DE102008004219A1 - Error handling method for e.g. motor vehicle, involves testing components of system i.e. motor vehicle, for errors according to sequences determined by decision tree, where sum of costs for handling errors is kept to be minimum - Google Patents

Abstract

The method involves determining cost for handling errors and a probability of occurrence of errors. Components (8-14) of a system (4) are tested successively for the errors according to different sequences, which are automatically selected, and determined by a decision tree (2), where sum of the costs for handling of the errors weighted with probabilities of occurrence of the errors is kept to be minimum. A sum of operating time is determined for handling the errors. The units are successively tested in a selected partial system order (16) until the error in a fault unit is repaired. Independent claims are also included for the following: (1) a device for determining a sequence of steps for handling errors within a system (2) a computer program comprising instructions for handling errors within a system (3) a computer program product comprising instructions for handling errors within a system.

Description

The The invention relates to a method for treating at least one Error within a system, a device for determining a sequence of steps of a method for treating at least an error within a system, a computer program and a computer program product.

State of the art

at modern vehicles, the functionalities u. a. provided by software. The spectrum typically ranges from the engine control through to to the comfort system. A computer architecture used here is formed as a distributed system. Depending on the vehicle type can 20 up to 80 ECU nodes be provided. These control device nodes are partially connected to up to four different bus systems. A program code can be several hundred thousand to several millions Include lines. It is expected that this degree of crosslinking in the vehicle steadily increases in the coming years, is added the increasing complexity hydraulic, pneumatic and mechanical vehicle components and thus the increasing variety of vehicles.

Out For this reason, the troubleshooting and repair in the workshop difficult. When maintenance in the workshop is the considered Diagnostic strategy usually symptomatic nature, that is, that a starting point for the workshop diagnosis on a certain amount of symptoms of a Misbehavior is usually based come from three sources of diagnostic information. As a diagnostic information source become from the ECU diagnostics originating information (online diagnosis), information from physical Measured variables (offline diagnosis), z. As voltage, pressure, exhaust gases, etc., as well as information from subjective Observations of the workshop staff, z. B. based on noise or a visual inspection, used.

There the symptomatic diagnostic search strategies, for example, on one already existing amount of data and fault diagnosis based on Pattern recognition is done, the different, earlier errors occurred in represent the system, can be a currently observed error with the symptom pattern of the collected Database compared and thus quickly and easily identified become. The time required for A symptomatic troubleshooting is low, it is especially then suitable when resorting to known incidents However, she is aware of previous incident experiences and the ability to to retrieve the existing information depends.

nowadays There are several diagnostic tools that exist in the workshop assist with troubleshooting. This is u. a. a tester to query one in one control unit (ECU, electronic control unit) stored code number for storage malfunctions or a value in a fault memory (DTC, diagnostic trouble code) of a controller for a guided troubleshooting offered the workshop personnel inspection steps in the form of decision trees and troubleshooting instructions. To the decision trees and To generate troubleshooting instructions, use the authors of diagnostic tools various software tools such as authoring systems or heuristic and model-based expert systems. The delivered decision trees are based not just on the evaluation of associative knowledge that is in shape of heuristics, ie rules of experience, represented in a rule base but also on the analysis and evaluation of vehicle models as well as their systems and subsystems. A combination of these two approaches allows a quick and accurate diagnosis because of the model-based Diagnosis high efficiency and high through diagnostic rules Diagnostic performance can be achieved.

The decision trees and troubleshooting guides are generated manually or semi-automatically today and map troubleshooting paths directly. The workshop staff suggested procedure is however specified by the author, d. H. a troubleshooting guide is in order the test steps, the duration of troubleshooting and thus their costs are not fully optimized. The order of the test steps is also anchored, that is, that intermediate results of the diagnosis used for optimization could become, not considered become.

Optimized decision-making in dynamic processes usually requires consideration of several decision-making criteria. The person who is in such situations for a step in a complex process, eg. B. repair process, diagnostic process, manufacturing process, decide is often due to their lack of or limited knowledge and because of their subjective opinion not in the Able to take a decision that has to take into account all the criteria and experiences that are important for this situation. This can lead to redundancies and unnecessary process steps, which can result in a longer processing time and increased process costs.

on the other hand is it for the case that a sound knowledge and a wealth of experience exist are not possible in many situations, the ones to be considered Prioritize decision criteria in the right order, what may be due to different causes, eg. A missing one Overall view or lack of understanding for the decisive quality criteria a process.

Of another is missing in the individual decision points of a complex Process understanding or the foresight for the consequences of a wrong or not optimal decision that only become clear in subsequent process steps. This is to explain that the person who should make decisions is usually a specialist for one Part of the process is trained and therefore not all effects to estimate their trade can.

moreover Dynamic decision making processes are characterized in that the importance of the individual decision criteria of level change to level which makes decision-making even more difficult.

