EP2810255A1

EP2810255A1 - Method for forecasting a fault or for fault detection in a transport machine, and transport machine

Info

Publication number: EP2810255A1
Application number: EP13713819.4A
Authority: EP
Inventors: Markus SCHÄTTIN; Michael Armbruster; Thomas Schmid
Original assignee: Siemens AG
Current assignee: Siemens AG
Priority date: 2012-04-25
Filing date: 2013-03-25
Publication date: 2014-12-10
Also published as: WO2013160039A1; DE102012206836A1

Abstract

What is explained is a method for forecasting a fault in a transport machine, involving: detection (2) of at least one type of detection values in the time domain in a first transport machine, storage (2) of the detection values or storage of frequency spectra for the detection values, waiting for a fault (4), preferably one that can be detected without the stored detection values or the stored frequency spectra, examination (6) of the frequency spectra stored prior to the occurrence of the fault, or of frequency spectra calculated from the detection values stored prior to the occurrence of the fault, for at least one feature that is an early indication of the occurrence of the fault, use of the feature for early detection of the fault or for fault detection in the first transport machine or in a second transport machine.

Description

description

Method for forecasting an error or for detecting errors in a transport machine and transport machine

The invention relates to a method for detecting errors, in particular to a method for the early prognosis of an error. In addition, an associated transport machine should be indicated, e.g. a vehicle, an airplane or a ship.

The invention relates to a method for forecasting an error in a transport machine, comprising:

Detecting at least one type of detection values in the time domain in a first transport machine,

Storing the detection values or storing frequency spectra of the detection values,

Waiting for an error, preferably for an error detectable without the stored detection values or the stored frequency spectra,

Examining the frequency spectra stored prior to the occurrence of the fault or frequency spectrums calculated from the detection values stored prior to the occurrence of the fault, for at least one characteristic indicative of the occurrence of the fault,

Use of the feature for early detection of the error or for error detection in the first transport machine or in a second transport machine. Furthermore, the invention relates to a transport machine, with:

- a vehicle electrical system containing at least two parts, a separating device arranged between the at least two parts, with which the two parts can be electrically separated from each other,

a first detection device on the vehicle electrical system, preferably on a first part of the at least two parts, - A control unit, which is connected to the separating device or with a warning unit, and which controls the separating device or the warning unit depending on the detected values detected by the detection device.

It is an object of the exemplary embodiments to provide a method for forecasting an error in a transport machine, wherein the method should be particularly easy to implement. In addition, a transport machine is to be specified, which is particularly suitable for the application of methods with error prognosis.

This object is achieved by a method according to claim 1 and by a transport machine according to the independent device claim. Further developments are specified in the subclaims.

One embodiment relates to a method for forecasting an error in a haulage machine, comprising:

Wait for a fault, preferably without the stored acquisition values or the stored frequency spectra, to be detected,

- Examine the frequency spectra stored before the occurrence of the error or frequency spectrums calculated from the detection values stored before the occurrence of the error on at least one characteristic which is the

Occurrence of the error indicates early

Use of the feature for early detection of the error or for error detection in the first transport machine or in a second transport machine.

The difficulty in forecasting errors consists, for example, in determining characteristic features which occur field of error, for example, would lead to the triggering of other safety devices. At a minimum, there should be a high probability between the occurrence of the feature and the occurrence of the error. If the probability is 100 percent, then there is a 1: 1 relationship or a deterministic or direct relationship.

The characteristic feature occurs, for example, only after several years of operation. If it then succeeds in associating an error with a characteristic feature from data that has been stored for a long time, this characteristic feature can be used in the vehicle repaired after the fault and, above all, in other vehicles of the same type for the early detection or prognosis of errors. This is possible by changing the operating software in a simple way, for example. In a workshop or via a wireless network.

The same applies to electromagnetic radiation in the electrical system in case of failure or imminent errors of connected to the electrical system modules or assemblies that are located close to the electrical system.

The determination of a characteristic feature or a characteristic feature group can not be over-estimated considering the general cost pressure, which is often associated with quality defects. Instead of global recall actions, errors can be selectively predicted and only the vehicles actually affected must be gradually called to the workshops.

