EP1018087A1

EP1018087A1 - Reduction process for knowledge data generation simulations

Info

Publication number: EP1018087A1
Application number: EP98948855A
Authority: EP
Inventors: Rainer Burkhardt; Andreas Heinzelmann; Bernd Rehfus
Original assignee: DaimlerChrysler AG
Current assignee: Mercedes Benz Group AG
Priority date: 1997-09-26
Filing date: 1998-08-21
Publication date: 2000-07-12
Also published as: DE19742450A1; WO1999017238A9; JP3439741B2; JP2001518671A; WO1999017238A1

Abstract

A process for reducing the number of simulation steps required in a simulation process can be used for many different simulation processes for generating, by means of a computer, knowledge data on a total system composed of several electrically controlled elements. The elements are decomposed into components which comprise one or several basic elements. The electrical connections between the basic elements and between the components are then detected, discrete electrical state values are associated to the basic elements, and operating states, as well as possible states of the individual components, are defined in relation to the electrical state values of the basic elements. The variables which can be measured at the basic elements and components and are required to generate knowledge data are then determined and the basic elements which have no influence on the thus determined variables are determined, gathered and eliminated. The number of required simulation steps and the computing time required to carry them out can thus be reduced by several powers of ten.

Description

Title: Reduction process for simulations for

Knowledge data generation

DESCRIPTION

The invention relates to a method for reducing the number of simulation steps required in a simulation method for computer-controlled generation of knowledge data about an overall system consisting of several electrically controllable components. Furthermore, the invention relates to a computer for carrying out a reduction process.

For a multitude of applications of complete systems consisting of electrically controllable components from the most diverse areas, e.g. The control of production lines or high-bay warehouses, robot technology, elevator control, circuit development, vehicle electronics or the diagnostic systems required for this, it is necessary to know the normal and faulty behavior of the individual components, the effects of this behavior on other components and in particular on the overall system .

Through data material about the overall systems, e.g. from their development phase, it is possible to describe the behavior of the components using models and to generate computer-controlled knowledge data about the behavior of the individual components by simulation. Known simulation methods allow the simulation of individual components and subsystems, but fail very often if the simulations are to be extended to the overall system. This is essential if the effects of individual component behavior or the behavior of subgroups on the 02s overall system are to be examined. This requires countless simulation runs, which require immense computer capacities and are also very time-consuming.

US Pat. No. 5,625,578 describes a control method for simulations which are used to investigate the electromagnetic behavior of electrical components of a printed circuit board. The simulations are based on a model equivalent to the circuit pattern of the circuit card. The number of equivalent lent circuit components is reduced by four orders of magnitude without incurring losses in terms of the accuracy of the equivalence model. The circuit pattern is represented by geometric elements. For the equivalence model, groups of main elements are compiled that contain only those selected elements whose distance from the neighboring elements exceeds a defined minimum wavelength λ. When setting up matrix equations that represent Maxwell's equations, field values are assigned to the selected elements. The field values of the elements not taken into account are defined via the field values of the selected elements, and a matrix representing the equivalence model is generated using a ranking method that takes into account the number of selected elements.

The object of the invention is to provide a reduction method for simulations for the generation of knowledge data, which considerably reduces the number of simulation steps required, without ignoring information relevant for the generation of knowledge data.

According to the invention, this object is achieved by the method described in claim 1 and by the computer having the features of claim 12.

The method reduces the number of simulation steps required in simulation methods that are used for the computer-controlled generation of knowledge data about complete systems, which consist of several electrically controllable components. The electrically controllable components are broken down into components that comprise one or more basic components. The electrical connections between the basic components and between the components are then recorded, discrete electrical status values are assigned to the basic components, and operating statuses and possible component statuses of the individual components are defined in relation to the electrical status values of the basic components belonging to a component. Subsequently, the sizes necessary for the generation of knowledge data and measurable on the basic building blocks and components are determined and those basic building blocks are determined which have no influence on the defined measurable sizes. Finally, the basic elements without influence are summarized and eliminated.

