WO2006128438A1

WO2006128438A1 - Method for evaluating signal sequences of sensors in order to determine positions and situations when displacing transport objects on conveyor devices

Info

Publication number: WO2006128438A1
Application number: PCT/DE2006/000937
Authority: WO
Inventors: Jörg MONECKE; Christian Heuschmann
Original assignee: Otto-Von-Guericke-Universität Magdeburg
Priority date: 2005-06-03
Filing date: 2006-05-31
Publication date: 2006-12-07
Also published as: WO2006128437A1

Abstract

The invention relates to a method which is used to evaluate signal sequences of sensors in order to determine positions and situations when displacing transport objects on conveyor devices, wherein the signal sequences of the sensors, which correspond to typical signal sequence patterns and which are detected during calibrated paths of the transport object, are defined as classes. The conveyor devices are allocated to typical classes, the signal sequences are continuously examined as to whether they can be allocated to a class which represents a conveyor device and, in the event of a possible allocation in relation to a class for the displaced transport object, detected parameters and displacement scenarios are allocated according to the position.

Description

Method for evaluating the signal sequences of sensors for position and position determination when moving transport objects on conveyors

The invention relates to a method for evaluating the signal sequences of sensors for position and position determination when moving transport objects on conveyors.

When moving or moving transport objects, such. As pallets, boxes, boxes, etc., the position and orientation of the objects on a conveyor of importance.

The prior art is additionally installed systems such as light barrier, laser and transponder systems, GPS positioning, radio frequency identification (RFID) or similar arrangements. All processes and devices are characterized by high technical installation and maintenance costs and high acquisition costs. The medium for locating the transport objects must have free access to the transport object to be observed in all the methods mentioned. Building parts, tunnels, shoring, control cabinets and similar unavoidable technical equipment usually lead to a temporary interruption of the position and position determination.

From EP 0 698 568 a device for detecting conveyed on conveyor tracks, in particular of piece goods in roller conveyors is known, are provided in which arranged between the rollers light sensors that measure perpendicular to the bottom of the transport objects and recognize them when crossing their sphere of action. The conveyor track is divided into blocks by the sensors. The disadvantage is the additional sensors on the conveyor and the discrete recognition of the transport objects. For conveyors with closed traction means, such as. As belt conveyors, this Einrich- not applicable.

EP 0 279 559 B1 relates to a device in which the edges of the transport objects are detected by means of an unspecified arrangement. The device has a counter, which is clocked at passing by transport objects at a lower rate, as in the spaces between the transport objects. From the sequence of counts, the relative position between the objects is determined. The disadvantage consists in the additional sensors on the conveyor system and in that the detection of transport objects is possible only at individual, fixed points of a conveyor system.

According to DE 102 26 673 A1, a conveyor system is divided into its individual conveyor. Upon entry of the transport object in a conveyor this is detected by a position detection sensor installed on the system and tracked on its way along the conveyor line. The sensors' information is routed via a bus connection to a "position manager." A control unit is used to influence the speeds of the individual conveyors, and the use of several position detection sensors in one conveyor is also proposed Sensor technology with the required extensive signal transmission and processing.

According to JP 2003306217 A, a synthetic resin conveyor belt is mixed with magnetic powder, and thus a moving object is detected by a metal detector.

According to DE 3 513 753 A1, the transport objects are equipped with a laterally mounted switching device consisting of at least two inductive sensors. The conveyor system is located at certain target Positions also inductive sensors that can be arranged spatially and in the switching distance differently. If the transport object reaches a destination point with a matching sensor arrangement, switching signals for further transport, transfer or arrest are triggered. The use of this solution requires the removal or retrofitting of the conveyor system with an additional sensor.

It is also possible to determine the position and position of moving transport objects on conveyors without the conveyor system having specific sensors or other special devices for position and position determination:

With sensors in or on the transport object, the constructive assemblies of conveyors can be detected contactless and wear-free.

Suitable sensors detect when moving or moving the transport object, the metallic components of the conveyor, such as rollers, frames, covers, branches, switches and similar devices directly or - as in continuous conveyors with endless circulating traction means (eg belt or apron conveyors) - through non-metallic elements.