The detection values can be stored. Alternatively, however, frequency spectra of the detection values are stored in order to reduce the amount of data, for example by a factor smaller than 10. Older data can also be erased or canceled out. For example. Deletes data older than one year. In the case of multiple storage per month, after the expiration of a period of time which, for example, is one month and one year, for out-of-period months only one record is stored for each month. Thus, data that has already been acquired and stored before the occurrence of a major error is used to determine the characteristic feature for the early detection of the error or for error detection. For example, data older than one day, older than one week, older than one month, or even older than one year may be used relative to the day the larger error occurred.

The detection values can be detected at a vehicle electrical system of the first transport machine. In on-board systems, early detection is particularly important because, for example, the power and voltage supply for important units must be secured. Faults in on-board networks can quickly lead to particularly serious errors, e.g. to fires. Even if conventional security measures were taken, such as fuses or automatic circuit breakers, another measure is preferable because the classical safety measures only trigger currents or voltages that are possibly at least an order of magnitude higher than in fault-free operation, which can lead to accompanying damage.

In addition, faults in on-board networks are particularly difficult to locate, let alone predict, so that the proposed method opens up new application areas.

In particular, in the on-board networks of electric cars, the discharge of a battery can be prevented by the early detection of errors. This can increase the acceptance of electric cars considerably, because they rely on a charged battery for driving. The method can be carried out for a large number of vehicles, for example also for all vehicles of a manufacturer. The database can be chosen as large as possible, since it is initially unknown in which vehicle the errors will occur. Alternatively, a relevant sample is used. If characteristics for early detection or error detection have then been determined on the basis of the random sample, these features can also be used in predicted vehicles for prognosis, in particular in vehicles that did not belong to the random sample.

Likewise, a wide frequency range can be detected in the spectrum, for example a frequency range which records more than 4 or more than 5 orders of magnitude. Thus, a frequency range can be used, for example from 0 Hz to 10 MHz (megahertz) or even up to 100 MHz. The largest frequency may typically be less than 1 GHZ. But even 50% or 80% of these ranges can be sufficient, especially if the percentages contain the upper frequency range. Again, it is not known or difficult to estimate where features change. Thus, a wide range ensures that changes are actually recorded. The acquisition of the detection values can take place on a physically existing vehicle / transport machine or on a physically existing subsystem of the vehicle / transport machine or in a simulation of the first transport machine or a simulation of a part of the first transport machine.

Simulation methods are becoming increasingly important and used as early as possible in the development process, see HIL (Hardware In The Loop), etc.

A probability measure may be determined that indicates the probability of early detection of the error correct is. On the one hand, there is a certain dispersion of the detection values and, on the other hand, features or feature combinations can also be associated with different errors or errors of different severity.

An error which can not be detected without the stored detection values or the stored frequency spectra can be specifically incorporated into a part of the transport machine or into a simulation of at least part of the transport machine. In a vehicle electrical system, for example, the insulation of a cable can be selectively thinned. Alternatively, a cable can be heavily loaded mechanically, for example by pressure, by kinking or by pulling. As a result of the load on the cable, it may then come after a certain time to larger errors, such as cable break or short circuit. In a simulation, for example, an aging of the insulation could be simulated or a higher temperature, which reduces the insulation capacity.

The detection values or the frequency spectra can be stored outside the transport machine.

In question, for example, a transmission via radio or a reading during maintenance, e.g. in a factory. The proposed method can be carried out inexpensively with M2M (Machine to Machine) methods, e.g. via existing radio networks or via specially designed radio networks. Thus, an existing mobile network could be used, such as UMTS (Universal Mobile Telecommunications System) or LTE (Long Term Evolution).

The acquisition values or the spectra are collected for several transport machines, eg for more than 10, for more than 100 or for more than 1000 transport machines. Thus, a database can be created for the vehicles of a vehicle type of a manufacturer or for all vehicles of a manufacturer with the same on-board networks. Alternatively or additionally, however, the detection values or the spectra are also stored in the transport machine, for example in order to avoid transmission via public radio networks or to design them effectively, for example the data is transmitted only on a monthly basis.

A current profile or a voltage profile or both a current profile and a voltage profile can be detected. If both current and voltage are detected, errors can easily be predicted, which are associated with an increase or decrease in electrical power or impedance.

The fault may be a cable break or cable short. Especially with these two types of error early detection is possible on the basis of characteristic frequency patterns. Before a cable break or short circuits, sporadic sparking or isolated breakdowns may occur, which are associated in particular with charging processes or high-frequency discharge processes.