Disclosure of the invention

The The invention relates to a method for treating at least one Error within a system. It is provided that the System has a number of units. When executing the procedure for every Unit is the cost of handling an error and a probability of occurrence for the Mistake for the respective unit provided. It is intended that different Sequences after which the units are successively checked for errors, be determined. With the procedure that order becomes automatic selected, at the sum of the costs to treat the possible Error weighted with occurrence probabilities the error is minimal is.

In Embodiment here includes treating at least one error a check at least one unit of the system for faultiness, so that in this regard checked and thus it is determined whether the at least one unit is an error having. To handle the at least one error may continue be provided that this in a suitable manner, eg. By repair or replacement of at least one unit, fixed and thus eliminated removed from the system.

Consequently Is it possible, with the procedure at least one error within the system to investigate. Also for the case that for certain units the cost of treatment and thus to verify the error and the probability of occurrence of the respective error known these are nevertheless automatically provided as part of the procedure. If values for the cost and / or probability of occurrence of an error should not be known, these are under the procedure for example, by estimates determined and thus also provided automatically. moreover can the test costs are calculated be expected to comply with each case in an order are. In an embodiment can Such costs are calculated according to the rules of probability being, taking costs of all possible Errors or error events with the respective occurrence probabilities the error events are weighted and summed.

In A variant of the method may be an algorithm for use come, using this algorithm automatically an order for fixing and thus for testing the error or a test order selected is at the expected test costs and thus the costs for testing be minimal. A check or checking the Units of error typically involves searching for an error in the units according to the selected order with the aim of finding this error. Furthermore, it is envisaged that by means of the algorithm takes into account that order is at the one to check and, if necessary, to correct as part of the treatment of a respective error to be operated effort is minimal. In addition, with the algorithm occurrence probabilities taken into account by errors so that as a rule priority units are checked, the most vulnerable for mistakes are.

In a further embodiment of the method, it is provided that the sequence is determined by a decision tree branched into decision options. It is envisaged that the system to be inspected for faults has a number of subsystems as units. Subsystems of this kind can, as further units, in turn comprise subsystems and / or components as further inputs include.

Of the Decision tree, which in an embodiment for carrying out the The method used may further include a structure of the system represent. This represents one node each of the decision tree as a subsystem and / or Component formed unit. About branches within the Decision tree for a respective system regularly becomes a node, for example for a subsystem, subdivided into subordinate subsystems and / or components. Through a respective subdivided or branched node as part of the procedure, the decision options for setting the order determined. Thus, an order to be selected becomes typically through a path within the decision tree represents.

at a procedure of the method is to be selected within a Order after a step a number of further steps provided for selection. In this case, that further step is selected regularly, the weighted from the selection of further steps taking into account costs with the probability of occurrence a minimum value of expected costs. If the procedure using the decision tree carried out will be considered further steps as decision options, which are executed starting from a branched node can.

In addition, in the Framework of the procedure for determining the total processing times be considered to correct individual errors.

The The method may include in detail a plurality of process steps, which are generally to be carried out successively.

at In a first such method step, the order with calculated at the minimum expected cost. Here are several possible Automatically determines sequences, taking into account those aspects, under which the process is to be optimized, d. H. Costs, occurrence probabilities etc., will then be the cheapest Order selected.

In In a second method step, the units in the selected or sequence found consecutively tested or checked until the at least one error is discovered. This can mean that the trained as subsystems and / or components units in the order or test sequence be tested one after the other. If a subsystem is to be checked as a unit, is regular Function to check. Therefore is first not provided, the amount of components of the subsystem in detail to check. Further Steps to be performed in the order can be one detailed examination components of the subsystem, if that subsystem identified as defective.

In a third method step is provided, the error in a fix unit identified as faulty. This can mean that this unit, for example, a component, repaired or is exchanged. If a faulty unit is a faulty unit Subsystem is identified, this subsystem is in detail examined. This may mean that individual child subsystems and / or components of this subsystem are checked, taking into account the selected one Sequence and / or the algorithm may possibly recursively.

In a fourth method step becomes a function of the former faultily identified unit tested, thus checking that the error within this unit has been corrected. It can the Function of a trained as a subunit unit possibly also on a parent Level to be tested.

The In a refinement, the method is ended by a fifth method step. This fifth process step According to that, it is envisaged that the examination of the system will be completed becomes if the at least one and thus all errors are solved or are. If in the system yet another not yet identified Error should be present, the procedure will be at the second Process step continued, so that as another unit another not yet tested Subsystem and / or another not yet tested components checked for defectiveness. In such a continuation of the method are therefore units considered, which has not yet been reviewed or have been repaired.

