WO2001020429A2

WO2001020429A2 - Computer aided engineering system with adjustable empirical models

Info

Publication number: WO2001020429A2
Application number: PCT/US2000/025519
Authority: WO
Inventors: Andrew J. Barber
Original assignee: Mts Systems Corporation
Priority date: 1999-09-17
Filing date: 2000-09-18
Publication date: 2001-03-22
Also published as: WO2001020429A3; AU7588400A

Abstract

A computer aided engineering system (50) includes memory (34, 38) for storing empirical models (54) of physical components defined by inputs and outputs. The system (50) further includes an output device (42) and a processor (32) operably connected to the memory (34, 38) and the output device (42). The processor (32) executes a module (52) having instructions for modeling a physical system from the empirical models (54) of the physical components. The module (52) is adapted to adjust at least one of the inputs, outputs or time scale of one of the models (54) forming the physical system.

Description

COMPUTER AIDED ENGINEERING SYSTEM WITH ADJUSTABLE EMPIRICAL MODELS

BACKGROUND OF THE INVENTION

The present invention relates to computer aided engineering, design, modeling or analysis systems used to model physical systems, and more particularly, to systems that use empirical models of physical components.

Computer-aided engineering and design (CAE/CAD) software and hardware tools are used to define (i.e., "model") and verify (i.e., "simulate") prototype system designs. Many CAE/CAD systems have models that are stored in memory and are built using theoretical (non-empirical) techniques. Typically, this type of modeling includes obtaining detailed physical information about each of the components of the physical system that will be modeled. Physical information can include the geometry of each of the components, the material properties, etc. This information can be obtained from physical measurements or specifications of existing devices, or from conceptual ideas of new prototypes having selected characteristics. In addition to the physical information about the components, this technique of modeling also requires an understanding of the basic physical principles (e.g. Newton's law of motion, Kirchhoff's laws, etc.) that control operation of each of the components as well as the physical system. The foregoing information is combined to yield a set of differential equations and algebraic equations that model each of the components in the physical system. The equations are solved by manual calculation or by a computer to complete the modeling.

An overriding goal of any CAE/CAD system is to improve performance, operation or design of the physical system or components thereof by modeling and analysis. Using non-empirical or theoretical models as discussed above (also known as a "whitebox" modeling) , the operator can vary one or more of the physical attributes or characteristics of the components and ascertain what effect this has on the complete physical system. Unfortunately, as explained above, a detailed understanding of each of the physical components as well as the physical system is required in order that the desired change in performance of the physical system is realized.

Another approach, also known as "blackbox modeling", uses a different modeling technique.

Specifically, each component is tested or otherwise ascertained in order to define a relationship between inputs and outputs of the component . A model

("blackbox") is then defined based on the relationship of the inputs and outputs. In short, the blackbox model is a collection of algebraic equations that define a relationship ("curve fit") between the input and the output data. Commonly, the model is specified by describing a family of algebraic functions and a set of corresponding coefficients that describe the placement (location, amplitude, scale) of these functions. The coefficients are typically obtained by a statistical procedure. In many instances, the coefficients have no simple connection to the physical parameters of the components or the associated system. However, a significant advantage of this form of modeling is the relative ease in which a physical component can be modeled. In particular, detailed information of the physical information and controlling principles are not required.

Neither of the above-described modeling techniques is clearly preferred. Thus, there is an ongoing need to improve CAE/CAD modeling. In particular, there is a need to improve CAE/CAD modeling that includes the benefits of each modeling technique .

SUMMARY OF THE INVENTION A computer aided engineering system includes memory for storing empirical models of physical components defined by inputs and outputs. The system further includes an output device and a processor operably connected to the memory and the output device. The processor executes a module having instructions for modeling a physical system from the empirical models of the physical components . The module is adapted to adjust at least one of the inputs, outputs or time scale of one of the models forming the physical system.