The fault may also relate to an electromagnetic radiation of a device of the transport machine into the electrical system of the transport machine. In this case, it is assumed that in vehicle manufacturing due to the increasing requirements of EMC (electromagnetic compatibility) for the error-free operation high-frequency emissions of control units are basically avoided or kept small, but occur in the event of a fault or in the run-up to an error or . occur increasingly and can then be recorded.

A feature of the frequency spectrum in the range of a frequency greater than 1 megahertz can be used. These frequencies occur, for example, more intensely only in the run-up to a cable break or short circuit or as a result of faulty electromagnetic radiation. table irradiation. The feature may be at a frequency less than 100 MHz (megahertz).

The detection values can be detected at a vehicle electrical system of the first transport machine. The electrical system can contain at least two parts, between which a separating device is arranged, through which the two parts can be electrically separated from each other. Thus, when predicting an error or when an error occurs by disconnecting a part in which the error occurs, the error can be prevented. For example. The driver can be given more time until the next workshop visit, for example at the expense of a reduced redundancy or a restriction of the driving comfort.

In the case of early detection or fault detection, a warning may be issued and / or at least one of the disconnecting devices (46) may be actuated. Thus, when detecting an error, a decision is made as to which measures are to be initiated.

An embodiment also relates to a transport machine, comprising:

a first detection device on the vehicle electrical system, preferably on a first part of the at least two parts,

- A control unit, which is connected to the separating device or with a warning unit, and which controls the separating device or the warning unit depending on detection values detected by the detection device.

This transport machine is particularly suitable for methods for error prognosis, because at a predicted Turn off parts of the vehicle electrical system so that the predicted fault can not occur at all. Additionally or alternatively, an automatic warning message can be generated. For the rest, the technical effects mentioned above for the method apply. A targeted switching off of individual segments can also be used to localize the error or the signals indicating the error.

The two parts can be redundant to each other. At the two parts can preferably be connected to each other redundant units of the transport machine. Redundancy is increasingly being used for automatic driver assistance systems, in particular for assistance systems which engage in vehicle control for a longer period, e.g. for more than one second or more than 3 seconds. Examples of such systems are: steering assistant, brake assist, overtake assistant, etc., i. Control units with central control tasks, control units for a steering system, control units for braking the transport machine.

Depending on the topology, for example, in the case of in-line networks intermeshed beyond a ring structure, the redundancy can still be ensured despite the separation of a faulty segment.

In one embodiment, the transport machine also contains a memory device for storing comparison features, comparison spectra, detection values acquired by the detection device / sensor, or spectra calculated from the detection values.

The signals detected by the detection unit can be transformed in one embodiment into the frequency domain. A transmitting unit can then send the signals detected by the detecting unit or the transformed signals to a collecting unit, for example to a server which is also called a service providing computer. In one embodiment, there is a communication connection, in particular a bus system, which lies between the control unit and the separation unit or the separation units. The same data transmission system or another data transmission system can be used for the transmission of the detection values, in particular before the transformation into the frequency domain. In a next embodiment, in addition to the separation unit, there is at least one further separation unit between the two parts in the vehicle electrical system. The further separation unit can switch on the occurrence of the predicted or predicted error or other serious errors and is eg a fuse or a circuit breaker. This avoids mispredictions and errors that are difficult or impossible to forecast. In another embodiment, an error is initially predicted. Subsequently, signal propagation time measurements are carried out in the vehicle electrical system in order to determine a segment of the electrical system in which the cause of the signals whose occurrence has caused the forecast of the fault. This segment can be separated from the rest of the vehicle electrical system as a precaution and / or it can be issued a warning message. A warning message is particularly sufficient in cases where the expected error is less serious or in which the error is expected only after a certain time, for example, in a period that is greater than a week.

In other words, a protection concept for on-board systems is specified by means of pattern recognition in the frequency domain.

This involves the detection and / or forecasting of errors in on-board networks of vehicles or other trans- regardless of the network topology used, ie both for simple topologies such as star networks, as well as for ring networks or heavily meshed networks. This applies in particular to the recognition and / or forecasting of short-circuits and the resulting parallel and serial

Discharges or arcs as well as over- or undervoltages or interference couplings.