If the process detects and corrects an error in a device, it may mean that the entire system is healthy. Thus, it makes sense to search for other errors that may possibly be present within the system, first higher-level units in a higher Ebe ne within the decision tree, for example the whole system as the highest superordinate unit. Because a failure in a device can typically affect higher-level units, the process can be shortened and, if necessary, terminated by checking at least one higher-level unit after a fault has been remedied in a subordinate unit.

With The method may be in an embodiment as a technical device trained system to be repaired. The system includes as at least a unit at least one subsystem, wherein the at least one Subsystem at least one component and / or optionally at least one further subsystem may comprise at least one unit.

The Invention also relates an apparatus for determining an order of steps of a Procedure for checking at least an error within a system, this system being a Number of units. This device is designed to for every Unit the cost of testing one Error as well as an occurrence probability for the error provide. Furthermore, the device is designed to different orders, after which the units are checked one after the other, and automatically select that order which weights the sum of costs to handle all possible errors Occurrence probabilities the error is minimal.

One within the decision tree based on the selected order As a rule, the suggested optimal decision path leads to one as possible short processing time and lowest possible Costs for fixing the at least one error.

For decision-making, the basis of the selected Order can be made in a further embodiment, three parameters are considered, namely effects and symptoms of one for one respective unit possibly occurring error, at least one error statistics of known and / or similar fault conditions and the costs associated with finding and / or correcting the problem can.

at the treatment, d. H. exam and / or rectification, error or treatment of a technical problem Problems will be considered the selected one Order and thus a decision path within the decision tree a strategy of treatment provided to nodes or Branches within the decision tree is oriented. At Each branch or node may have a number of options to choose from. If now a node within the decision tree as a subsystem represents a trained unit, represent the other options available are usually the components this subsystem. To determine the order is therefore provided select in which order the individual components of the subsystem be checked one after the other. This order is chosen such that a minimal effort it is to be expected what u. a. again by optimal temporal and / or financial aspects is affected.

The The device described is adapted to all steps of the presented Procedure to perform. It can individual steps of this procedure also of individual modules of the Device performed become. Furthermore you can Functions of the device or functions of individual modules the device can be implemented as steps of the method. The Device is in an embodiment as a so-called. Diagnostic tool, for example for a text of a trained as a vehicle system, in a workshop educated.

The Invention further relates to a computer program with program code means, to perform all the steps of a described method, if the computer program on a computer or equivalent Arithmetic unit, in particular in a presented device, accomplished becomes.

The Computer program product according to the invention with program code means stored on a computer readable medium is to be carried out all steps of a method described, if the computer program on a computer or equivalent Computing unit, in particular in a device according to the invention, accomplished becomes.

As part of the method algorithms for hybrid trees or decision trees, ie provided with components and / or subsystems in a plane. The costs or test costs used include optimization criteria for troubleshooting and / or troubleshooting, and the test costs can now be optimized. This can include labor costs of the workshop, actual personnel costs, material costs, test time, etc., but also virtual test costs, which are weighted by linear or non-linear weighted Combinations are formed.

Such costs can, for example, by the formula K = a · (personnel costs + material costs) + b · inspection time, be represented, in which case the proportion "b · test time" can represent a kind of "penalty" for waiting times, which means that waiting times can also be minimized. In a so-called. "Money does not matter" mode for urgent customers would be z. B. a = 0 and b = 1 to set. Within the scope of the method, the still-to-be-expected duration of the search process for finding the error can be continuously displayed, for example, via a display module of the device.

The Calculating the expected cost may also be different in other ways consideration other parameters and their weighting, but to comparable Results.

possibly missing statistical information for components and subsystems from static data, for example, failure rates, etc., are comparable by or parent Subsystems of the system approximated and therefore automatically provided.

decision steps in a complex decision-making process can also take the form of a decision tree to be provided. Dynamic decision making changes the shape of the decision tree depending on various environment variables. An optimal decision path in a decision tree leads to the Rule to as possible short processing times and lowest possible Process costs.

The inventive method is in the following in a possible Variant using decision trees for error identification and troubleshooting.

For decision-making typically three essential parameters are considered in this variant, namely Error effects and symptoms, at least one error statistic of known and / or similar fault conditions as well as costs incurred during fault finding or troubleshooting.

For others Fields of application, d. H. when using decision trees for others Purposes, can additional parameters or criteria are used. The presented Accordingly, the method serves, for example, the decision-making in decision trees.

at the variant of the method are in particular by the computer program weighted decision trees, z. For example Diagnostic equipment, used. It is u. a. the adjustable weight, z. B. by Probabilities or costs used. The database of Decision tree is using the paths selected by the computer program combined.

