EP1145171A2

EP1145171A2 - Method and arrangement for designing a technical system

Info

Publication number: EP1145171A2
Application number: EP99936427A
Authority: EP
Inventors: Claus Hillermeier
Original assignee: Siemens AG
Current assignee: Siemens AG
Priority date: 1998-06-17
Filing date: 1999-06-07
Publication date: 2001-10-17
Also published as: EP1145171A3; WO1999066434A3; WO1999066434A2

Abstract

In order to design a technical system having a predetermined number of target functions, each target function is provided with a weight parameter, said weight parameters being changed in a predetermined manner. In order to conduct multitarget optimization, an efficient parameter vector that may include operational points and layout parameters of the technical system is detected. The technical system is designed, adapted or adjusted on the basis of said parameter vector.

Description

description

Process and arrangement for the design of a technical system

The invention relates to a method and an arrangement for designing a technical system.

For a complex technical system, significant sizes such as costs for the manufacture or efficiency of the system are of interest in an early planning phase or during operation. A dependency of each of these variables on a predetermined set n of parameters influencing them (operating parameters), summarized in a parameter vector x of dimension n, is detected by a target function.

If there are several target functions with competing targets, it is to be ensured in the context of a so-called multi-target optimization that the parameters simultaneously represent a solution with sufficient good for all given target functions.

There are numerous numerical methods used to solve

Systems of equations are used. Implementations of such methods m algorithms that can run on computers are widespread. The following numerical methods are particularly noteworthy:

From [1] is a numerical method for non-linear

Optimization, e.g. the BFGS variant of the Quasi-Newton method, known. An SQP method for nonlinear optimization with constraints also emerges from [1]. [2] describes the Newton method for solving a system of non-linear equations.

The object of the invention is to provide a method and an arrangement for designing a technical system, at least in the context of multi-target optimization a parameter vector is determined, which is a solution with sufficient good with regard to the design of the technical system.

This task is carried out according to the characteristics of the independent

Claims resolved. Further developments of the invention also result from the dependent claims.

To solve the problem, a method is specified in which the technical system has a predetermined set of k target functions

(f _λ , f ₂ _,. -., f _k ⁾ = f ^T

each target function preferably of a predetermined set of n parameters

summarized m a parameter vector x, can be influenced. Each target function is also provided with a weight parameter. After changing the weight parameters in a predetermined manner, parameters for the design of the technical system are determined. These parameters preferably represent a so-called efficient parameter vector of the technical system.

The technical system is optionally designed, which results in the described determination of the parameter vector with n parameters, or a parameter vector of an already implemented system is designed by determining a parameter vector for the existing system with regard to the given target functions, which parameter vector has a high Guaranteed good of the system.

This good is evaluated by means of the k objective functions, the objective functions competing with one another, ie with the Improving a value for one objective function, the value of at least one other objective function deteriorates. Examples of competing objective functions are:

f ^τ = (f _] _, ± 2) = (system efficiency, investment costs)

or

f ^T = (f] _, f2, 1: 3) = (product throughput, probability of failure, pollutant emissions).

For a successful design, i.e. a successful compromise between the competing target functions, several alternative solutions are automatically determined. Each of the alternative solutions is "efficient", i.e. this alternative solution (= parameter vector) can no longer be changed without resulting in a deterioration for at least one value of a target function.

As part of a remote design of the technical system, e.g. a technical system, sometimes a parameter vector (as an operating parameter of the system or a design parameter for the system) cannot be implemented, which is why an alternative solution is indispensable.

The invention enables an automatic generation of a number of alternative solutions, the ede of which represents an efficient implementation with regard to the setting or design of the technical system or the technical system.

Individual alternative solutions are determined by step-by-step modification of already determined or predetermined efficient solutions, in particular by changing the weight parameter vector α. A further development of the invention then consists in the following steps for designing the technical system: a global objective function is specified in accordance with

9α α ^τ • f

where τ denotes a transposed vector with k weight parameters, f denotes a vector with k target functions. The k objective functions tension while a k-space aauuff .. eess wwiirrdd eeiinn eerrsstteerr VVeekkttoorr xx _ss mmii "t parameter values determined by the equation system

will be solved. A new vector ct _new with weight parameters is determined by the relationship

is determined, where c denotes a predetermined constant and δ denotes a predetermined direction in k-space. Along the direction δ m of the k-space, stationary values for the parameters (the parameter vector) are determined using an approximate value

is determined, whereby

the gradient referred to and this approximate value x as a starting value for a method for solving a non-linear system of equations

is specified. Finally, the solution to this non-linear system of equations determines the parameters for the design of the technical system.

