximation functions. The proposed model has a total of 16 regression parameters that need to be adjusted on the basis of experimental data. The model validation is based on four sets of engine mapping data, each set pertaining to one particular valve lift mode but otherwise spanning the entire engine operating envelope in terms of speed, load and cam-phasing positions.

This paper presents a comprehensive Matlab/Simulink-based framework that affords a rapid, systematic, and efficient engine control system development process including automated code generation. The proposed framework hinges on three essential pillars: 1 ) an accurate model for the target engine, 2) a toolset for systematic control design, and 3) a modular system architecture that enhances feature reusability and rapid algorithm deployment. The proposed framework promotes systematic model-based algorithm development and validation in virtual reality. Within this context, the framework affords integration and evaluation of the entire control system at an early development stage, seamless transitions across inherently incompatible product development stages, and rapid code generation for production target hardware.

The fierce competition and shifting consumer demands require automotive companies to be more efficient in all aspects of vehicle development and specifically in the area of embedded engine control system development. In order to reduce development cost, shorten time-to-market, and meet more stringent emission regulations without sacrificing quality, the increasingly complex control algorithms must be transportable and reusable. Within an efficient development process it is necessary that the algorithms can be seamlessly moved throughout different development stages and that they can be easily reused for different applications. In this paper, we propose a flexible engine control architecture that greatly boosts development efficiency.

There is a clear sense of urgency among the automotive industry to cut the cost and time spent on all aspects of vehicle development, and move as much as possible towards virtual design practices. In the area of embedded engine control systems development this move requires that the industry gradually accept a paradigm shift towards a rigorously model-based design practice. This paper elaborates on current shortcomings in the development process and highlights key prerequisites that enable virtual design for engine control applications.