Decision trees can be used in many ways become. One application is z. B. Troubleshooting trees used in diagnostic testers are implemented and with the help of which workshop personnel error causes, eg. defective components, searches. about a decision tree will systematically all potential Error sources checked. In an embodiment of the method troubleshooting trees are referenced. The procedure is however for any kind of decision trees, eg in medical diagnostics, customer surveys etc., applicable.

In A variant is the automated interactive processing of a Decision tree executed. At a decision node in a decision tree is with the device and / or the method a user of several Decision options that are designed as branches or nodes and / or are shown, the cheapest selected and typically suggested. If a branch is checked as negative successively reviewed alternative branches. Becomes a branch tested positive, this branch is followed up, and the user of the method arrives at the next decision node.

• 1: "Sichtprüfung der Birne"
• 2: "Prüfe Lampenspannung"

zur Überprüfung vorschlagen. Nach der Rückmeldung des Ergebnisses an das System, z. B.

• 1: "Birne nicht defekt"
• 2: "Lampenspannung = 0"

verfolgt das Verfahren in einem nachfolgenden zweiten Zweig den Fehler weiter, z. B. durch die Schritte

• 2-1: "Prüfe Lampensicherung"
• 2-2: "Prüfe Lampenzuleitung"
• 2-3: "Prüfe Bordnetzspannung".

Hereinafter, a possible example for implementing the method will be described. For example, when searching for the cause of a headlamp malfunction, the system may provide the user with the following decision options:

• 1: "Visual inspection of the pear"
• 2: "Check lamp voltage"

to suggest for review. After the feedback of the result to the system, z. B.

• 1: "Pear not defective"
• 2: "lamp voltage = 0"

the method continues to track the error in a subsequent second branch, e.g. B. through the steps

• 2-1: "Check lamp fuse"
• 2-2: "Check lamp lead"
• 2-3: "Check vehicle electrical system voltage".

results the review z. B. too low board voltage, so will after the same Principle continued in this branch until the cause of the error, z. B. a defective battery was found.

This Example shows that the order of execution of the branches the price for the troubleshooting determined. Are z. B. first the costly but unlikely options tested, troubleshooting expensive. In the example, the most cost-effective diagnosis results when on First try option 2 and then option 2-3.

The inventive method includes u. a. an optimal evaluation of the decision options that the total cost of error handling and thus troubleshooting minimized.

A starting point for providing the sequence based on the decision nodes of the decision tree is the fact that at least one of the units of the system designed as component C ₁ ,..., C _N is defective. The N components form here as a subsystem formed unit. This is a group of components whose function can be checked in one step during a subsystem check.

P_κ

die Wahrscheinlichkeit, dass die Komponente k defekt ist, dies entspricht hier der absoluten Ausfallrate

P_T

die Wahrscheinlichkeit, dass ein Defekt im betrachteten Teilsystem vorliegt; diese kann aus den P_κ-Werten errechnet werden.

K_κ

ein Maß für die Kosten, die bei der Überprüfung der Komponente κ anfallen. Diese Kosten werden hier als Prüfkosten bezeichnet.

K_T

ein Maß für die Prüfkosten, die bei der Funktionsüberprüfung des betrachteten Teilsystems anfallen, d. h. für die Überprüfung der Funktion eines Teilsystems typischerweise, nicht für die Überprüfung seiner Komponenten. Im obigen Beispiel war dies Schritt 2: "Prüfe Lampenspannung". Demnach wird in diesem Beispiel die Funktion des diagnoserelevanten Teilsystems geprüft, das hier Lampensicherung, Lampenzuleitung und Bordnetz umfasst.

Now it has to be decided in which order the components will be tested. The following information is regularly available:

P _κ

the probability that the component k is defective, this corresponds here to the absolute failure rate

P _T

the probability that a defect exists in the considered subsystem; this can be calculated from the P _k values.

K _κ

a measure of the costs incurred in checking component κ. These costs are referred to here as test costs.

K _T

a measure of the cost of testing that occurs during the functional review of the subsystem under consideration, ie typically for the verification of the function of a subsystem, not for the verification of its components. In the example above, this was Step 2: "Check Lamp Voltage". Accordingly, the function of the diagnostics-relevant subsystem is examined in this example, which includes lamp protection, lamp supply and wiring system here.

Weterhin the following 2 ^N - 1 events are defined:

den Fall, dass die Komponenten 1 und 3 defekt, und die Komponenten 2 und 4 in Ordnung sind. Die Menge aller 2^N – 1 Ereignisse beschreibt das Ereignis, dass mindestens ein Fehler im betrachteten Teilsystem aufgetreten ist.This z. For example, if N = 4, the event

the case that the components 1 and 3 are defective, and the components 2 and 4 are in order. The amount of all 2 ^N - 1 events describes the event that at least one error has occurred in the considered subsystem.