With this method, the number of simulation steps required and the amount of computing time required to carry them out can be reduced by several powers of ten. The DΞS process can be used for many different simulation processes that are used for the computer-controlled generation of knowledge data about an overall system consisting of several electrically controllable components. The division of knowledge data into knowledge data types enables the selection of one or more knowledge data types that are relevant to a specific question. This is advantageous because the quantities that can be measured on the basic building blocks and components are only to be specified for the selected knowledge data types and the number of simulation steps required is reduced again.

It is particularly advantageous to summarize previously determined basic modules which influence the same spatially limited knowledge data area. The formation of local founders takes into account the knowledge of the components and the overall system by introducing the knowledge that not all operational and component states affect all knowledge data. There are spatially limited areas of knowledge data in which only certain component states have an effect. Determining and summarizing the locally influencing basic building blocks into basic clusters makes it possible to reduce a simulation that initially encompasses the entire system to the essential simulation steps and to limit the global simulation to a limited range of knowledge data.

Further advantageous refinements and developments of the method according to the invention and the computer according to the invention are set out in the subclaims.

Preferred embodiments of the invention are shown in the drawings and are described below.

1 shows schematically the sequence of the reduction process;

2 shows the relationship between the components and the states of the individual components;

3 is a compilation of different components with the assigned discrete electrical state values as well as the possible operating and component states;

Fig.4 shows the relationship between the basic building blocks and the components;

5 shows the creation of a reduction graph;

6 shows the principle of the creation of a reduced subgraph from any Teiigraphen;

Figure 7 illustrates the work of binding operators;

Fig. 8 shows an extract from a relation table. The sequence of the method for reducing the number of simulation steps is shown in FIG. 1 Represent procedure.

The main components that occur in the periphery of automation devices are, as summarized in the table in FIG. 3, resistors, switches, lines, connectors, lamps, bundles of cables, sensors, relays, fuses, solenoid valves, motors and signal transmitters. Some of the components can be summarized in aggregate A. The aggregate information with the structure relationships is usually available in a data processing form. This information is already available during the product development phases e.g. present in CAD systems or can be derived from model-based programs.

An aggregate A is a set of components C. with a local arrangement. An aggregate describes a smallest exchangeable unit. The individual units A are connected to one another by the pins Pl, the components C _j .

The cardinalities of the relationships of aggregates are shown in the left half of Fig. 2. The relationships can be defined as follows for the general and system-independent presentation of components:

• An aggregate A; consists of zero or more components C _j . A component C _j belongs exactly to an aggregate A ,.

• An electrical component C _j has zero or more ports PO _k . A port PO _k is part of a component C _j .

• Each port PO _k can contain one or more pins Pl _m . Each pin PI _m is assigned to exactly one port PO _k .

• A connection CH, refers to two pins Pl _m . A pin Pl _m can be related to none or a connection CNi.

Common to all the components considered here is the linear behavior in certain working points. Nonlinear components such as Diodes are replaced by processor-controlled automation devices. On the one hand, the software functions simulate the non-linear behavior and, on the other hand, the non-linear components are integrated into the automation devices, so that the behavior can be described linearly outside the systems within the framework of the relevant information.

The errors associated with these components are mainly limited to discrete errors. Experience has shown that the proportion of discrete or discretisable errors with a static error predominates in the periphery of the automation devices Behavior. Continuous errors cannot be ruled out, but can generally be neglected.

For the simulation it is sufficient to take the static properties of the components into account. Dynamic properties of the components considered here play a subordinate role, since the dynamic behavior properties of the components considered do not provide a significantly higher information gain.

Due to their practical relevance, the following considerations mainly focus on the passive periphery. Active components can be analyzed using the same method, but generally play a subordinate role. With a purely passive periphery, all components show an electrical resistance with different resistance values. Current and voltage sources are added to these resistors in active circuits. These primary components are called basic building blocks or atomic units.

Each basic building block U, refers to a component C ,. The basic building blocks or atomic units differ depending on the type in active and passive basic building blocks. The active basic components are divided into current sources Ql and voltage sources QV, whereas the passive basic components contain simple resistors R _k . The type of the basic module is characterized by type (U. The basic module U _j can assume different discrete values V _k depending on the state of the component. The individual discrete values V _k of component C are related to one another by the states of the components via a behavior description B _k coupled

The passive basic building blocks are considered in the further considerations. The active building blocks are listed for completion.

