Education and demonstration – Vehicle operator instruction or testing
Reexamination Certificate
1998-02-24
2002-07-30
Cheng, Joe H. (Department: 3713)
Education and demonstration
Vehicle operator instruction or testing
C434S030000, C434S037000, C703S004000, C703S006000
Reexamination Certificate
active
06425762
ABSTRACT:
The present invention relates generally to computer systems for simulating the operation of electronic circuits, software, mechanical systems, hydraulics, pneumatics, electromechanical systems, and combinations thereof, particularly to a system and method for integrating the operation of simulation objects representing various portions of a heterogenous system.
BACKGROUND OF THE INVENTION
A cosimulation is defined to mean the coordinated simulation, using N simulators, of N sub-models that together form a comprehensive model of a corresponding comprehensive system, in a way that the cosimulation behaves as closely as possible to one hypothetical simulation of the comprehensive model by one hypothetical simulator capable of simulating the comprehensive system.
The development of complex systems is almost always an interdisciplinary task involving several development teams that work on various components of the system in parallel. An example is the development of the electronically controlled transmission in a car: while one team might be concerned with increased fuel economy by improving the software implementing complex, potentially adaptive control algorithms, another team might investigate the ramifications of replacing an electronic sensor by a cheaper product, and a third might be working on the mechanical or hydraulic components of the transmission.
Any detailed simulation of the transmission's operation logic needs to include not only models of the mechanical and hydraulic components, but also electromechanical models of the sensors and actuators, and a model of the electronic control unit that implements the control logic. These different components will likely have been developed in different groups using different software tools, for example:
a tool for modeling and simulating abstract dynamic systems and designing embedded control software (such as MATRIXx, a product of Integrated Systems, Inc.),
a tool for modeling and simulating the mechanical aspects of the transmission,
a tool for modeling and simulating the electronic circuitry in the transmission, and
a tool for modeling and simulating the hydraulic components of the transmission.
Two fundamental approaches are possible when attempting to simulate such a heterogeneous system: either, one simulator is used that is able to simulate the entire comprehensive system, or a number of simulators is used in a coordinated manner so that the collective simulation results reflect the behavior of the transmission as a whole. The latter approach is called cosimulation.
Neither standard data formats that could represent a system as heterogeneous as the one in this example nor simulation tools that could simulate it exist today. Such simulation tools are not expected to exist in the foreseeable future. Thus a system simulation of a heterogeneous system requires a cosimulation of N simulators with different capabilities. Preferably the cosimulation will be between the simulators of all the tools that have been used when developing the components.
In order to allow this cosimulation of N simulators from different vendors, simulating models using different modeling paradigms (e.g., electronics, software, mechanics, etc.) and with various simulation methods (e.g., event-queue-based, differential equation based, etc.) without significant restrictions, a sophisticated cosimulation architecture is necessary.
Several cosimulation approaches have been proposed and implemented, partially as commercial products, in the past. See L. Maliniak, “Backplanes Mesh Simulators into One Environment,” Electronic Design (Aug. 8, 1994); and S. Schmerler et al., “A Backplane for Mixed-Mode Cosimulation,” in Proc. Eurosim 1995, Vienna, Austria. They range from ad-hoc integrations of two simulators, to complex simulation backplanes. Depending on the specific approach or product, various problems have effectively prevented their wide-spread adoption. In particular, it has been difficult to extend the prior art cosimulation techniques to cosimulations having multiple simulators using different simulation techniques (such as multiple differential equation solvers) and to incorporate extensions such as new data types.
The present invention avoids problems associated with prior cosimulation techniques. In particular it:
Uses a standardized protocol for exchanging signals between simulation objects, rather than a proprietary one. This makes it easier to ntegrate simulation products from different companies.
Can handle a virtually unlimited number of simulators operating together.
Does not require the use of a fixed set of synchronization techniques to synchronize the operations of the cosimulating simulators. In fact, multiple synchronization methods can be used concurrently in different parts of the cosimulation.
Can be used on one CPU, or across networks on different platforms.
Allows parallel execution of simulations.
Is not limited with respect to the simulation methods that participating simulators may use. Thus the cosimulation is not limited to discrete-event simulations only. Also there is no limit on the number of differential equation solvers used.
The numerical precision of the cosimulation degrades gracefully if some of the participating simulators do not support the full cosimulation protocol, in particular rollback capabilities. This allows virtually any simulator to participate in a cosimulation, albeit with less than ideal precision in some circumstances, in spite of not having rollback capabilities.
Cosimulation, according to the definition given above, should be part of a comprehensive development process whose goal is to create one comprehensive system, rather than different, non-integrated models for various components. The comprehensive model may be arbitrarily heterogeneous. In particular, its sub-models may use a wide variety of modeling languages and techniques, and may use a wide variety of simulation techniques.
Most previous cosimulation approaches have advocated a bus-line or star-like structure for cosimulation. An example of the star structure, also known as the backplane structure, is shown in FIG.
1
. In the backplane type of cosimulation system, the backplane controls the simulation time steps used by all simulators, and essentially tells all the simulators when to perform each incremental simulation step (i.e., when all the simulators are ready to perform that next step). In backplane based cosimulation systems, there are no parent/child relationships between simulators; all simulators are equal “siblings” that simply share signal values with each other. All the simulators run independently of each other, synchronized by signals generated by the backplane. There are several problems with the backplane cosimulation structure. For instance, like most centralized systems, it does not scale up well. When the number of simulators increases, the number of interactions between simulators increases, and the backplane becomes a bottleneck. The bottleneck problem can be solved by using a peer-to-peer setup, using a virtual object bus, as shown in FIG.
2
. However, this cosimulation structure is, in general, entirely unrelated to the structure of the comprehensive model. It also violates the abstract design principles of encapsulation and locality. Thus, like the backplane based cosimulation structure, the peer-to-peer structure also treats all simulators as equal, parallel partners in the simulation, all running independently except for periodic data exchanges.
Many traditional approaches to cosimulation, including the backplane and peer-to-peer structures mentioned above, assume that the comprehensive model has a flat structure, such as the one shown in FIG.
3
. In a flat structure, all simulators are equal partners in the cosimulation. No simulator can invoke another simulator in order to generate its results. For this reason, each component of the system must be modeled using only one tool, and the system being simulated must be partitioned in terms of simulation tools, rather than in terms of functions. Also, in violation
Cheng Joe H.
Harris Chanda
Pennie & Edmonds LLP
Wind River Systems, Inc.
LandOfFree
System and method for cosimulation of heterogeneous systems does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for cosimulation of heterogeneous systems, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for cosimulation of heterogeneous systems will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2870153