Furthermore, in the following
P (E ...)
the conditional probability for the occurrence of a special event
E ...
describe, on the condition that an error has occurred, ie that at least one designed as a component and / or subunit unit is defective. The values of these probabilities can be calculated according to the rules of probability theory and estimated on the basis of real field data. It can be shown that their sum is 1.

In the following, the sequence 1, 2, 3,..., N provided here especially for the treatment and testing is considered, ie the components are checked in the order 1, 2, 3,. replaced, until there is no more error in the system, it would be conceivable N! - 2 more test sequences. It now describes
K _{1-2-3 -...- N} (E ...)
the total cost of testing for testing the components in the underlying test order 1, 2, 3, ..., N at the occurrence of the specific fault event
E ...
would arise.

werden die Einheiten bzw. Komponenten in der Reihenfolge 1, 2, 3, 4 so lange geprüft, bis kein Fehler mehr im betrachteten System vorliegt. Bei dem betrachteten Fehlerbild ist dies der Fall, nachdem die defekte Komponente 3 getestet wurde, weil danach kein Fehler mehr im System vorliegt. Neben den Kosten für das Überprüfen der Komponenten fallen noch an zwei Stellen Kosten für die Überprüfung des Gesamtsystems an, nämlich nach Austausch oder Reparatur von der ersten Komponente 1 mit dem Ergebnis, dass immer noch Fehler im betrachteten Teilsystem vorhanden ist und der dritten Komponente 3 mit dem Ergebnis, dass das betrachtete Teilsystem fehlerfrei ist.The example for the case N = 4 with the event

the units or components are checked in the sequence 1, 2, 3, 4 until there is no longer an error in the system under consideration. This is the case with the considered error picture after the defective component 3 has been tested, because afterwards there is no more error in the system. In addition to the cost of checking the components are still in two places costs for the review of the entire system, namely after replacement or repair of the first component 1 with the result that there are still errors in the considered subsystem and the third component 3 with the result that the considered subsystem is error-free.

This results in the total testing costs for the event under consideration in conjunction with the test sequence 1, 2, 3, 4

By the way, all test sequences starting with 1-3 -... were optimal for the considered event, because they would lead to minimal test costs K ₁ + K ₃ + 2K _T.

Before the test and when determining the order for the test, the occurrence of the error event is regularly unknown. The algorithm therefore chooses an order in which the expected testing costs become minimal. In the following, therefore, the test costs are calculated, which are to be expected in compliance with the special test sequence 1, 2, 3, ..., N. They are calculated according to the rules of probability by weighting and summing the costs of all possible error events with the occurrence probabilities of the error events. For the specific test sequence 1, 2, 3,..., N, formula (1) is obtained:

The proposed algorithm now optimizes the sequences or test sequences by this the costs according to formula (1) for all N! - 1 possible test sequences calculated. Thereafter, the most favorable order is selected and the user subsequently performs troubleshooting, the system resorting to the order with the least expected cost.

• 1. Prüfreihenfolge mit minimalen erwarteten Prüfkosten errechnen.
• 2. Die Komponenten in der gefundenen Prüfreihenfolge testen, bis ein Fehler gefunden wird.
• 3. In der fehlerbehafteten Komponente den Fehler beheben oder Komponente austauschen.
• 4. Die Funktion des Teilsystems testen.
• 5. Falls das Teilsystem fehlerfrei ist, ist die Prüfung abgeschlossen.

A determination of the probabilities P and the costs K takes place in each case taking into account the events that can occur when following an order during an examination. Thus, the probability P and the cost K are functions of the variables defined as events and thus dependent on them. An implementation of the algorithm would therefore be implemented by the following method steps:

• 1. Calculate test sequence with minimum expected test costs.
• 2. Test the components in the test order found until an error is found.
• 3. Correct the error in the faulty component or replace the component.
• 4. Test the function of the subsystem.
• 5. If the subsystem is error free, the test is complete.

If In the subsystem there are still errors second process step with the next component, d. H. that after just repaired, continued.

In usually neither probabilities nor costs are fully known for all components be. In this case, estimates to use, the z. B. derived from statistical analysis and accordingly can be provided.

A further embodiment of the invention provides in a second case alternatively or in addition a test order for branches, the other levels, d. H. Subsystems as units, contain, in front. Checked Then not only components but also subsystems, for example. higher Two subordinates.

In this second case the algorithm is similar Form, as in the first case 1. In place of the cost of checking the Move components only the cost of the verification of the functioning the considered subsystems and their failure probabilities. Then proceed as described above. Once a mistake within a subsystem is detected, the algorithm jumps into this Subsystem and continues its troubleshooting as described in the first case Proceeding. The test sequence for the components contained therein will again be based on the principle according to the formula (1) optimized.