The behavior of the components is characterized by discrete states. As shown in Fig. 4, this results in discrete values for the basic building blocks, which represent different electrical states of the components. The individual discrete states of the basic building blocks have a specific relationship to the component states, which include the standard behavior and the fault behavior of a component. In addition, it can be used for many applications such. B. for simulation methods for generating diagnostic knowledge, be of interest in the operating state of the component. Electrical and non-electrical operating states can be distinguished depending on the type of physical quantity that causes the operating state change. The electrical operating states are directly influenced by electrical quantities of the basic modules to be examined

Each component C can assume different operating states O. Depending on the influencing factor, the operating state transitions depend on electrical and / or non-electrical events and variables. If there are several operating states, the state transition is primarily either electrical or non-electrical result. The non-electrical operating states ONE can be achieved, for example, by mechanical influences. A component C can assume electrical operating states OE if an electrical effect causes an operating state change. A change between the operating states can be carried out as often as required without restriction.

Each component has at least two component states F _m , which are not changed by electrical variables or by external influences in normal operation. The component states F _m hide at least one faulty state and the normal state. It is therefore of primary importance for simulation processes for the generation of diagnostic knowledge to recognize the component state. A component state change from the normal state to an error state occurs only once. The sporadic errors are an exception.

An incandescent lamp e.g. can be defective or not defective regardless of the operating voltage. This behavior, which is permanently assigned to the component, is the component state. In the component state not defective, different operating states are possible, depending on the electrical quantities available. In general, a distinction between active mode (the lamp is on) and passive mode (the lamp is off) is sufficient.

Depending on the location, the individual components can be assigned to the complete overall system or the periphery of the overall system. In principle, the complete system as a whole can be included in the model, but the information of interest can mainly be found in the periphery.

The periphery of the overall system can be modeled as a graph, in which the basic building blocks represent the edges. To connect the basic building blocks, connection points are required that appear as nodes in the graph. A graph G is given by a set of nodes N and edges E. Each node N, is the junction of two or more edges E _r An edge E, connects two nodes N, and N _r The graph G is called undirected if there is no node order an edge exists. The graph elements GE include both the edges E and the nodes N,

The degree or degree (NJ of the node is determined by the number of edges E, at this node. The edges E _j connected to the node N, are characterized by the edges (N,).

If an edge is a loop, ie if the two nodes of edge N are identical, ioop (NJ returns TRUE, otherwise the result is FALSE. The result of the chain (N _r N,) function is a set of edges E “die directly in a chain mix the elements N, and N _j . Each inner knot of a chain that is not closed has a knot grade 2, the outer knots have a knot grade 3 or larger.

For the periphery in the elementary basic level, ie in the deepest modeling level, a graph has the following properties: The periphery can be described as an undirected graph G _ep = ( _j , Z;). The graph is represented by the edges E; with the atomic units U, and the nodes N _j . A node N is the connection point of two or more basic building blocks U. The entire behavior of the electrical periphery is characterized by the elementary peripheral graph G _ep .

In an elementary peripheral graph G _ep , the node potentials P _t or the currents l _j can be measured by a basic building block. The measurability of the node potential P; or the branch current I depend on the local position and the structure of the entire system and can assume the values measurable or not measurable. The potential P, at the node N, also contains a fixed value V _fi or an unknown value V _vi . For branch currents I, fixed or unknown values result analogously. The measurability of a physical quantity can be clearly assigned to a basic building block U, or a node N. The periphery can thus be represented as an undirected graph.

The relationships shown in FIG. 2 result between the information which can be extracted from, for example, CAD data and which describes the structure and the individual components and the states of the individual components.

• At least one component state F _{m can be} assigned to each component C-. Each component state F _m belongs to one component C ,.

• The operating state O is exactly one component Cj. assigned. A component C; can contain one or more operating states O.

• The operating state O can be clearly caused by measurable quantities.

• At least one node belongs to each component C _j .

• Each component state F _m is clearly described by the discrete values of the basic building blocks or atomic units.

