EP0824735A2

EP0824735A2 - Device and process for processing sheet articles such as bank notes

Info

Publication number: EP0824735A2
Application number: EP96919717A
Authority: EP
Inventors: Paul Dr. Smith; Walter Herrmann; Bernd Dr. Wunderer; Dieter Stein
Original assignee: Giesecke and Devrient GmbH
Current assignee: Giesecke and Devrient GmbH
Priority date: 1995-05-11
Filing date: 1996-05-08
Publication date: 1998-02-25
Anticipated expiration: 2016-05-08
Also published as: JPH11506555A; WO1996036931A3; EP1168252A3; EP1168252A2; DE19618541A1; CN1078724C; AU5815896A; ATE214182T1; ES2171685T3; WO1996036931A2; CN1187894A; US6074081A; US6151534A; EP0824735B1; DE59608844D1; RU2168210C2

Abstract

In a device for checking sheet articles, at least one sensor device is provided with a store in which data sets for several sheets can be maintained. Each data set is provided with regions in which data from at least one other sensor unit can be stored. The sensor unit is preferably provided with a measurement unit and an evaluation unit, the store of the sensor unit being provided in the evaluation unit.

Description

Device and method for processing sheets. such as B. banknotes

The invention relates to an apparatus and a method for processing sheet material, such as. B. banknotes.

DE 2760166 shows such a device which is constructed from various units. In a separating device, the sheet material present in a stack is separated sheet by sheet and transferred to a transport path which transports the separated sheet material through the device.

A plurality of sensor units are attached along the transport path, with each sensor unit detecting certain features of the sheet material and combining them into a measurement result. The structure of the sensor units used here is shown in DE-PS 2760 165. Each sensor unit has a transducer which detects certain features of the sheet material and converts it into an electrical signal. This signal is converted in a signal processing stage. In general, the mostly analog signal is converted into digital measurement data here. The measurement data are then finally transformed into an evaluation unit in the sensor unit

Yes / No statement transformed. This then forms the measurement result of the sensor unit and is stored in a central memory.

The central memory is used as a connection for data exchange between the units of the device. All units can access it and write or read the data necessary for processing the sheet material. A data record is stored in the central memory for several sheets.

From the measurement results of the sensor units for each sheet material stored in the central memory, evaluation information is first generated in a central evaluation unit. By means of a The target units for the sheet material in question are determined from the evaluation information using the stored decision table.

The target units can be, for example, stackers for stacking the sheet material or shredders for destroying the sheet material. The target units for the corresponding sheet material are stored in the central memory. On the basis of the stored target unit, the sheet material is appropriately guided and deposited by the transport unit. After the sheet material has been transported to the target unit, the transport unit writes positive or negative information about the output of the processing to the central memory.

The machining process in the device is controlled by means of a control unit. This too has access to the central memory and can monitor and log the processing operation on the basis of the information stored there. Furthermore, the control unit serves to initialize the units of the device depending on an operating mode set by the operator. This includes, for example, storing the correct decision table for the selected operating mode in the central evaluation unit.

In the known system, each sensor unit can only derive its measurement result from the measurement data of the sheet material that it has recorded.

Proceeding from this, the object of the invention is to propose a device for processing sheet material which enables the

To improve the quality of the derivation of a measurement result of the sensor units.

This task is solved by the features of the main claim. The basic idea of the invention essentially consists in using data from other sensor units about the corresponding sheet material when deriving a measurement result from a sensor unit. For this purpose, a memory is provided in at least one sensor unit, in which data records of several sheets can be managed. Areas are provided in each of these data sets in which data from at least one other sensor unit can be stored.

It is an advantage of the invention that the sensor unit has data from other sensor units available that it can take into account when deriving its own measurement result. By knowing these data, the sensor unit is able to derive its measurement result faster and more accurately from these data.

The sensor unit preferably has a measuring unit and an evaluation unit, the memory of the sensor unit being provided in the evaluation unit. Furthermore, the measurement results of the sensor unit are not restricted to a yes / no statement, but rather are provided with a higher information content. The measurement results can be, for example, the length or the width of the sheet material in mm, a measure for the contamination or the correspondence of the printed image with a reference image, the distance of a metal thread from the leading edge of the sheet material, an identification number for the type or position of the sheet material or the like to be.