• 1. Prüfreihenfolge mit minimalen erwarteten Prüfkosten errechnen.
• 2. Die erste Komponente oder das erste Teilsystem in der gefundenen Prüfreihenfolge testen, bis ein Fehler gefunden wurde. Im Falle eines Teilsystems wird nur dessen Funktion, jedoch nicht die Menge seiner Komponenten, geprüft.
• 3. In einer fehlerbehafteten Komponente wird der Fehler behoben oder die Komponente ausgetauscht. Falls der Fehler in einem Teilsystem vorliegt, wird in dieses Teilsystem gesprungen und weitergeprüft, was mit diesem Algorithmus, ggf. rekursiv, erfolgen kann.
• 4. Dann wird die Funktion des Systems auf einer übergeordneter Ebene getestet.
• 5. Falls das System fehlerfrei ist, ist die Prüfung abgeschlossen.

The implementation of the algorithm in a device designed as a tester, for example, comprises the following method steps:

• 1. Calculate test sequence with minimum expected test costs.
• 2. Test the first component or subsystem in the test order found until an error is found. In the case of a subsystem, only its function, but not the quantity of its components, is checked.
• 3. In a faulty component, the error is corrected or the component is replaced. If the error is present in a subsystem, it jumps into this subsystem and continues to check what can be done with this algorithm, if necessary recursively.
• 4. Then the function of the system is tested at a higher level.
• 5. If the system is error free, the test is complete.

If the system should still have errors, the procedure will in the second step with the next component or next Subsystem, d. H. that after the component just repaired, continued.

This Algorithm is recursive, d. H. if in lower levels again as subsystems trained units is also here proceed as described. This will be down to the lowest level the only components trained units, and therefore none Subsystems are present, carried out.

at the determination of costs must take into account those costs the at the exam of a component or a subsystem. In the last second Case is the cost of verifying the operation of a subsystem but not to review his Components.

The calculations to be carried out with the formula (1) shown usually lead to the optimal solution while minimizing the costs to be expected. Suboptimal calculations can easily be deduced by z. B. is assumed that no multiple errors occur. This simplifies the calculation considerably. This is typically useful when testing reliable systems where multiple errors are very unlikely.

Further Advantages and embodiments of the invention will become apparent from the Description and attached drawing.

It it is understood that the above and the following yet to be explained Features not only in the specified combination, but also usable in other combinations or alone are without departing from the scope of the present invention.

Brief description of the drawings

1 shows a schematic representation of a first example of a decision tree used in an embodiment of a method according to the invention.

2 shows a schematic representation of a second example of a decision tree used in the context of an embodiment of a method according to the invention.

3 shows a schematic representation of an embodiment of a device according to the invention when carrying out a variant of the method according to the invention.

Embodiments of the invention

The Invention is based on embodiments schematically shown in the drawings and will be in the following described in detail with reference to the drawings.

The in 1 shown decision tree 2 serves to illustrate an algorithm which is suitable for carrying out an embodiment of the method according to the invention. Furthermore, through this decision tree 2 schematically part of a system 4 shown, this system 4 in a first level one through a decision node 6 represented unit, which is designed here as a subsystem having. Starting from the decision node 6 includes the decision tree 2 as a branch of the decision node 6 in a second level several as components 8th . 10 . 12 . 14 trained units.

In the context of the embodiment of the method described here, a selected sequence is checked for checking the units upon reaching the decision node 6 the subsystem is checked. In this case, this subsystem is assigned a probability P _T that there is an error in this subsystem considered. Furthermore, the subsystem is also assigned a value K _T as a measure of the costs incurred in the testing of the subsystem.

If it is determined during the test that the subsystem is in order, the procedure will proceed from the decision node 6 finished and the subsystem as "okay" 16 classified.

If the subsystem is not "OK" 16 should be, there is a review of the components 8th . 10 . 12 . 14 that are assigned to the subsystem. In accordance with a selected order, in the context of the method, the units to be checked first become a first component 8th C ₁ , then a second component 10 C ₂ , a K-th component 12 C _K and finally an Nth component 14 C _N checked. All these components 8th . 10 . 12 . 14 are probabilities of having an error and costs associated with verifying a particular component. For the first unit 8th C ₁ is the probability P ₁ and the cost K ₁ , for the second component 10 C ₂ is the probability P ₂ and the cost K ₂ , for the Kth component 12 C _K is the probability P _K and the cost K _K and for the Nth component 14 C _N is the probability P _N and the cost K _N.

The order according to which the units and thus in this case the components 8th . 10 . 12 . 14 be checked in succession, is automatically selected such that the sum of the costs K ₁ , K ₂ , K _K , K _N for checking all possible errors weighted with occurrence probabilities P ₁ , P ₂ , P _K , P _N, the error of the units and thus of the components 8th . 10 . 12 . 14 is minimal.

The in 2 schematically represented decision tree 20 who here in a second example egg ner embodiment of the method according to the invention is used, is also schematically a system 21 within which a mistake is to be treated.