On the basis of the elementary peripheral graph G _ep introduced above, a graph reduction can be carried out which contains the essential behavioral properties with regard to the observable physical quantities. As shown in FIG. 5, a reduction graph G _{r can} consist of several subgraphs. These include the elementary peripheral graph G _ep , several intermediate graphs G _t and a head graph G _h . The nodes and edges of the individual subgraphs are connected to each other and represent the reduction rule.

The basic operations known from network analysis are used to construct the reduction graph. The basic consideration is that under certain conditions branches in an electrical network can be represented by an equivalent branch of a simpler structure. In addition to the individual reduction operations, the efficient construction of a reduction graph requires equivalency operation. in which the network structure is converted into an equivalent form of representation and this ultimately forms the starting point for a further reduction step. The equivalence operations do not reduce the network structure, ie the number of components in the parent graph does not change compared to that of the current graph

The operations required for this are detailed below. It should be noted that the individual values of the basic building blocks or atomic units are not just a single value, but a set of possible values. These values describe the behavior of the component.

O-Resistance Reduction:

y R = {0} N,> N

Loop reduction:

Series reduction:

u ₁₂ u ₂ "u se

* 1 »2 N Y, N n

The series reduction is applicable to basic modules of the type resistors R _κ and voltage sources Q _v . The following applies to these basic building blocks U, with the values V _k :

r _* = ∑ ^ -i) (0-1)

; = 1 for resistors V = R for voltage sources V = U The following applies to the discrete values V = {0, ∞):

³ * V, = * = * V ^• e - co (0.2)

^V. ) ^{= Te} ^ ^{a 0} (0.3)

Parallel reduction:

The above reduction method can be used for the type of the basic module: resistors R _k and current sources Q:

for resistors R = 1 / V for current sources Ql = V

The following applies to the discrete values V = {0, ∞}:

(O.6J

Stem polygon reduction:

n-beam resistance star complete n-corner The applicability of the stem polygon reduction is limited to resistors R. In general, the following applies to the calculation of the corner resistance value:

With the leg resistors:

R _m = {R m = μvm = v} (0.8) and the residual resistances:

R _SΛn = {o \ 0≤m≤n \ {ß, v}} (0.9)

The following relations apply to the discrete values R = {0, ∞} after considering the limit value:

3R _Saι = ∞ => R _μv = ∞ (0.10)

VR _Snι = 0 ^ R _μv = 0 (0.11)

with k = number of O-value residual resistors

If a star consists of n edges and u nodes, the reduced Teiigraph comprises n (n-1) / 2 edges and u-1 nodes. This leads to a reduction effort for adding and removing edges in the order of magnitude O (n ² ).

The stem polygon reduction can be carried out according to a sorted list in which the higher order nodes are processed first. However, the practical effects on the reduced graph are not significant, since the order on the stem polygon reduction only changes the shape of the reduction graph, but not the header graph. Voltage source reduction:

The above source reduction applies to resistors R _k and voltage sources QV, with the node degree n> 2 in the arrangement shown.

Power source reduction:

For a combination of resistors R _k and current sources Ql, the current source reduction operation shown below applies. The current source reduction also applies to higher order nodes of the same structure.

Equivalence operations for voltage sources with resistors:

The aim of the equivalence operation for voltage sources and resistors is to sort the individual basic modules according to their type and to prepare them for a subsequent series reduction.

The values of the individual graph elements GE represent operating and component states. The primary goal is to determine the component status and from this the system status. The set of all system states is thus the permutation of all individual component states, which in turn are represented by the discrete values. The overall or partial system state is composed of the permutations of the discrete values Vi of the individual basic components U,. To see the overall behavior of a child graph in the parent To transfer graphs, it is necessary to create the permutations between the individual values.

The O resistance reduction, loop reduction, voltage source reduction, and current source reduction operations remove graph elements without adding new elements. No special procedures are necessary for these operations. The operations series, parallel and stem polygon reduction insert new values in the graph, so that the creation of the permutations and the corresponding processing is necessary.