Further features and advantages of the invention result from the secondary claim, the subclaims and the description of an embodiment of the invention with reference to the figures. Show it:

1 is a schematic diagram of an embodiment of the invention, Fig. 2 representation of the memory content in the evaluation unit

sensor

3 representation of the memory content in the central evaluation unit,

4 shows a flow diagram of a first exemplary embodiment of the method according to the invention,

5 shows the representation of the measurement results in discrete classes,

6 representation of the control matrix of the first embodiment,

7 representation of the change conditions for sorting classes,

8 shows a flow chart of a second exemplary embodiment of the method according to the invention,

9 representation of the mapping of the measurement results in overlapping classes with membership function,

10 representation of the control matrix of the second exemplary embodiment,

11 representation of the membership functions of the sorting classes,

12 graphical derivation of a resulting membership function of a sorting class,

Fig. 13 representation of the resulting membership functions. Fig. 1 shows a schematic diagram of an embodiment of the invention. The sheet material is separated from a stack sheet by sheet in a separating unit and transferred to a transport path which transports the sheets through the device and is controlled by a transport unit 30. The transport route is divided into individual sections, each of which is controlled by decentralized subunits 30.1-30.M of the transport unit 30.

During the separation, an identification ID is assigned to each sheet, by means of which the sheet is uniquely recognized by the units of the device. The data required for processing a sheet are exchanged using a connection 100 using the identification ID of the sheet. The connection 100 connects both the subunits 30.1-30.M and also a central evaluation unit 10, a plurality of sensor units 20.1-20.N and a control unit 40.

The sensor units 20.1-20.N each consist of a measuring unit 21.1-21N and an evaluation unit 22.1-22N. Each measuring unit 21. N has a measuring sensor which detects certain features of the sheet material and converts them into electrical signals. These electrical signals are then converted into digital measurement data and can optionally be standardized and / or transformed before further processing. The evaluation unit 22.n of the sensor 20.n receives the measurement data from the measurement unit 21.n and derives a measurement result using the measurement data.

A memory is provided in at least one evaluation unit 22n, the content of which is shown in FIG. 2. The evaluation unit 22.2 was selected here as an example. Several data sets can be managed in the memory of the evaluation unit 22.2. Each record is a sheet of a particular agreed identification ID assigned. The memory shown here is able to manage a number L of data records.

Each data record has an area for external data ED. Either measurement data MD or measurement results ME of other sensor units are stored in it. 2, for example, the measurement data MD of the sensor unit 20.3 and the measurement results of the sensor unit 20.1 are stored in each data record. For example, the measurement data of the sensor unit 20.3 for the sheet with the identification ID = 2 are designated here with MD.23, the first index corresponding to the identification ID of the banknote = 2 and the second index corresponding to the index of the sensor unit = 3. The same procedure is followed for the other data in the designation.

Preferably, the measurement data MD supplied by the measuring unit 21.2 for each sheet are also stored in the memory of the evaluation unit 22.2.

From the own measurement data MD and the external data ED of a data set, the evaluation unit 22.2 derives a corresponding measurement result ME for each sheet, which can optionally be stored in the corresponding data set.

If the measurement result for a sheet is determined, this is written onto the data line 100 with the corresponding identification ID of the sheet. If necessary, the measurement result can now be read by other sensor units and stored in the memory of the evaluation unit of this serisor unit. If it is necessary to know certain measurement data from one sensor to derive the measurement result from another sensor unit, the latter must also write the corresponding measurement data onto the data line 100 so that it can read the other sensor unit. Alternatively, the measurement data can also be written only after receipt of a corresponding signal from the other sensor unit. Furthermore, the device has a central evaluation unit 10 with a memory, the content of which is shown in FIG. 3. The central evaluation unit 10 reads the measurement results of all sensor units 20.1-20.N from the data line 100 and stores them under the identification ID of the corresponding sheet. If the measurement results of all sensor units are known for an identification ID, the central evaluation unit 10 derives a sorting class KL for the corresponding sheet material from the measurement results and writes the identification ID and the associated sorting class KL on the data line 100 Sorting class KL can be saved in the memory under the corresponding identification of the sheet.