A second aspect of the present invention is a computer implemented method of engineering analysis of a physical system formed from empirical models of physical components defined by inputs and outputs. The method includes receiving values indicative of adjusting at least one of the inputs, outputs or time scale of one of the models; adjusting said at least one of the inputs, outputs or time scale of one of the models as a function of the received values; and outputting results of analysis of the physical system having the adjusted said at least one of the inputs, outputs or time scale of one of the models. BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram of an exemplary computing environment for implementing the present invention .

FIG. 2 is a simplified block diagram of a CAE/CAD system.

FIG. 3 is a simplified block diagram of an empirical, blackbox model, including adjusters represented of one aspect of the present invention.

FIG. 4 is an input-output plot of a hypothetical shock absorber, under realistic loading conditions .

FIG. 5 is a first modified plot of the characteristics illustrated in FIG. 4. FIG. 6 is a second modified plot of the characteristics illustrated in FIG. 4.

FIG. 7 is a third modified plot of the characteristics illustrated in FIG. 4. FIG. 8 is a input -output plot of a hypothetical coiled spring.

DETAILED DESCRIPTION OF THE ILLUSTRATIVE EMBODIMENTS FIG. 1 and the related discussion provide a brief, general description of a suitable computing environment in which the invention may be implemented. Although not required, the present invention will be described, at least in part, in the general context of computer-executable instructions, such as program modules, being executed by a computer 30. Generally, program modules include routine programs, objects, components, data structures, etc., which perform particular tasks or implement particular abstract data types. Tasks performed by the program modules are described herein and with the aid of diagrams and flowcharts. Those skilled in the art can implement the description, diagrams and flowcharts to computer-executable instructions. In addition, those skilled in the art will appreciate that the invention may be practiced with other computer system configurations, including multiprocessor systems, networked personal computers, mini-computers, main frame computers, and the like. The invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computer environment, program modules and/or data may be located in both local and remote memory storage devices.

The computer 30 illustrated in FIG. 1 comprises a conventional computer having one or more processing units (CPU) 32, memory 34 and a system bus 36, which couples various system components, including the memory 34 to the CPU 32. The system bus 36 may be any of several types of bus structures, including a memory bus or a memory controller, a peripheral bus, a network bus and a local bus using any of a variety of bus architectures. The memory 34 includes read only memory (ROM) and random access memory (RAM) . A basic input/output (BIOS) containing the basic routine that helps to transfer information between elements within the computer 30, such as during start-up, is stored in ROM. Storage devices 38, such as a hard disk, a floppy disk drive, an optical disk drive, etc., are coupled to the system bus 36 and are used for storage of program modules and data. It should be appreciated by those skilled in the art that other types of computer readable media that are accessible by a computer, such as magnetic cassettes, flash memory cards, CD-ROM, digital video disks, random access memories, read only memories, and the like, may also be used as storage devices. Commonly, programs are loaded into memory 34 from at least one of the storage devices 38 with or without accompanying data.

An input device 40 such as a keyboard, pointing device (mouse), or the like, allows an operator to provide commands to the computer 30. A monitor 42 or other type of output device (printer, etc) is further connected to the system bus 36 via a suitable interface and provides feedback to the operator. Computer 30 can communicate to other computers, or a network of computers (generally designated at 43) , such as an intranet or the Internet through a wired or wireless communications link, and an interface 44, such as a modem, network card, or the like. In one embodiment, computer 30 can organize, present and solicit information to and from a user through a "Website" commonly used on intranets or the Internet. In such a situation, the computer 30 is identified as a server, while remote computers are identified as clients. Remote users can access the Website using a conventional desktop computer or other Internet device and a browser, such as MICROSOFT INTERNET EXPLORER or NETSCAPE NAVIGATOR.