6 shows the basic processing of the individual values in order to generate a reduced partial graph G _{sub red} from any subgraph G _{sub 0Pg} . A sequential permutation method is used to calculate the permutations. However, for subsequent derivation methods it is essential that the order of the individual values remains and that a lexicographic order is not carried out.

The number of values of the result z of a mathematical function of the form

/ (* ι, * ₂ , * „) = z (0.16) with x, as parameter and the discrete values

x, e {x _n , x _a x (0.17)

results from v, = [v, ≤ my (v,) ^« (0.18)

Equations (0.2), (0.5) and (0.10) reduce the number of discrete result values after the reduction operations series, parallel and stem polygon reduction. Generalized applies:

V rea • uoper (v.) "(0.19)

As can be seen from Fig. 3, the discrete values 0 and ∞ predominantly occur in the electrical components, so that the following applies to practical arrangements:

Every stem polygon reduction with a node degree greater than 3 leads to an increase in the number of edges in the graph. The stem polygon reduction complements the original component states with further combinations, which are referred to below as pseudo states. With this reduction, the increase of n (n-1) / 2 edges leads to an exponential growth of the pseudo states. On the one hand, this can lead to a rapid increase in the computation or storage effort from one reduction step to the next. During processing as shown in FIG. 5, the number of states in the overall system increases with each star polygon reduction step.

To limit the permutations of unrealistic combinations of values caused by the stem polygon reduction, a so-called binding operator is introduced. The binding operator marks all permuted values as a so-called permutation block. The calculation of the values in a reduction step is ultimately not based on the individual values, but rather the entire permutation block with all bindings is taken into account in the calculation

When generating the permutations, elements must also be taken into account that are not necessary for the calculation but are linked by the binding operator. Fig. 7 shows the processing of the binding operations. Despite increasing the elements in a stem polygon reduction, the binding operator prevents pseudo-component states.

This method of calculating the individual discrete resistance values enables any combinations to be taken into account during the reduction process and other undesired combinations to be excluded from the outset. For example, certain types of knowledge data, such as Single errors can be specified and reduced by this method.

By abstracting and reducing the peripheral resistance graph, one or the other behavioral characteristic, represented by the individual discrete values, can occur several times. However, it is sufficient to consider only one behavioral characteristic at a time.

Values that have no binding can be reduced to a single occurrence. In the case of elements with a bond, a combination is sufficient for the further representation of the resistance. The reduction in value can take place directly after a reduction operation, but it can also take place only after several reduction steps.

The starting point for the reduction process is an undirected elementary basic graph G _ep = (U, NJ with node N, and the atomic units U, as edges. The network is reduced with the basic operations described. After each reduction run, the number of nodes N, For the number of atomic units U, in the reduced graph, a reduction caused by the stem-polygon reduction is not guaranteed Basic graph G ^ is processed until no node or edge can be removed from the graph.

At the end of the reduction process, the head graph consists of nodes with fixed or measurable potentials. All nodes with unknown potentials are removed in the head graph G _h . Atomic units U | which contain a current measurement remain in the reduced graph.

For an elementary basic graph G _ep = (U, N) with n nodes N, and n atomic units U _| is the time complexity of the algorithm O (n ² ). For complex overall systems with a large number of components, the time required for creating the reduction graphs is considerable. In order to counter this behavior, it is advantageous to carry out a clustering before the actual reduction process, in which the locality of the individual components is determined.

The basic idea behind clustering is that not all operating and faulty conditions of components affect all system sizes. There are thus spatially independent areas in which certain states of components have an effect, and areas which are not influenced by all components. In order to minimize the resource requirements for the simulation, it is expedient to identify such spatial areas with their elements. Clustering makes it possible to reduce a system-wide problem to the essential elements in the system and thus to limit the global problem to a limited local area.

A cluster C consists of a certain number of elements. All states of the elements in the cluster only affect sizes and error messages in this local cluster area.

The individual areas can refer to different levels of abstraction. The elementary peripheral graph forms a level of abstraction. Due to the mutual coupling of the individual basic building blocks or atomic units via the corresponding components, there is a second level, which contains Ciuster areas at component level.

Clustering on the level of the elementary peripheral graph comprises the basic building blocks as elements. In the overall system, every possible observation belongs to a cluster.