The sorting class KL is evaluated by the subunits of the transport unit, which control the transport of the sheet into the target unit. If the corresponding subunit 30.m is not responsible for processing the sheet, it is forwarded to the subsequent subunit 30.m + l. In the other case, the sheet is guided to the corresponding manipulators of the subunit 30.m and processed. After the sheet material has been processed, the processing unit writes corresponding positive or negative information about the completion of the processing on the data line 100. This information is read, for example, by the control unit 40 and used in the logging of the machining process.

Furthermore, each subunit 30.m can write error messages on the data line if, for example, a sheet jam occurs in the transport system of the subunit 30.m. These error messages can be interpreted by other units of the device and appropriate measures can be initiated.

The subunits 30.m are preferably designed such that they control the electrical and mechanical functions of the transport route. For this count among other things the drive of the transport route, the switching of the switches within the transport route, the measurement of the position of the sheet material by means of light barriers etc. Furthermore, the sub-units 30.m can also control special electrical or mechanical manipulators within the units perform the device. This includes, for example, the control of the singling components, the stacking wheels and the shredder rollers etc.

The control unit 40 serves to control and log the processing operations of the sheets. It is able to send control information via data line 100, which is interpreted accordingly by the individual units. Such control information can be used, for example, to bring the device into a processing status selected by the operator. Furthermore, the control unit 40 can have special programs or reference data stored by the control unit 40 in the other units of the device via the data line 100. For this purpose, the control unit 40 has mass memories in which this data is managed.

On the basis of the data from the sub-units 30.1-30.M, the sensor units 20.1-20.N and the sorting class SL of the central evaluation unit 10, the control unit 40 can monitor and log the processing operation of each individual sheet. During the actual processing of the sheet material, the function of the control unit is only limited to listening to the data line 100.

The data line 100 is designed as a data bus. A CAN bus is preferably used. This is particularly well suited for so-called real-time applications, as they are mainly available here. Optionally can Further data lines 101 and 102 are provided parallel to the data line 100, so that the data line 100 is relieved.

The data line 101 can also be implemented by means of a CAN bus and serves to improve the data exchange between the sensor units 20.1-20.N and the central evaluation unit 10. This is particularly useful if a lot of measurement data is exchanged between the sensor units 20.n, which often have a high data volume.

The data line 102 is used especially by the control unit 40 for so-called non-real-time applications. In this way, for example, when the device is initialized into a specific operating state, programs or larger-scale reference data can be written into the sensor units 20 or the central evaluation unit 10. A connection to the subunits 30.m can also be dispensed with, since the amounts of data transmitted there are generally small.

The sorting class of a sheet can be derived from the measurement results of the sensor units, for example using freely configurable tables and / or matrices which are managed in a memory of the central evaluation unit 10. During the derivation, continuous measurement results are initially mapped to classes. Discrete measurement results are assigned directly to a class. Individual classes are combined into one property of the sheet with different characteristics. By means of a rule matrix, any, but selected combinations of different forms of a set of properties can be assigned to a sorting class.

4 shows a flow diagram of a first exemplary embodiment of the method according to the invention for processing sheet material, here specifically Banknotes. The measurement data MD of the banknote are recorded by the sensors 20.n. Using these measurement data MD, measurement results ME of the banknote are derived and stored in the evaluation unit 10 according to FIG. 3.

In the first embodiment, the measurement results ME are initially mapped to discrete classes. An example of such a mapping is shown in FIG. 5. In this case, the measurement result should represent the area of the banknote in mm ² which is covered by spots. If, for example, 140 mm ^{2 is} determined as the measurement result in a first measurement Mi, this measurement result is mapped to the class with the class identifier 4. The number of classes and the location of the class boundaries can be configured as desired. Classes 0 to 5 can be combined to form the "stain" property. Each class thus represents an expression of the "stain" property "," Much "etc. occupied.

6 shows the control matrix of the first exemplary embodiment. For the individual properties "double deduction", "disruption", etc., the associated classes with the verbal and class identifiers are given. For a better overview, various properties are summarized again in a higher-level group.

To derive the sorting class of the banknote, a property vector is first formed from the classes of all properties. 6 four class vectors V_ to V are shown as an example. In each property the class is marked that corresponds to the measurement result of the banknote. With the property "stains", the measurement result of the leaf material belonging to the class vector Vi lies, for example, in the class "little", while the measurement result of the property "dog ears" is in the class "very little". The class vector thus classifies the characteristics of all the properties of a banknote.