It must be ensured in the vehicle or in the transport machine that these errors are avoided by suitable mechanisms and countermeasures are implemented in order to reduce or avoid damage to the equipment. Faulty components should be switched off reliably in order to avoid fault propagation in the electrical system and thus a failure of the entire system.

In the vehicle or in the transport machine usually errors in on-board networks, in particular short circuits, by

Fuses and circuit breakers are detected and repaired. In the event of a short circuit, the fuse will cause a higher current than permitted during normal operation. With fuses this leads to a thermal overload and thus to the separation of the nets. An automatic circuit breaker uses the additional energy to activate a switch and thus switch off the affected network. These

Protective devices are usually accommodated in vehicles / transport machines in a central fuse box.

An undervoltage contactor is usually integrated in the individual components of the vehicle, so that after detection of an undervoltage in the device, the component turns off or via an implemented energy buffer, which is installed in the components, for example in the form of batteries or buffer capacitors, the other function can be maintained for a period of time. An overvoltage contactor is usually provided by a diode or an overvoltage arrester, eg surge arrester or lightning discharge protection, which becomes conductive at a certain voltage.

All of these mechanisms only detect errors. It is not possible for you to predict mistakes.

The invention relates to, for example, vehicles / transport machines with electrical wiring systems, which detect errors in

On-board networks of vehicles, in particular short circuits, on the basis of frequency patterns allows. The frequency patterns result from the detection of electrical characteristics, e.g. Current and voltage, and their mapping in the frequency domain. The basis for detecting or detecting errors in on-board networks is that each component has a characteristic fingerprint in the frequency domain. Typical errors, such as Serial arcs occurring in the event of a cable break - precursor to a short circuit - also show a characteristic frequency response.

A suitable procedure for fault finding and troubleshooting in a vehicle electrical system is, for example, in the following procedure:

- Sampling of the current and the voltage at defined locations in the electrical system of a vehicle / a transport machine. This should be done as possible distributed over the vehicle.

- Transformation of the sampled values in the frequency domain by a Fourier transform or similar methods.

- Identification and comparison with characteristic features or fingerprints of the devices and errors. By means of frequency changes and shifts and amplitude changes, a possible error, eg in the line network or in the component, eg aging, can be detected or even predicted with suitable methods, eg previous knowledge or statistical methods. - initiate countermeasures. Eg information of the user up to the shutdown of affected network segments.

In advance, it is necessary to record a data basis of the frequency patterns as well as the characterization of individual components and fault patterns. This can also happen at runtime. For this purpose, fault scenarios and associated frequency images are recorded and evaluated for optimization. Subsequently, these evaluations can be centrally collected, analyzed and correlated with each other to improve the database regarding the functioning of on-board networks and their components. The resulting error syndromes can be used for the proactive diagnosis of errors in other vehicles / transport machines with similar on-board networks / components as well as for the further development of on-board networks.

The method described is suitable for detecting errors which are particularly difficult to detect, such as e.g. Cable breaks, i. serial

Short circuits, as well as typical parallel short circuits, e.g. The chafing of cables causes one to predict ahead of time. The aim is to prevent damage to other components of the electrical system or the entire system and the necessary replacement of components and lines, e.g. are subject to aging, that they can be consumed more efficiently.

The state of components in the electrical system can be monitored and recorded so that the quality of the electrical system and its components, in particular its availability, can be improved.

Interference coupling, eg with regard to electromagnetic compatibility, can be measured and stored at runtime and analyzed at runtime or after operation and detected errors are mitigated or remedied by countermeasures.

In the following, embodiments of the invention will be explained with reference to the accompanying drawings. FIG. 1 shows method steps for constructing a database with characteristic features, e.g. for forecasting errors,

FIG. 2 shows the chronological sequence during the construction and use of the database;

FIG. 3 shows a vehicle electrical system, e.g. of a passenger vehicle,

FIG. 4 shows an infrastructure for generating the database,

FIG. 5, method steps in error monitoring

and / or forecasting errors,

FIG. 6 shows a typical spectrum for components and / or

Consumers on the electrical system,

Figure 7 shows a typical spectrum upon the occurrence of an error, e.g. a short circuit,

FIG. 8 shows a voltage curve and a current profile at a detection device on the electrical system, and