It is particularly advantageous that, starting from a parameter vector x along a predetermined direction δ in k-space (see Equation (3)), i.e. in the space spanned by the k target functions, a new stable parameter vector according to Equation (5 ) can be determined and thus the technical system can be designed for a previously unknown stable parameter vector. At this multidimensional operating point (parameter vector), stability of the technical system is ensured.

Examples of target functions are costs caused by the system, exhaust gas emissions, probability of failure and efficiency of the technical system.

A further development consists in that iteratively along further directions δ _p (with p as an index for each of the further directions) in each case for each direction

Parameter vector for the design of the technical system is determined. This can be done in an automated manner, in that in each case a stable one in each direction δ

Operating point is determined.

One embodiment of the invention is that a numerical method, for example the Newton method, is used to solve the systems of equations. In a further embodiment, the technical system is implemented or set on the basis of the determined parameters. This has the advantage that the parameters m of a parameter vector, which was determined by means of the invention, identify a stable operating point and the setting of the system to this operating point ensures a safe state of the system.

Another development is that the parameters of the target function can be operating parameters or design parameters. Operating parameters identify possible adjustable sizes, while design parameters in particular identify physical dimensions of the technical system and cannot be adapted or changed at all or only with great effort during operation.

It is also a further development of the invention that secondary conditions are taken into account for solving the systems of equations (2) and (5).

Furthermore, an arrangement is specified to solve the task, which has a processor unit, which processor unit is set up in such a way that the following steps can be carried out:

- The technical system has a predetermined set of k target functions

(f !, f ₂ , ..., f _n ) = f ^T , with each objective function preferably having a predetermined set of n parameters

(xτ_, x ₂ , .... x _n ) = x ^T ,

summarized m a parameter vector x, can be influenced. - Furthermore, each target function is provided with a weight parameter.

- The weight parameters are changed in the specified manner.

- From the target functions provided with the changed weight parameters, parameters for the design of the technical system are determined, which parameters in particular the stability of the technical

System.

In a further development, the arrangement with the processor unit is designed such that the following steps can be carried out:

- For the technical system, a global target function is specified in accordance with equation (1), the technical system having a predetermined set of k target functions and each target function being able to be influenced by a set of n parameters (collectively m a parameter vector x);

- by loosening a system of equations (2), a first parameter vector x _{s is} determined;

- a new vector with a weight parameter vector —new ^wιri determined by equation (3);

- along the direction δ becomes a stationary one

Parameter vector (stationary parameter values for this parameter vector) is determined by determining an approximation parameter vector x using equation (4), this approximation parameter vector x as a starting value for a method for solving the non-linear

System of equations according to equation (5) is used; - The solution to equation (5) determines the parameters for the design of the technical system.

This arrangement is particularly suitable for carrying out the method according to the invention or a further development described above.

Exemplary embodiments of the invention are illustrated and explained below with reference to the drawing.

Show it

Fig.l is a block diagram with steps to design a technical system;

2 shows a processor unit.

Fig.l shows a block diagram with steps to design a technical system. The technical system comprises a predetermined set of k target functions, one target function representing the costs for the technical system, another target function representing the efficiency of the technical system. Each objective function comprises a set of n parameters, these parameters being combined as a parameter vector. The parameters (also: parameter values) are, in particular, operating parameters or design parameters.

A global objective function according to equation (1) is determined by

9α = ^α l ^f l + α ₂ f2 + - - • + ^θ - ^t da).

Corresponding to the k target functions, there are therefore k weight parameters α ^, combined m a weight parameter vector α. In a step 101, a first (parameter) vector x _{s is} determined by solving equation (2). A numerical method for non-linear optimization, for example the BFGS variant of the quasi-Newton method (cf. [1]), is used in particular for this solution.

The result of the solution to equation (2) is a stable parameter vector x _s .

Starting from the weight parameter vector α (in k-space, that is, the space that is spanned by the target functions), in k-space along a predetermined direction δ with a discretized length | δ | a new weight parameter vector ct _newly determined in accordance with equation (3) (cf. step 102).

In a step 103, an approximation value (approximation parameter vector) x is determined. This approximate value x serves as the starting value for a method for solving a non-linear system of equations according to equation (5)

(see step 104). To solve this system of equations e.g. used the Newton method (cf. [2]).

Step 105 shows the determination of the parameters for the design of the technical system from the solution of equation (5).

Alternatively, as indicated in FIG. 1 with block 106, additional conditions are also taken into account.