The rule matrix consists of a number of rules, which are designated here with the numbers 1 to 5. Each rule consists of a rule vector which, like the class vector, is formed from the classes of all properties. In contrast to the class vector, it is possible, however, for several classes of a property to be marked, for example for the property “soiling” in rules 1 to 5. Each of rules 1 to 5 is assigned a sorting class, which here with the the respective sorting destination is designated “stacker 1”, “stacker 2” etc. In general, the same sorting class can also be assigned to several rules.

The statements of the individual rules can be formulated verbally roughly as follows. According to rule 1, those banknotes are assigned the sorting class “stacker 1”, the denomination of which is $ 50, which are oriented upwards, have all security features, are clean and have very few defects. According to rule 2, those banknotes, de - whose orientation is directed downwards and which otherwise have the same properties as the banknotes according to rule 1, assigned to the sorting class "stacker 2". The sorting class "stacker 3" is assigned all $ 1 and $ 2 banknotes which have at least one correct security thread, which are clean and have few defects. The sorting class "stacker 4" is assigned to those banknotes which are clean regardless of the denomination are few defects and for which neither the property "watermark" nor the property "security thread" is correct. The "Shredder" sorting class is assigned all banknotes that have the correct security features and are dirty, regardless of their denomination and defects. To derive the sorting classes, the markings of the class vector, e.g. B. Vi, compared with the corresponding markings of Regelvekto¬ ren 1, 2, 3, 4, 5 in sequence. The sorting class, which is assigned to the first rule vector, which is marked in all classes of the class vector, is assigned to the sheet material as a sorting class.

If the markings of none of the rule vectors match all of the markings of the class vector, the sheet material is assigned an arbitrary but firmly selected sorting class.

For the examples in FIG. 6, this means that the sorting class "stacker 2" is assigned to the sheet material for class vector Vi. The sheet material for class vector V ₂ is assigned the sorting class "stacker 4". The sorting class "shredder" is assigned to the sheet material to class vector V3. Since the marking of no rule vector matches all the markings of class vector V ₄ , any arbitrary but firmly selected sorting class is assigned to this sheet material, which is referred to as "reject" should.

After the sorting class has been assigned to the sheet material, it is transported to the corresponding target unit based on the sorting class. The sheets with the "Reject" sorting class are generally stacked in a so-called reject compartment, where they can then be removed from the device by the operator and inspected.

In order to prevent an unauthorized change in the rules of the rule matrix, a security level SL is assigned to each class. It can be used to specify which users can make changes in this class. Here, for example, the value 3 stands for the developer, 2 for the supervisor and 1 for the operator of the device. It is thus possible for the operator of the device to change the classes of the "denomination of the banknote" property while changing the properties of the group "Security features of the banknote" may only be changed by the supervisor.

It is also possible to assign a weight G to at least certain classes. Using these weights G, the rules of the rule matrix can be checked for consistency, for example, or the sorting class derived from the rule matrix can be changed if necessary.

As an example, only one possible meaning for the weights G of the classes of the group "security features of the banknote" is to be explained here. In addition to the two properties "watermark" and "security thread" shown here, there are generally a number of other properties in this group which are not mentioned here for reasons of clarity.

For the assessment of a banknote, it may be of interest not only to check the individual properties of the security features, but also to carry out a weighting of the individual properties to one another, for example in order to distinguish meaningful properties from less meaningful ones. Here, for example, the correctness of the "security thread" property is rated 5 higher than the correctness of the "watermark" property 3. In the case of a large number of such properties, appropriate weights can be used to fine-tune the individual properties.

A minimum weight can now be determined for each rule from the weights of the individual classes in the group “Security features of the banknote” by adding up the weights of the individual classes of each property of the group with the lowest marked weight of the rule For the example in FIG. 6, this means that rule 1, 2 and 5 in the "Security features of the banknote" group is assigned a minimum weight of 8. For rule 3, the minimum weight is 5 and for rule 4, the minimum weight is 0.

The minimum weight determined in this way for each rule in the group "Security features of the banknote" thus gives a measure of the security of the banknote. A high minimum weight stands for high security and a low minimum weight for low security. For a fit banknote the desired security can thus be defined by a predetermined minimum weight in the "security features of the banknote" group.