FIG. 9 shows the spectrum with error feature calculated from the voltage curve and the current profile. FIG. 1 shows method steps for constructing a database with characteristic features, eg for forecasting errors. The method begins in a method step 1, hereinafter also referred to as a step for short. In a subsequent step 2, detection data in the time domain are first detected in a large number of cars / vehicles and then transformed into the frequency domain, for example voltage data, current data, performance data, and the like. The transformed data is then transmitted via a radio network or in another way, eg via a service or maintenance network by means of wired transmission, to one or more storage computers and stored there. Alternatively, only the data in the time domain can be transmitted and stored. A transformation into the frequency domain then takes place at a central location or at several central locations. In this case, after saving, only the data relevant for an error detection or error prognosis can be transformed, for example, only the data of a vehicle within a specific period before the occurrence of a fault in this vehicle.

By points process steps 3 are indicated, in which further data from the plurality of vehicles are transmitted to the database. The time sequence of the data generation will be explained in more detail below with reference to FIG. Spectra or data are transmitted in the time domain.

In a method step 4, an error occurs in one of the vehicles. Thereupon, in a step 5, the data of this vehicle are read from the database, preferably the spectra stored for that vehicle, in particular the spectra in a certain time before the occurrence of the error, e.g. within one week or within one month recalculated from the date of the error. However, it is also possible to use data from the vehicle which are located further back, in order, for example, to have a basis of comparison for a course of the spectra without errors. Alternatively, the spectra transmitted by other vehicles may be used, e.g. to have comparative values for error-free spectra.

In a step 6 characteristic features of the spectrum are sought, which allow time as far back as possible from the time of error still a safe detection of a deviation. This examination can be done manually, computer assisted or automatically. In particular, statistical evaluations can be carried out. When searching for characteristic features, it is preferable to examine a plurality of vehicles with the same error, so that, if necessary, further maintenance is required before the determined characteristic feature for forecasting errors can be released. It must therefore be ensured that the deviations found in the spectrum are actually related to the error that has occurred and that the change in the spectrum did not occur only in the one vehicle due to its individual production or its individual properties. However, probabilities of occurrence of the error in the chosen change may also be used. If a characteristic feature has been determined, then this can be used for the error prognosis and the error detection in future cases, which will be explained in greater detail below with reference to FIG. The search of characteristic features can be carried out with the aid of the service providing computer on which the database is stored. Alternatively, another computer or another data processing system can be used for the examination.

The method for determining the features is in one

Step 7 finished.

FIG. 2 shows the chronological sequence during the construction and use of the database. On a timeline 8, the time t is plotted. For example. a record is always generated when the vehicle has been running for 10 minutes, but at most once a day. Alternatively, the data are detected, for example, shortly after starting, for example within 30 seconds. The acquisition of the data may also be tied to other conditions, such as driving at a certain speed or in a certain range of speed. In other embodiments, data is continuously recorded while the vehicle is running. Even at standstill of the vehicle detection is possible.

At a time t 1, for example, a data set DSla is generated for a first vehicle. The structure of the data records DSla etc. will be explained in greater detail below with reference to FIG. After that, the first vehicle, for example, does not drive for a few days, so that no data records are recorded. Only at a time t2 a data set DSlb for the first vehicle is generated. One day later, a data set DSlc is generated for the first vehicle.

At a time t4, which is after the time t3, a data set DS2a for a second vehicle is generated and stored centrally or in the vehicle. The generation of further data sets is indicated by interruption of the time beam 8.

At a time t5, another record DSld for the first vehicle is generated and stored, e.g. more than two weeks after time t4.

At a time t6, an error occurs in the first vehicle, for example on the same day as the record DSld has been generated. The step 6 for searching for characteristic features explained above with reference to FIG. 1 is then carried out. The software for error detection is then updated on the basis of the feature found in the first vehicle and in the second vehicle, so that the feature found in the forecast of errors in the electrical system of these vehicles can be considered in the future. Subsequently, at time points t7 and t8, further data sets DS2b or DSle are generated by the second vehicle or by the first vehicle and transmitted to the database, for example to find deviations in other errors that occur later.