A basic cluster or atomic cluster C _a is a coherent path of basic building blocks or atomic units. Each basic building block U, and the nodes H- in between, can be assigned to exactly one basic cluster C _a . All values that the atomic units and the nodes assume affect only the sizes in the basic cluster C " _a a.

The related components can be recognized with any graph 1 transfer method. In various references Find algorithms for transversing graphs. In contrast to the usual transversal methods, each node and edge traversed must be entered in a list and marked as processed.

The result of clustering at the atomic level is a list of clusters, each of which contains references to the individual basic building blocks and the neighboring nodes that belong to these clusters. A complex elementary peripheral graph is broken down into spatially limited areas of knowledge data. The graph elements in a cluster have no atomic-level relationships with graph elements in another cluster. Each cluster thus forms a completed subproject and can therefore also be described as a focus area.

If the creation of the reduction graph does not use the entire elementary peripheral graph, but rather the individual clusters are taken into account, the time complexity with 2n graph elements in a cluster O (n ² ) is still considered locally. From the global complexity analysis, however, it can be seen that with k Clustem the call of the reduction method only has a time complexity of O (k). This is of essential practical importance for complex systems, since it enables realistic processing of systems with several thousand components

Clustering also offers the possibility of creating the reduction graph in parallel for the overall system. In the best case scenario, all clusters can be processed in parallel. Under these conditions, the time complexity of O (n ¹² ) arises for the creation of the reduction graph with n 'local graph elements.

Individual components can have a relationship to several basic building blocks. The states of the components also relate to one or more basic modules. This creates relations between basic building blocks of different basic clusters.

A component-oriented cluster C _c comprises one or more basic clusters C _E with its basic building blocks U _f and nodes Nj. The basic clusters C _a are connected to one another by the relationships between the components and the basic building blocks. However, only the components in the periphery are taken into account.

The same creation procedures apply for the component-oriented cluster as for the basic clusters. The result of clustering at component level is a list of clusters, each with references to the basic clusters and references to components. The component-oriented clusters form the basis for the relationships between the basic components in the header graph and the states of the components.

After a simulation has been carried out, the simulation results must be assigned to the component states again. A table of relations provides the necessary information. For every component-oriented Cluster describes a relation table of all possible component states of the cluster with the corresponding discrete values of the basic building blocks. The table can also be limited to selected combinations of component states. This table is continuously updated throughout the reduction process. At the end of the reduction process, the table shows the component states specified at the beginning with the discrete values of the header graph. Fιg.8 illustrates an extract from a table of relations.

With the help of the relation table, the associated component states can be determined relatively easily from a known graph using the header graph. If the simulation result is added after the last column of the relation table, the result is e.g. the classic error table with the component states and the corresponding system sizes when generating diagnostic knowledge. The size of the table can be limited to certain combinations.

An example of a diagnosis system for a complete vehicle system is to estimate the necessary simulations without and with the reduction method. For the sake of completeness, single and multiple errors are shown for the possible error assumptions. First, the error states are considered. Then the controls are discussed, since they play an important role in the error effect and diagnosis.

Assumptions: there are 26 control units in the vehicle.

Single I multiple error assumptions

Fault conditions of an entire vehicle 6370 2.1 10 892

Assumptions: At least every second simple error has the same effects and is therefore redundantly recognized and removed. Simple assumption of errors

Fault conditions of an entire vehicle 3185

Controls

Assumptions: The locality of the effects of a control is recognized by the hierarchy, so that on average only half of the possible controls per control device are coupled to another

In addition to the above reduction properties, the number of components is reduced due to the substitute value formation. Various experiments show:

Assumptions: At least 50% of all components are removed in the reduction process. A quadratic relationship between the number of components and the simulation time results in a reduction factor of 25%.

Assuming that a simulation step takes one second, the following values are set for the simulation times:

A comparison of the two numbers thus results in a reduction of the simulation effort with the method described above in the amount of 10 ⁴⁵ for an entire vehicle.

Claims

Patent claims

1.