Such a predetermined minimum weight for the security of a fit banknote is to be determined from the weights of the property "denomination". The predetermined minimum weight of a rule in the group "Security features of the banknote" for fit banknotes results here as the maximum of the weights of the marked ones Classes of the rule in the property "Denomination". For rule 1, 2, 4 and 5 there is a minimum weight for the security of an unfit banknote of 8 and for rule 3 of 3.

A comparison of the specified minimum weight for the security of a fit banknote according to the property "Denomination" with the weight in the group "Security features of the banknote" for each rule shows that the minimum weights for the security of a fit banknote in the rules 1, 2, 3 and 5 are greater than the minimum weights in the "Security features of the banknote" group. The relation and thus the criterion for a fit banknote is only not fulfilled in rule 4. The consi - should be checked for each rule. The introduction of a minimum weight for each rule in the "Security features of the banknote" group creates a criterion with which even banknotes with different security features can be compared with one another. If necessary, other evaluation algorithms for the individual weights of the classes are of course also conceivable.

As shown in FIG. 7, the sorting class determined with the aid of the control matrix can optionally be changed again in retrospect depending on a specific condition. Such a subsequent change can be helpful, for example, when maintaining the device or when designing the control matrix.

The conditions can be derived from the rule matrix, such as the minimum weight MG for a rule in the group "security features of the banknote". Furthermore, the conditions can also depend on measurement results from the sensors or the class identification of a specific measurement result All data available to the evaluation device can be used in any combination in any condition.

It is also possible to randomly redirect certain banknotes in a randomly distributed manner using a random number generator RND. In the example shown in FIG. 7, statistically 20% of the banknotes are diverted from the "Shredder" sorting class into the "Reject" sorting class. Such a procedure makes it possible, for example, to continuously check the sorting quality of the banknotes by the operator of the device personally examining the banknotes redirected with the "Reject" sorting class. If necessary, the latter can then use the visual inspection to determine the class limits Change classes appropriately. Furthermore, by subsequently changing the sorting class, it is easily possible to divert the corresponding banknotes in the event of a malfunction without having to make major changes in the rule matrix.

A second exemplary embodiment of the method according to the invention for processing sheet material is shown in FIG. 8. Here too, as already explained for the first exemplary embodiment, measurement data are first acquired by the sensors 20.n and MD measurement results ME are derived using this measurement data.

In contrast to the first exemplary embodiment, the measurement results ME are mapped or fuzzified to overlapping classes. An example of such a mapping is shown in FIG. 9. For the sake of clarity, only the properties “dirt”, “dog ears” and “stains” were used from the properties of the first exemplary embodiment. In this example, the measurement results of the sheet material can each take values between 0 and 1. Each property is three overlapping classes For the property "pollution" these are the classes "strong" with measurement results in the interval from 0 to 0.5, "medium" in the interval from 0 to 1 and "low" in the interval from 0.5 to 1. The classes "strong", "Medium", "low" are used as fuzzy classes in the following.

Each fuzzy class is assigned a membership function shown in FIG. 9. The number of overlapping fuzzy classes and the form of the various membership functions can be set as desired. The functionality of the method can be optimized for the respective application by a suitable choice of the membership functions.

9 each shows the measurement results of two measurements, M. and M2, with the membership results resulting from the membership functions. values entered. The Mi measurement is a banknote with little contamination, a relatively large number of dog ears and few stains. The measurement M2 is heavier than the measurement Mi and has more dog ears. Furthermore, it shows fewer spots than the measurement Mi.

A rule matrix is defined by means of the fuzzy classes, which is shown in FIG. 10. The possible combinations of the individual classes of the properties “pollution”, “dog's ear” and “stains” are plotted in the columns of the rule matrix. The last column of the rule matrix forms a “sorting” property with three fuzzy classes, that with “stacker "," Shredder "and" Reject ". In the rows of the rule matrix, a rule 1 to 8 is plotted, which in each case assigns a possible combination of fuzzy classes of the three properties to a fuzzy class of the" sorting "property. The respective value of the measurements M_ and M ₂ determined in FIG. 9 is given under the respective designation of the fuzz class. The determination of the values specified for the fuzzy classes of the "sorting" property is explained below.

The rules of the rule matrix can be represented verbally as follows. Rule 1, for example, says that a banknote with little soiling, many dog-eared ears and little stains of the fuzzy class "reject" should be assigned a property "sorting". According to rule 2, a banknote with medium soiling, many dog ears and few stains is the

Fuzzy class "Shredder" assigned to the "Sorting" property, etc. The rule matrix is limited to 8 rules here, since no further meaningful combinations occur in the measurements Mi and M ₂ . In principle, however, it is not necessary that the rule matrix rules for all possible combinations contains contains. It is sufficient if it only contains rules for relevant combinations.