The data sets DSla to Dsle and the data sets DS2a and Ds2b have the structure explained below with reference to FIG. FIG. 3 shows a vehicle electrical system 10, e.g. of a passenger vehicle 110, see FIG. 4. The vehicle electrical system 10 contains:

Segments 12 to 26, each of which contains a plus lead and possibly a minus lead, and which are connected in a ring structure in the order named,

for example, two batteries or accumulators 28 and 30, called battery for short, the accumulator 28 being connected to the segment 26 and the accumulator 30 to the segment 18,

Separators 32 to 46, e.g. mechanical, electromechanical or electronic switches, e.g. Power field effect transistors, in particular MOSFETs (Metal Oxide Semiconductor Field Effect Transistor), which are connected in this order between the segments 12 to 26,

- A data transmission connection or a bus system 50a, 50b for controlling the separation devices 32 to 46 via

Control lines 52 to 64 and 51 and for interrogation of current and / or voltage sensors via sensor lines 72 to 82, as well

a control and communication unit SE on the bus system 50a, 50b.

Usually, the body is used as a return conductor in an automobile. An explicit return conductor is thus not necessary, but may be more advantageous, for example, with regard to interference couplings. The segments 12 and 24 are mutually redundant and serve, for example, the power supply of control units with central tasks. These control units can in each case be designed redundantly, ie two control units on the segment 12 and two control units on the segment 24.

The segments 14 and 22 are also mutually redundant and serve the power supply of control units to assist the steering of the vehicle, see, for example, the Steuergeräte 90 and 92 on the segment 22nd

The segments 16 and 38 are also mutually redundant and serve the power supply of control units, which serve, for example, the automatic braking of the vehicle.

The measurement data for the method explained with reference to FIG. 1 are detected, for example, only at the segment 12. Alternatively, measurement data can be acquired at all segments 12 to 16, 20 to 24 and optionally also at segments 18 and 26. However, the equipment of each segment 12 to 16, 20 to 24 with a detection unit is only necessary, for example, for detecting transit time differences in the localization of a segment in which a fault is being made. In another embodiment, instead of a

Ring topology selected a more meshed topology. Even with on-board networks without redundancy, the invention is used.

FIG. 4 shows an infrastructure 100 for generating the database. The infrastructure 100 includes cars or vehicles 110 to 114 which are equipped with transmission units, for example transmission unit 102 of the vehicle 110. Furthermore, the infrastructure 100 includes a server 116. The server 116 is a service-providing computer (DER), for example the following ones Ingredients contains: a memory Mem, for example a RAM (Random Access Memory) and / or a ROM (Read Only Memory),

a processor MP, e.g. a microprocessor or a microcontroller that executes instructions stored in memory Mem,

a transmitting / receiving unit R / S (Receive / Send), the transmitting part being optional,

a bidirectional data bus 18 between the transmitting / receiving unit R / S and the processor MP, and

a data / command bus 20 between the processor MP and the memory Mem.

Mem Mem stores a plurality of data sets for a plurality of vehicles, see the two data sets DSla and DS2a. The data record DSla relates to the vehicle 110. The data record DS2a, on the other hand, concerns the vehicle 112.

The data sets, see DSla, DS2a, have the same data structure:

an identification date, e.g. ID1, specifies the vehicle 110, 112, 114 from which the record comes,

- spectrum data, e.g. Spl include the spectrum for the values detected within a given time window in the respective vehicle 110, 112 or 114.

a time date, e.g. Tl, indicates the day and / or time the record was created.

The data sets DSla, DSlb, etc., DS2a, Ds2b, etc., are transmitted from the vehicles 110 to 114 to the server 116, see data link 122, 124 and 126, respectively. The data sets DSla etc. can also be electronically encrypted and / or electronically signed by the vehicles 110 to 114 are transmitted to the server 116.

Alternatively, data transmission from the server 116 to the vehicles is possible, for example, to confirm receipt the records, to update program components of the vehicles 110 to 114 or for other purposes.

FIG. 5 shows method steps in the error monitoring and / or the prediction of errors. The method begins in a method step 200, hereinafter also referred to as a step for short. In a subsequent step 202, typical patterns or characteristics for the prognosis are determined, as explained above with reference to FIG. These features are stored in a program, in the execution of which are executed by a processor, in particular the steps 206 and 208 explained below. This processor may be located in or outside the vehicle, e.g. in a service-providing computer of a workshop or service center.

In a step 202, detection values are recorded on the vehicle electrical system 10; in particular, the sampling of voltage U (t) and current I (t) is carried out at defined locations of the vehicle electrical system 10, for example at the segment 12 or 22.