Method for reducing the number of simulation steps required in a simulation method for the computer-controlled generation of knowledge data about an overall system consisting of several electrically controllable components with the steps

• Dismantling the electrically controllable components into components that include one or more basic modules

• Recording the electrical connections between the basic blocks and between the components

• Assigning discrete electrical status values to the basic components

• Defining the operating states and the possible component states of the individual components in relation to the electrical state values of the basic blocks belonging to a component

• Determination of the variables necessary for knowledge data generation and measurable from the basic building blocks and components

• Determination of basic building blocks that have no influence on the defined measurable variables and

• Combining and eliminating the identified uninfluential basic building blocks.

2.

Method according to claim 1, characterized in that before the step of determining the measurable variables, the knowledge data to be generated are divided into knowledge data types and the variables measurable on the basic modules and components are only determined for one or more knowledge data types.

3.

Method according to claim 1 or 2, characterized in that after the step of determining the measurable variables, those basic building blocks are determined and combined that influence the same spatially limited knowledge data area to form local basic clusters.

4.

Method according to claim 3, characterized in that component clusters are formed which consist of one or more basic clusters.

5.

Method according to one of claims 1 to 4, characterized in that the knowledge data is reduced to the knowledge data which relate to the component states considered to be essential by introducing empirical knowledge about the electrically controllable components or the overall system.

6.

Method according to one of claims 1 to 5, characterized in that the knowledge data is diagnostic knowledge.

7.

Method according to one of claims 1 to 6, characterized in that the components are resistors, switches, lines, lamps, line bundles,

Include connectors, relays, fuses, solenoid valves, motors, switches, ground nodes and power sources.

8th.

Method according to one of claims 1 to 7, characterized in that the basic components include resistors, current sources and voltage sources.

9.

Method according to one of claims 1 to 8, characterized in that the discrete electrical state values are resistance values,

Current values or voltage values.

10.

Method according to one of claims 1 to 9, characterized in that the operating states include the states open or closed, lamp on or off and relay contact active or passive.

11.

Method according to one of claims 6 to 10, characterized in that the component states include the states good, defective, closed and no contact, open and contact, interruption, short circuit and blocked rotor.

12.

Computer for carrying out a method for reducing the number of simulation steps required in a simulation method for the computer-controlled generation of knowledge data about an overall system consisting of several electrically controllable components, containing:

• Means for dismantling the electrically controllable components into components that include one or more basic modules

• Means for detecting the electrical connections between the basic building blocks and between the components

• Means for assigning discrete electrical status values to the basic components

• Means for defining the operating states and the possible component states of the individual components in relation to the electrical state values of the basic building blocks belonging to a component

• Means for determining the variables necessary for knowledge data generation and measurable from the basic building blocks and components

• Means for determining basic building blocks that have no influence on the defined measurable quantities and

• Means for combining and eliminating the identified uninfluential basic building blocks.

13.

Computer according to claim 12 additionally containing

Means for dividing the knowledge data into knowledge data types and for determining the quantities that can be measured on the basic building blocks and components only for one or more knowledge data types.

14.

Computer according to claim 12 or 13 additionally containing Means for determining and combining those basic building blocks that influence the same spatially limited knowledge data area to form local basic clusters.

15.

Computer according to claim 14 additionally containing means for forming

Component clusters that consist of one or more basic clusters.

16.

Computer according to one of claims 12 to 15 additionally containing means for introducing empirical knowledge about the electrically controllable components or the overall system and means for reducing the knowledge data to the knowledge data which relate to the component states considered essential.

17.

Computer according to one of claims 12 to 16, characterized in that the knowledge data is diagnostic knowledge.

18.

Computer according to one of claims 12 to 17, characterized in that the components are resistors, switches, cables, lamps, cable bundles,

19.

Computer according to one of claims 12 to 18, characterized in that the basic components include resistors, current sources and voltage sources.

20.

Computer according to one of claims 12 to 19, characterized in that the discrete electrical state values are resistance values,

Current values or voltage values.

21.

Computer according to one of claims 12 to 20, characterized in that the operating states include the states open or closed, lamp on or off and relay contact active or passive.

22.

Computer according to one of claims 17 to 21, characterized in that the component states include the states good, defective, closed and no contact, open and contact, interruption, short circuit and blocked rotor.