FIG. 2 schematically illustrates in block diagram form components of a CAE/CAD system 50. As illustrated, CAE/CAD system 50 includes the input device 40 and the output device 42 described above, a simulation/analysis module 52 and one or more models 54. Simulation/analysis module 52 may be a software program and/or hardware accelerator, which cooperates with, or is executed by, CPU 32. Simulation/analysis module 50 may be a conventional simulation environment such as ADAMS™ by Mechanical Dynamics, Inc. of Ann Harbor, Michigan or numerous others. As should be known or obvious to persons of ordinary skill in the relevant art, simulation/analysis module 50 serves to predict or verify behavior of physical systems in response to applied stimuli or input signals given a structural description of the components forming the physical system. The models 54 include individual models 54A, 54B and 54C that herein are empirically based and are typically nonlinear. Although, it should be understood that some of the individual models can include whitebox models used in conjunction with the blackbox models to model the desired physical system. As described below, characteristics of the blackbox models can be varied to understand or test different modifications of the modeled physical system. Individual models can be formed for any type of physical component including mechanical, electrical, optical or other physical components. Generally, each "blackbox" model has one or more inputs and one or more outputs. Various types of modeling techniques can be used such as polynomial or spline curve fits, autoregressive moving average

(ARMA) , frequency response function (FRF) , or neural networks; however, it should be understood that the particular type of empirical model used is not important to the present invention wherein any form of empirical modeling can be employed.

The models 54 can be obtained experimentally with statistical curve fitting techniques. Co-pending application entitled "Method to Generate Drive Signals for Simulation Testing Using Neural Networks", U.S. Serial No. 09,420,023, filed October 18, 1999, which is hereby incorporated by reference in its entirety, describes techniques for obtaining each of the models 54.

Using the input device 40, the user can construct a physical system from the models 54. As appreciated by those skilled in the art, additional lists or information may be solicited from the simulation/analysis module 50, which may depend, for instance on the type of physical system being modeled. This information may be stored in the simulation/analysis module 50, or in databases accessible to the simulation/analysis module 50. Such information may also be dependent on the type of simulation/analysis module 50 being used.

FIG. 3 schematically illustrates the model 54A, by example, having a plurality of inputs 60 and a plurality of outputs 62. In a first aspect of the present invention, means are provided for adjusting one or more of the inputs and outputs of one of the models forming the physical system. As illustrated herein, the means for adjusting includes a gain adjuster 66, preferably one gain adjuster for each of the inputs to the model 54A. Similarly, a gain adjuster 68 can also be preferably provided for each of the outputs from the model 54A.

Effects of input and output gain adjustment may be best understood graphically as re-scaling the axes of an input - output plot. Fig. 4 illustrates force versus displacement characteristics of a mechanical damper such as an automobile shock absorber In this example, the model of the shock absorber includes a single input (displacement or velocity) and a single output (force) ; however, depending upon the desired physical component to be modeled a plurality of inputs and/or a plurality of outputs may be present.

Fig. 5 illustrates adjustment of the input adjuster 66, while the output adjuster 68 remains unity. As illustrated, a gain value greater than one of the input adjuster 66 in effect stretches the original model of Fig. 4 along the horizontal axis. For a gain value less than one of the input adjuster 66, the plot would be compressed along the horizontal axis .

A similar result is illustrated in Fig. 6 for an adjustment of the output adjuster 66, while the input adjuster 66 remains unity. In particular, a gain value greater than one of the output adjuster 68 in effect stretches the original model of FIG. 4 along the vertical axis. Whereas, for a gain value less than one of the output adjuster 68, the plot would be compressed along the vertical axis. As described above, in one embodiment, each individual input and output signal to the blackbox model includes an independent gain adjuster 66 or 68. Each adjuster functions as a multiplier for the associated signal. In one embodiment each adjuster is nominally set to 1.0, but other default values may be used. As appreciated by those skilled in the art, the range and scale of each adjuster may vary depending on the physical component being modeled as well as each corresponding signal of the model. In addition, the scale can be linear, logarithmic, etc.

Simulation/Analysis model 50 receives values indicative of adjusting the adjusters 66 or 68. For instance, dialog boxes allowing the user to vary each of the adjusters can be provided. Suitable labeling is provided, if known, for each of the signals. After adjustment of one or more of the adjusters as a function of the received values, simulation/analysis module 50 executes instructions to obtain an output of the results of analysis of the physical system with the modified models. The adjusters 66 or 68 can be stored in the simulation/analysis module 50, in another database accessible to the simulation/analysis module 50, and/or within each module of the plurality of models 54.