In a step 204 thereafter, the transformation of the information from U / l (impedance) by Fourier transformation in the frequency domain. Alternatively, the power U * I can be transformed, only the voltage or only the current profile. For example, a signal processor which is contained in the control and communication unit SE is used for the transformation. The transformation can also be performed outside the vehicle, e.g. in a service or maintenance center.

In a step 206, the frequency range (amplitude response) calculated in step 204 is compared with typical patterns as deposited in step 201, and from this identification of errors or prognosis of errors in step 208. The identification or prognosis of Errors thus occur in the frequency domain. If it is determined in step 208 that there is no error, the method is continued, for example, in method step 202, for example without a break or at the next start of the vehicle, on the next day of travel or in the next month of driving. Thus, the method is in a loop from the method steps 202 to 208. This loop is only exited in step 208 if an error is found or predicted. However, steps 202 through 208 may be performed only during maintenance.

In the event of an error, immediately after step 208, a step 210 follows, which relates to the initiation of countermeasures, for example switching off the affected segment of vehicle electrical system 10 and / or issuing warning messages to the driver of the vehicle. The segment to be switched off can be determined, for example, via travel time measurements in which several of the sensors on the vehicle electrical system are included. Alternatively, it is tried sequentially or according to a search strategy which segment 12 to 26 is the defective segment or the segment for which an error is predicted. If the procedure is only performed during maintenance, the error can be rectified immediately. The method is then completed in a step 212. Alternatively, the method can also be continued in step 202, see arrow 214, in order, for example, to forecast further errors. FIG. 6 shows a typical spectrum 220 for components and / or consumers on the vehicle electrical system. On the horizontal axis, the frequency is plotted with negative and positive values. The spectrum 220 has three maxima in its left part. The right part of the spectrum 220 also has three maxima, of which, however, two maxima have larger amplitude values than the maxima on the left side. FIG. 7 shows a typical spectrum 230 when an error occurs, eg a short circuit. Compared to the spectrum 220, the spectrum 230 has a similar course on the left side, ie three maxima. On the other hand, the right side of the spectrum 230 has only two maxima, with the first maximum being in the range of the two first maxima of the spectrum 220 and the third maximum at high frequencies on the right side of the spectrum 230 being the corresponding maximum of the spectrum 220 at that frequency significantly surpasses and even represents the global maximum. The global maximum was assigned to an error in study 6, see FIG. 1, which occurred later or had already occurred.

FIG. 8 shows a voltage curve 250 and a current curve 252 at a detection device on the vehicle electrical system 10, as it is determined, for example, in step 202, see FIG. 5, in the time domain. Initially, the voltage is at low levels, then increases and then drops again. The current starts at high values, then drops and again alternates several times.

FIG. 9 shows the spectrum 260 with error feature 262 calculated from the voltage curve 250, see FIG. 8, and the current profile 252, see also FIG. 8, in step 204, see FIG. 5. high amplitude at a high frequency, then an error is predicted or detected. The spectrum 260 is slightly changed compared to the spectrum 230 of FIG. 7, which is due to vehicle-specific deviations.

Current and voltage can, for example, be multiplied only in the time domain, whereupon the spectrum is then calculated. Alternatively, first the spectrum of the voltage profile and the spectrum of the current profile are calculated. Thereafter, a convolution of the spectra in the frequency domain. A calculation of the impedance spectrum is possible in two ways. Likewise, only the voltage spectrum or only the current spectrum can be leranalyse or to predict errors.

Suitable error characteristics of the frequency spectrum are, for example, suitable: the position and the amplitude value of a global minimum,

the position and the amplitude value of a global maximum,

the number of local minima or local maxima,

the position and the amplitude value of a local minimum or local maximum,

- the location of a turning point,

the displacement of said features on the frequency axis,

- and similar Characteristics. The signal energy, i. the area under the curve in a certain frequency range is a typical characteristic that can be evaluated in the frequency domain and that is often much more meaningful than, for example, a minimum or maximum consideration. The steepness of the curve or the waviness of the curve may be other criteria, in particular relative to a certain frequency range.

The transformation into the frequency domain may, for example, be a Fourier transformation, a Laplace transformation, a Z transformation or another transformation.