To derive a sorting class of a banknote, as shown in FIG. 11, the property "sorting" is first assigned a corresponding membership function to each fuzzy class.

The resulting fuzzy classes “stacker”, “shredder”, “reject” of the sorting are first derived from the fuzzy classes with their membership functions and the rule matrix using a so-called inference machine.

The fuzzy classes of the "sorting" property resulting from the rules are obtained by first linking, as shown in FIG. 10, the corresponding membership values of the measurement results within a rule and assigning the result to the linking of the sorting. Here, as a simple case for a link, the selection of the smallest membership value was selected (sometimes framed).

12 shows, for example, the fuzzy class "shredder" of the "sorting" property resulting from the corresponding rules. The membership function of the fuzzy class “shredder” is generally cut off at the corresponding height in accordance with the result of the linkage. FIGS. 12a and 12b show this process for the results of measurement M2 of rule 4. According to the one shown in FIG. 10 The rule matrix provides the

Rule 4 for the measurement M2 and the fuzzy class "Shredder" of the property "Sorting" the value 0.2. Consequently, the membership function of the fuzzy class "shredder" is cut off at the value 0.2. The portions of the individual rules obtained in this way are linked to one another. For the sake of simplicity, the maximum covered area of the selected individual areas. The result of the link is shown in Fig. 12c.

If the analog method is carried out for all fuzzy classes of the "sorting" property and all rules, the resulting fuzzy classes of the "Sorting" property shown in FIGS. 13a and 13b with their membership functions are obtained for the measurement Mi and the measurement M2. The result determined in FIG. 12c, for example, can be found here in FIG. 13b.

In a last step, a discrete sorting class must be derived from the resulting fuzzy classes of the "sorting" property, or the "sorting" property must be denazified. A simple possibility for such a derivation is to assign the sorting class to the sheet material whose fuzzy class has the largest area. In the case of the measurement results Mi, the sorting class “Reject” would be assigned to the sheet material and the sorting class “Shredder” would be assigned to the sheet material with the measured values M2.

A more complex method for deriving the sorting class from the resulting fuzzy classes of the “sorting” property consists, for example, of first combining the individual resulting fuzzy classes “stacker”, “shredder”, “reject” of the “sorting” property, for example by combining and to calculate the position of the center of gravity from the resulting area. By rounding this value can be mapped to a discrete sorting class.

Of course, it is also possible to transfer further possibilities known from the prior art in dealing with fuzzy logic to the problem of processing sheet material in the sense described above. Analogous to the first exemplary embodiment of the method, it is also possible here to provide the individual rules with security levels. The use of weights for each class is also possible by, for example, linking the respective membership function of a fuzzy class with the corresponding weight, for example by multiplication. The treatment of the safety level and the weights can be carried out analogously to the first embodiment.

Claims

claims

1. Device for processing sheet material, such as Banknotes, with a separating unit that separates the sheet material from a stack sheet by sheet, a transport unit that transports the separated sheets through the device, a plurality of sensor units, each with a measuring unit and an evaluation unit that determine measurement results for each sheet central evaluation unit, which is used by the

Sensor units ascertained measurement results assign a specific target unit to each sheet, a plurality of target units in which sheets are stacked or destroyed, a control unit which controls and / or logs the processing of the sheets and a connection for data exchange between the units,

characterized in that

A memory is provided in at least one sensor unit (20.n) in which at least one data record is managed, at least one area (ED) being provided in this data record in which data from at least one other sensor unit can be stored.

2. Device according to claim 1, characterized in that data records of several sheets are managed in the memory of at least one sensor unit (20.n).

3. Apparatus according to claim 1, characterized in that the memory of the sensor unit (20.n) in the evaluation unit (22.n) of the sensor unit (20.n) is provided.

4. The device according to claim 1, characterized in that in each data set areas (MD, ME) are provided in which the measurement data determined by the measuring unit of the sensor unit (20.n) and / or by the evaluation unit of the Measured results determined by the sensor unit (20.n) can be stored.