In a first level 22 includes the decision tree shown here 20 a first decision node 24 , the unity of the system formed as a superordinate subsystem 21 represents. In a second level of this decision tree 20 three are the parent decision node 24 associated child decision nodes 26 . 28 . 30 intended. These child decision nodes 26 . 28 . 30 simultaneously represent units of the system formed as subordinate subsystems 21 that are subordinate to the parent subsystem. In a third level 32 includes the decision tree 20 several components 34 . 36 . 38 . 40 . 42 . 44 . 46 . 48 . 50 . 52 as the child decision node 26 . 28 . 30 and thus also the lower subsystems associated units of the system 21 ,

The procedure takes into account that the parent subsystem that is in the decision tree 20 through the parent decision node 24 is represented, an occurrence probability for an error P _SK and costs for the treatment of the error K _{SK are} assigned. Your first child subsystem, here through the first child decision node 26 are represented, a probability P _T1 for the occurrence of an error and costs K _{T1 associated} with the treatment of the error. A second child subsystem, here through a second child decision node 28 , a probability P _T2 for an error and a cost K _T2 for handling the error are assigned. An Nth subsystem, here by an Nth subordinate decision node 30 are represented, a probability P _TN for an occurrence of an error and costs K _{TN associated} with the elimination of the error.

The first child decision node 26 and thus the first subordinate subsystem are components as entities 34 . 36 . 38 C ₁₁ , C _1J and C _1L assigned. For a first component 34 C ₁₁ , a value for a probability P _{11 is provided} for an occurrence of an error and a cost K ₁₁ for the treatment of the error. For a Jth component 36 C _1J , a probability P _{1J is provided} for occurrence of an error and cost K _1J for correcting the error. For an Lth component 38 C _L are expected to cost K _1L to treat the error and a probability P _1L for the occurrence of the error.

The second child decision node 28 and thus the second subordinate subsystem of the system 21 are as units a first component 40 C ₂₁ , a second component 42 C ₂₂ and an Nth component 44 Assigned to C _2N . For this first component 40 C ₂₁ is the cost K ₂₁ , for the second component 42 C ₂₂ the cost K ₂₂ and the Nth component 44 C _2N the cost K _2N for each treatment and thus search and remedy a mistake provided. For the possible errors within these three components 40 . 42 . 44 C ₂₁ , C ₂₂ , C _2N , the occurrence probabilities P ₂₁ , P ₂₂ , P _2N are respectively estimated.

The Nth decision node 30 representing the Nth subordinate subsystem are as units a first component 40 C ₃₁ with cost K ₃₁ and an occurrence probability P ₃₁ for an error within this first component 46 C ₃₁ provided. For a second component 48 C ₃₂ , which as a unit the third subordinate subsystem 30 is assigned, the costs K ₃₂ for handling a fault and the probability of occurrence P ₃₂ for this fault are to be expected. For a Kth component 50 C _3K , the occurrence probability P _{3K is provided} for an error. To correct this error of the K-th component 50 C _3K are the cost K _3K expected. At the Nth component 52 C _3N , which is assigned as a unit to the third subordinate subsystem, an occurrence probability P _{3N is} to be expected for an error. The expected costs of treating this error are K _3N .

In the case of 2 presented example for treating at least one error within the system 21 be different orders after which the units described within the system 21 successively checked for errors determined. Here, for each unit, ie for the higher-level subsystem, the subordinate subsystems and the components assigned to the subordinate subsystems, the sum of the above-mentioned costs for the treatment of the at least one possible error weighted with the above-enumerated occurrence probabilities, the error is taken into account. Among these sequences determined in this embodiment of the method, the order is selected according to which the sum of the costs for handling the at least one error weighted with the occurrence probabilities for this at least one error is minimal.

It is provided in the present example that within the system 21 the Nth component 44 C _2N , the second subsystem 28 , represented by the second subordinate decision node 28 , has an error.

According to this selected order, first the first parent subsystem that is within the decision tree 21 through the parent decision node 24 is represented, treated. Since a check reveals that the superordinated subsystem is faulty, it is provided according to the selected sequence that the subordinate subsystems associated with the superordinated subsystem are now checked in such a way that first the first subordinate subsystem, represented by the first subordinate decision node 26 then the second subordinate subsystem represented by the second subordinate decision node 28 and finally the Nth subsystem, represented by the Nth subordinate decision node 30 , treated and checked for errors. The result is that the first subordinate subsystem is identified as error-free in the course of the treatment and thus as "in order". 54 is classified as the three components 34 . 36 . 38 C ₁₁ , C _1J , C _{1L are} also faultless. Due to the fact that the first subordinate subsystem is error-free, another check is the components assigned here 34 . 36 . 38 C ₁₁ , C _1J , C _1L not required.