The following equality constraints apply

h _x (x) = 0, ι = l, 2, ..., m (6)

and / or the following inequality constraints

h _x (x) <0, ι = m + l, ..., m + r (7) given. The parameter space is thus expanded: m + r are added to the n parameters (combined in the parameter vector x)

Lagrangian multipliers λ ₁ (with ι = l, 2, ..., m + r) and rr slip variables. These slip variables are suitable for obtaining equations from inequalities in accordance with equation (7), in particular in the form

In step 101 there is now a system of equations

m + r

Vg _α ( ^χ ) - ∑ λ_L Vh _x (x) = o: n equations

1 = 1

m Equality constraints

λm + 1 ⁱ + l (x)

λm + r hm + r (x)

r Inequality constraints

to solve. To solve the system of equations (8), an SQP method for linear optimization with constraints is preferred (see [1]). Equation (8) is to be used instead of F m equation (4) in order to obtain the approximate value (approximation parameter vector) x. Likewise, the

Supplementary conditions (6) and (7) in step 104 (equation (5)) are taken into account by ö ^χ _m δx _n .

by F _{α + C} § is used analogously to the explanations and equation above. In this case, equation (5) becomes a + cδ = ° A processor unit PRZE is shown in FIG. The processor unit PRZE comprises a processor CPU, a memory SPE and an input / output interface IOS, which is used in different ways via the interface IFC: an output is visible on a monitor MON and / or on a printer via a graphic interface PRT issued. An entry is made using a mouse MAS or a keyboard TAST. The processor unit PRZE also has a data bus BUS, which ensures the connection of a memory MEM, the processor CPU and the input / output interface IOS. Furthermore, additional components can be connected to the data bus BUS, for example additional memory, data storage (hard disk) or scanner.

Bibliography :

[1] R. Fletcher: "Practical Methods of Optimization", John Wiley & Sons, Winchester 1987, pages 48-55 and 304-307.

[2] H. R. Schwarz: "Numerical Mathematics", Teubner Verlag, Stuttgart 1997, pages 213, 214.

Claims

claims

1. procedure for the design of a technical system,

a) in which the technical system has a predetermined set of k target functions;

b) in which each objective function is provided with a weight parameter;

c) in which a predetermined change in the weight parameters is carried out;

d) the parameters for the design of the technical system are determined on the basis of the changed weight parameters.

2. The method according to claim 1,

a) in which each objective function is influenced by a predetermined set of n parameters;

b) where a global objective function

g _α = α ^{τ •} f

is specified, whereby

T 1 is a transposed vector with the

Weight parameters and f denote a vector with k objective functions;

where the first vector x _{s is determined} with parameter values by the system of equations

is solved;

d) with the new vector αn _n eu. is determined according to

where c denotes a predetermined constant and δ denotes a predetermined direction in k-space;

e) in which stationary values for the parameters are determined along the direction δ by using an approximate value according to the relationship

is determined, whereby

F the gradient referred to and this approximation as an initial value for a method for solving a non-linear system of equations

is specified;

f) where by the solution of the non-linear

Equation system the parameters for the design of the technical system are determined.

3. The method according to claim 2, wherein parameter values for the design of the technical system are determined iteratively for further directions.

4. The method according to claim 2 or 3, in which is used to determine the first vector x _s em QuasiNewton method.

5. The method according to any one of claims 2 to 4, in which e Newton method is used to solve the system of equations from step e).

6. The method according to any one of the preceding claims, in which the technical system realizes on the basis of the parameters for the design.

7. The method according to any one of the preceding claims, in which an adaptation of the technical system takes place.

8. The method according to any one of the preceding claims, wherein the parameters of the target function are operating parameters or design parameters of the technical system.

9. The method according to any one of claims 2 to 8, in which additional equality constraints and / or inequality constraints are specified in the solution of the system of equations from step d).

10. Arrangement for the design of a technical system, with a processor unit which is set up in such a way that

a) the technical system has a predetermined set of k target functions; b) each target function is assigned a weight parameter;

c) a predetermined change in the weight parameters can be carried out;

d) parameters for the design of the technical system are determined on the basis of the changed weight parameters.

11. The arrangement according to claim 10, wherein the processor unit is set up in such a way that

a) the technical system has a predetermined set of k target functions;

b) each objective function can be influenced by a predetermined set of n parameters;

c) a global objective function

g _α = e ^{τ •} f

can be specified, where α ^τ denotes a transposed vector with weight parameters, f denotes a vector with k target functions;

d) e first vector x _s with parameter values can be determined by the equation system

is solved; A new vector can be _newly determined according to

f) stationary values for the parameters can be determined along the direction δ by using an approximate value according to the relationship

is determined, whereby

is determined;

g) the parameters for the design of the technical system can be determined by the solution of the non-linear system of equations.