5. The device according to claim 1, characterized in that a memory is provided in the central evaluation unit (10) in which data records are managed by several sheets, areas (ME) being provided in each data record, in which the measurement results of Sensor units (20.1 - 20.N) can be saved for a specific sheet.

6. The device according to claim 5, characterized in that a memory is provided in the central evaluation unit, in which freely configurable tables and / or matrices are administered which are used in deriving a sorting class of a sheet.

7. Apparatus according to claim 5 or 6, characterized in that means are provided which derive a sorting class for this sheet from the measurement results of the sensor units for a particular sheet.

8. The device according to claim 5, characterized in that an area (SK) is provided in each data set in which a sorting class of the sheet can be stored.

9. The device according to claim 1, characterized in that the transport unit (30) has a plurality of decentrally operating sub-units (30.1-30M).

10. The device according to claim 9, characterized in that each subunit (30.1 -30.M) controls a portion of a transport route of the transport unit (30).

11. The device according to claim 9, characterized in that each subunit (30.1 -30.M) controls a specific unit of the device.

12. The apparatus according to claim 1, characterized in that the connec tion for data exchange (100) is a data bus.

13. The apparatus according to claim 12, characterized in that the data bus is a CAN bus.

14. The apparatus according to claim 12, characterized in that further connections (101, 102) are arranged parallel to the connection for data exchange (100), which at least partially connect the units.

15. The apparatus according to claim 14, characterized in that the further connections (101, 102) are data buses.

16. Methods for processing sheet material, such as Banknotes that perform the following steps:

Acquisition of measurement data by means of several sensor units, derivation of measurement results from the acquired measurement data, mapping of the measurement results to classes, Grouping of individual classes into a property of the leaf material with different characteristics,

Establishment of a freely configurable rule matrix, by means of which any but selected combinations of different characteristics can be assigned to a set of properties of a sort class,

Derivation of the sorting class of the sheet material using the rule matrix,

Assigning the sorting class to the sheet material, - Transporting the sheet material to a target unit, based on the sorting class of the sheet material.

17. The method according to claim 16, characterized in that the measurement results are at least partially mapped to discrete classes.

18. The method according to claim 17, characterized in that a class vector is formed from the classes of all properties and in each case exactly the class of each property is marked, which corresponds to the respective measurement result, - the rule matrix contains a number of rules, each rule consisting of one Rule vector is formed from the classes of all properties and at least one class of each property is selected, each rule is assigned a sort class.

19. The method according to claim 18, characterized in that the Ablei¬ line of the sorting class, the markings of the class vector with the corresponding markings of the rule vectors are compared in their order, and if If a rule vector contains all the markings of the class vector, the sorting class assigned to this rule vector is assigned to the sheet material; if no rule vector contains all the markings of the class vector, the sheet material is assigned any but selected sorting class.

20. The method according to claim 18, characterized in that the classes are at least partially assigned weights and the rules are checked at least partially based on these weights for consistency.

21. The method according to claim 16, characterized in that the measurement results are at least partially mapped to overlapping classes.

22. The method according to claim 21, characterized in that each of these overlapping classes is assigned a fuzzy class with a membership function, each sorting class is assigned a fuzzy class with a membership function and the rule matrix contains the rules of an inference machine.

23. The method according to claim 22, characterized in that a sorting class is derived from the fuzzy classes and assigned to the sheet material by an inference machine.

24. The method according to claim 22, characterized in that the classes are at least partially assigned a weight and the affiliation functions of the corresponding fuzzy classes can be linked to this weight.

25. The method according to claim 16, characterized in that the assigned sorting class is changed depending on any but fixed conditions.

26. The method according to claim 25, characterized in that the classes are at least partially assigned a weight and at least one condition depends on the weight of at least one class.

27. The method according to claim 25, characterized in that at least one condition depends on at least one measurement date and / or measurement result.

28. The method according to claim 25, characterized in that the at least one condition depends on at least one random variable.

29. The method according to claim 16, characterized in that the rule matrix contains a number of rules, each rule one

Security level is assigned, with which an unauthorized change of the rule is prevented.

30. The method according to claim 16, characterized in that the classes are at least partially assigned weights and by means of the weights a fine gradation of the individual properties is carried out against each other.

31. Process for processing sheet material, such as. B. banknotes, characterized in that the derivation of at least one measurement result of a sensor unit is carried out using data from at least one other sensor unit.