In accordance with the selected sequence, the second subordinate subsystem, which here by the second subordinate decision node, will subsequently be used 28 is represented, checked. It follows that this second subordinate subsystem is faulty. As a result, it is necessary to have the components associated with this second subordinate subsystem as units 40 . 42 . 44 C ₂₁ , C ₂₂ , C _2N , according to the selected order one after the other, checking for errors and correcting any errors. In the present example, the first two components become 40 . 42 C ₂₁ , C ₂₂ identified as faultless. Only in the last, Nth component 44 C _2N an error is detected and fixed. After rectifying this error, the second subordinate component is also error-free and is thus considered "okay" 56 classified.

After correcting the error, it is possible in the design of the method, within the decision tree 20 starting from the second child decision node 28 in the second level 24 take a step up to the first level 22 to the parent decision node 24 to go and now the functionality or accuracy of the parent subsystem, here by the parent decision node 24 is represented, check. If this parent subsystem is identified as healthy, it means that the entire system 21 error-free and thus a further review of other subsystems is not required. Another review of the third subordinate subsystem, here by the third subordinate decision node 30 represented and as a rule can only be carried out theoretically, it will be shown that the third subordinate subsystem is error-free and thus "in order" 58 is to be classified.

3 shows an embodiment of a device according to the invention 60 in carrying out a variant of the method according to the invention. The device shown here 60 is for treating and thus eliminating at least one error within a system also shown schematically 62 , which is designed here as a motor vehicle and thus as a technical device suitable.

This system 62 includes as units a subsystem 64 as well as other units of this subsystem 64 child components 66 . 68 , To treat the at least one error, the device 60 with the system 62 connected. As part of the process are by the device 60 for each of the units and thus for the subsystem 64 as well as for the components 66 . 68 Costs for handling an error as well as an occurrence probability for the error within the units provided. To treat the at least one error are from the device 60 Furthermore, different orders, after which the units are checked one after another, are determined, and the order in which a sum of the costs for handling all possible errors weighted with occurrence probabilities of the errors is minimal is automatically selected.

The device 60 includes in the presently described embodiment, a computing unit 70 , with which the different orders are determined and the most favorable order with the minimum sum is selected. Furthermore, the device 60 designed to handle the at least one fault within the system 62 at least semi-automatically perform and here by the presence of a compound 72 to the system 62 individual units of the system 62 at least semi-automatic, usually automatic, to test and thus check.

Claims (10)

Method for dealing with at least one error within a system ( 4 . 21 . 62 ) comprising a number of units, each unit being provided with the cost of dealing with an error and a probability of occurrence of the error, by which various orders, after which the units are successively checked for errors, are determined and in which Order is selected automatically, in which the sum of the costs to treat the error weighted with occurrence probabilities of the error is minimal. Method according to Claim 1, in which an order is in each case passed through a decision tree ( 2 . 20 ), which branches into decision options. Method according to claim 1 or 2, wherein inside one to be selected Order after a step a number of further steps is available for selection, with that further step being selected, taking into account the selection of further steps of costs weighted with the probability of occurrence has minimal value. Method according to one of the preceding claims, in to determine the total processing times for the treatment of individual Error considered become. Method according to one of the preceding claims, in the - the Order is calculated with the minimum expected cost, in which - the Units in the selected Order one by one until the at least one Error is discovered at the - the error in a faulty one Unit is fixed at which - a function of this unit is tested, and in which - the exam is completed when all bugs are fixed. Method according to one of the preceding claims, with which a system designed as a technical device ( 4 . 21 . 62 ) is treated. Method according to one of the preceding claims, in which the system ( 4 . 21 . 62 ) comprises as at least one unit at least one subsystem, wherein the at least one subsystem as at least one unit at least one component ( 8th . 10 . 12 . 14 . 34 . 36 . 38 . 40 . 42 . 44 . 46 . 48 . 50 . 52 . 66 . 68 ). Apparatus for determining a sequence of steps of a method for dealing with at least one fault within a system ( 4 . 21 . 62 ) having a number of units in which the device ( 60 ) is arranged to provide, for each unit, the cost of handling an error and an occurrence probability of the error, determining various orders of sequentially checking the units for errors, and automatically selecting that order in which the sum of the costs to treat the errors weighted with occurrence probabilities the error is minimal. Computer program with program code means for carrying out all the steps of a method according to one of Claims 1 to 7, when the computer program is stored on a computer or a corresponding computer unit ( 70 ), in particular in a device ( 60 ) according to claim 8. A computer program product comprising program code means stored on a computer-readable medium for carrying out all the steps of a method according to one of claims 1 to 7, when the computer program is stored on a computer or a corresponding processing unit ( 70 ), in particular in a device ( 60 ) according to claim 8.