Via run-time measurements of signals, it is also possible to determine in which segment of a vehicle electrical system an error occurs. This makes it possible to switch off this segment in a targeted manner. Alternatively, it can be determined by probing in which segment the fault is located, wherein, for example, the segments are separated from the electrical system in sequence. If a faulty segment is switched off, then the fault feature on other segments can no longer be detected, whereby the faulty segment is determined. It is also possible to use a strategy which leads to a faster detection of the faulty segment, eg disconnection of half the electrical system, detection In which part of the error lies and then continue to search in this half, for example, by dividing this half again into two approximately equal segments, etc. The electrical systems lead, for example, voltages less than 100 volts or even less than 60 volts, but greater than 5 volts or greater than 10 volts. In electric vehicles, however, the methods described can also be used in the power grid. In this case, the power grid may carry voltages greater than 100 volts or greater than 400 volts. The voltages in the drive network can typically be less than 1000 volts. Due to the use of batteries or rechargeable batteries, mainly DC mains are used. In contrast to signal lines, the voltage on DC on - board networks is comparatively constant and fluctuates, for example, by less than plus / minus 10 percent of a mean value or nominal value. However, the methods are not limited to DC networks and can thus also be used in AC networks.

The embodiments explained with reference to the figures can be combined in particular with the exemplary embodiments mentioned in the introduction. The embodiments are not to scale and are not restrictive. Modifications within the scope of expert action are possible. Although the invention has been further illustrated and described in detail by the preferred embodiment, the invention is not limited by the disclosed examples, and other variations can be derived therefrom by those skilled in the art without departing from the scope of the invention.

Claims

claims

1. A method for forecasting an error or error detection in a transport machine (110 to 114), with

Detecting (2) at least one type of detection values in the time domain in a first transport machine (110),

Storing (2) the detection values or storing frequency spectra of the detection values,

Waiting for a mistake (4),

Examining (6) the frequency spectra stored prior to the occurrence of the error or examining (6) frequency spectra calculated from the detection values stored prior to the occurrence of the error for at least one characteristic indicative of the occurrence of the error (4);

Using (206) the error detection or error detection feature in the first transport machine (110) or in a second transport machine (112).

2. The method according to claim 1, wherein the detection values are detected at an on-board network (10) of the first transport machine (110).

3. The method of claim 1 or 2, wherein the method for a plurality of vehicles (110 to 114) is performed.

4. Method according to one of the preceding claims, wherein the detection of the detection values takes place in a simulation of the first transport machine (110) or a simulation of a part of the first transport machine (110).

5. The method according to any one of the preceding claims, wherein a probability measure is determined which indicates the probability with which the early detection of the error or the error detection is correct.

6. The method according to any one of the preceding claims, wherein selectively one without the stored detection values or the stored frequency spectra undetectable error in a part of the transport machine (110) or in a simulation of at least part of the transport machine (110) is installed.

A method according to any one of the preceding claims, wherein the detection values or the frequency spectra are stored outside the transport machine (110).

8. The method according to any one of the preceding claims, wherein a current waveform (252) or a voltage waveform (250) or both a current waveform (252) and a voltage waveform (250) are detected.

9. The method of claim 1, wherein the fault relates to a cable break or cable short.

10. The method according to any one of the preceding claims, wherein the error relates to electromagnetic radiation of a device of the transport machine (110) in the electrical system (10) of the transport machine (110).

11. The method according to any one of the preceding claims, wherein a feature of a frequency spectrum is used, the at

Frequencies greater than 1 megahertz.

12. The method according to any one of the preceding claims, wherein the detection values on an electrical system (10) of the first transport machine (110) are detected, and

wherein the electrical system (10) at least two parts (12, 24), between which a separating device (46) is arranged, through which the two parts (12, 24) can be electrically separated from each other.

13. The method of claim 12, wherein a warning is issued in an early detection or error detection and / or at least one of the separation devices (46) is actuated.

14. Transport machine (110), with

an electrical system (10), which contains at least two parts (12, 24),

a separating device (46) arranged between the at least two parts (12, 24), with which the two parts (12, 24) can be electrically separated from one another,

a first detection device (72) on the vehicle electrical system (10), a control unit (SE) which is connected to the separating device (51) or to a warning unit and which, depending on the detection values detected by the detection device (72), disconnects the separating device (46 ) or the warning unit activates.

15. Transport machine (110) according to claim 14, wherein the two parts (12, 24) are mutually redundant, and

wherein at the two parts (12, 24) preferably mutually redundant units of the transport machine (110) are connected.