In another aspect of the present invention, simulation/analysis model 50 includes means for adjusting the time scale of one or more of the models 54. All dynamic models (blackbox or whitebox) have an associated time scale. The time scale may be represented implicitly, in the parameters of a model. For example, a fluid viscosity parameter involves a velocity gradient term, which includes a time dimension or parameter. Alternatively, the time scale may be defined explicitly. For instance, non-linear blackbox models can be defined using discrete time signals. The discrete time signals have a corresponding time increment or sample rate that defines the time scale for the entire model. Time scale adjustment can be achieved using one adjuster for the entire model. However, it should be noted that the discrete input and output signals may require interpolation for sample rate conversion in order to match the model time scale. In addition, different time scales can be used for different models of the physical system.

Although the effects of input and output gain adjustment can be understood graphically, as discussed above, the adjustment of time scale is less obvious. Technically, there is no simple graphical interpretation of adjusting the time scale. Rather, the input -output plot may exhibit complex changes in the "shape" when the time scale is adjusted as illustrated in Fig. 7.

One of the advantages of providing gain and/or time adjustment of a model allows the user to explore other possible physical components to be used in the physical system, concentrating more on what is desired rather than trying to ascertain which parameter of the component model will achieve the desired effect. For many components, a given blackbox adjustment may represent the combined and simultaneous change of several physical parameters, thereby making it difficult in some circumstances to determine which parameters of the physical component in a real-life prototype will be needed. In some cases, detailed understanding of the internal dynamics of the physical component is needed to make the necessary associations. Nevertheless, these disadvantages do not negate the value of being able to adjust a blackbox model in order to explore possible components or systems in empirically based modeling techniques.

Although associating the adjusters of the blackbox model with physical parameters of each physical component may be difficult in view of the wide variety of models that can be built and corresponding inputs and outputs that can be measured, some characteristics can still be drawn. In particular, changing the time scale of a model can significantly affect the model wherein the physical component has a time or frequency dependent parameter. In the shock absorber blackbox model described above, changing the time scale can be effectively equivalent to changing the kinematic velocity of the fluid within the shock absorber since the corresponding physical dimensions of kinematic velocity are dependent upon time. Thus, a strong correlation can be obtained between changing the time scale of models of physical components where the physical component has one or a few parameters related to time or frequency. In other words, by changing the time scale of a blackbox empirical model the same effect as changing the time dependent parameter of a theoretical, physical component model can be realized. Although illustrated herein where a fluid viscosity of fluid in a shock absorber has been described, it should be understood that other time dependent parameters can also be used. For instance, other time or frequency dependent parameters related to fluid flow of other mechanical physical components can also be used. Likewise, time or frequency dependent parameters related to voltage or current

(e.g. capacitance or inductance) of electrical components can also be used. Similar correlations of time dependent parameters may also be found in other physical component models of optical or hybrid components and should therefore be considered within the scope of the present invention.

It should also be understood that in some situations, adjustment of the input and/or output adjusters may also have strong correlations to parameters of a physical component. For instance, Fig. 8 illustrates an input-output plot for a coiled spring, having non- linear displacement characteristics associated with "bottoming out" of the coils as indicated at 72. For a blackbox model of the coil spring having an input adjuster and an output adjuster, adjustment of the input gain can be equivalent to stiffening or softening the spring, without changing the saturation load (as indicated with dashed lines 74) . In the coiled spring, this would correspond to increasing or decreasing the number of coils without changing the wire diameter or the overall cylindrical diameter of the coiled spring. Likewise, adjustment of the output gain is equivalent to stiffening or softening the spring, without changing the saturation displacement (as indicated with dashed lines 76) . Generally, this would correspond roughly to changing the cylindrical diameter of the coiled spring, without changing the wire diameter or the overall spring length.

In summary, the present invention allows an operator to vary the input, output or time scale of empirical models used in a CAE/CAD system. This invention thereby allows the operator to explore different components within the physical system in order to understand effects as well as improve performance of the physical system. Unlike theoretical models, the operator is not required to know what physical parameter needs to be changed in the physical component of a theoretical model in order to achieve the desired result. If desired changes have been realized in the physical system by varying one or more of inputs, outputs or time scale of one of the component models, further analysis then can be performed to replicate the desired characteristics by component parameter changes . Although the techniques described herein can be used with linear and non- linear models, a significant benefit is obtained when applied to non-linear models, which were historically hard to model accurately. Although the present invention has been described with reference to preferred embodiments, workers skilled in the art will recognize that changes may be made in form and detail without departing from the spirit and scope of the invention.

Claims

WHAT IS CLAIMED IS:

1. A computer aided engineering system comprising: memory for storing empirical models of physical components defined by inputs and outputs; an output device ; and a processor operably coupled to the memory and the output device, the processor executing a module having instructions for modeling a physical system from the empirical models of the physical components, the module including means for adjusting at least one of the inputs, outputs or time scale of one of the models forming the physical system.

2. The computer aided engineering system of claim 1 wherein the means for adjusting includes a gain adjuster for at least one of the inputs to said one of the models.

3. The computer aided engineering system of claim 2 wherein the means for adjusting includes a gain adjuster for at least one of the outputs to said one of the models.

4. The computer aided engineering system of claim 1 wherein the means for adjusting includes a gain adjuster for each of the outputs to said one of the models .

5. The computer aided engineering system of claim 1 wherein said one of the models includes a gain adjuster for each of at least one of the inputs and outputs and the means for adjusting selectively adjusts the gain adjusters .

6. The computer aided engineering system of claim 1 wherein the module includes means for adjusting time scales of a plurality of model differently.

7. The computer aided engineering system of claim 1 wherein said one of the models for adjusting the time scale is of physical component having a time dependent parameter .

8. The computer aided engineering system of claim 7 wherein the time dependent parameter comprises a parameter related to fluid flow.

9. The computer aided engineering system of claim 8 wherein the time dependent parameter comprises fluid viscosity.

10. The computer aided engineering system of claim 7 wherein the time dependent parameter comprises a parameter related to one of voltage and current

11. The computer aided engineering system of claim 10 wherein said one of the models is of a component having inductance .

12. The computer aided engineering system of claim 10 wherein said one of the models is of a component having capacitance .

13. The computer aided engineering system of claim 8 wherein said one of the models is of a fluid damper.

14. A computer-implemented method of engineering analysis of a physical system from empirical models of physical components defined by inputs and outputs, the method comprising: receiving values indicative adjusting at least one of the inputs, outputs or time scales of one of the models; adjusting said at least one of the inputs, outputs or time scales of one of the models as a function of the received values; and outputting results of analysis of the physical system having the adjusted said at least one of the inputs, outputs or time scales.

15. The computer-implemented method of claim 14 wherein adjusting includes adjusting a gain adjuster for at least one of the inputs to said one of the models .

16. The computer-implemented method of claim 15 wherein adjusting includes adjusting a gain adjuster for at least one of the outputs to said one of the models .

17. The computer-implemented method of claim 14 wherein adjusting includes adjusting a gain adjuster for at least one of the outputs to said one of the models .

18. The computer-implemented method of claim 14 wherein said one of the models includes a gain adjuster for each of at least one the inputs and outputs and adjusting includes adjusting the gain adjusters.

19. The computer- implemented method of claim 14 and further comprising adjusting a time scale of at least one of the models.

20. The computer-implemented method of claim 19 wherein adjusting a time scale includes adjusting time scales of a plurality of model differently.

21. The computer- implemented method of claim 29 wherein said one of the models for adjusting the time scale is of physical component having a time dependent parameter.

22. The computer-implemented method of claim 21 wherein the time dependent parameter comprises a parameter related to fluid flow.

23. The computer-implemented method of claim 21 wherein the time dependent parameter comprises fluid viscosity.

24. The computer- implemented method of claim 21 wherein the time dependent parameter comprises a parameter related to one of voltage and current .

25. The computer- implemented method of claim 24 wherein said one of the models is of a component having capacitance .

26. The computer- implemented model of claim 24 wherein said one of the models is of a component having inductance .