The signals or signal sequences generated in the sensors must now be evaluated with regard to their content of position information.

The object of the invention is therefore to develop a method that enables the evaluation of the signal sequences of sensors for position and position determination when moving transport objects on conveyors with little effort.

This task is accomplished by the design of the method for the evaluation of the Signal sequences of sensors for position and position determination when moving transport objects on conveyors according to the features of the main claim solved. In the associated subclaims particularly advantageous embodiments of the method are claimed.

It is part of the essence of the invention that for each conveyor from its typical system elements also typical signal sequences of one or more sensors, which are characterized by pulse widths, pulse width ratios, pulse sequences, frequencies or even amplitudes result. Even transitions and branches between individual system elements of a conveyor are clearly characterized by corresponding signal sequences.

By using sensors with different detection depth, further features can be integrated into the signal sequences. The sensors preferably operate inductively, with resonant circuits, with high-frequency alternating electromagnetic fields or similar methods, but can also be designed as capacitive sensors, optical sensors, ultrasonic sensors and radioactively operating sensors.

The core idea of the invention is that typical signal sequences of the sensors moved with the transport object are classified, so that the conveyors or transitions between conveyors of a conveyor system can be assigned, and in the course of journeys of the transport object an expense-minimal, but sufficiently exact movement determination of the transport object possible becomes.

Data analysis systems can be used to classify the signal sequences according to the conveyors they represent. Data analysis systems are used to find general structures in data, such as signal sequences, time series and matrices, and to use the structural knowledge for evaluation. The data analysis is preferably subdivided into the areas of data acquisition, data preprocessing, feature extraction, feature selection, classifier formation and classification.

Data analysis systems, such. As fuzzy systems, artificial neural networks and their combinations in various forms (fuzzy C-Means, Kohonen networks, fuzzy-Kohonen networks, multilayer perceptrons, etc.), are able to signal sequences (time series) with their concise features and links to record, store and recognize relationships and structures in and between these.

Defined outputs of a data analysis system are activated as a function of the incoming signal sequences (approximation). The data analysis systems obtain their information by learning typical signal sequences and / or by specifying observed features and relationships (pulse widths, pulse width ratios, pulse trains, frequencies and amplitudes).

For unknown signals (series of measurements), data analysis systems are interpolatable, i. H. one or more outputs are activated according to the signal sequences applied to the inputs. Data analysis systems are very robust and error-resistant in decision-making, as they rely on typical and generalizing features. Individual untypical deviations in signal sequences are suppressed or ignored.

The method according to the invention for evaluating the signal sequences of sensors for position and position determination when moving transport objects on conveyors can be carried out with the aid of a position generator with data analysis. sesystemeen, in which typical signal sequences of a conveyor system with their concise features, links, relationships and structures are integrated, can be realized.

It is also part of the essence of the invention that, in addition to the use of the signal sequences, the procedure according to the detection of system element typical sensor signals, and to determine continuous position and position information, other signals can be detected and processed: Be with other sensors, which also are arranged in the transport object, but not to the underground of the conveyor track to sense typical system elements, such. B. measured data for acceleration, vibration, temperature of the transport object detected, so changes to a conveyor system can be determined in conjunction with the position data obtained according to the method. This can be used advantageously for locating disturbances, wear and abnormal operating data. This can be made possible by repeatedly performing control runs with a pattern transport object and by identifying deviations from known, stored, signal sequences from the position generator.

The method according to the invention can furthermore be used to detect structures of unknown complex conveyor systems. This is possible by comparison with patterns of the signal sequences of known conveyors stored in the position generator (straight roller conveyors, inclined roller conveyors, tapered roller conveyors, belt conveyors, curved conveyors, chain conveyors, plate conveyors, etc.).

The invention will be explained in more detail using an exemplary embodiment.

The accompanying drawings show in FIG Fig. 1: a scheme of a cargo handling system

2 shows a diagram of a transport object with sensors

3 shows a scheme for the integration of a position generator

4 shows a diagram of typical signal sequences of belt and roller conveyors

Fig. 5 is a diagram of the typical signal sequence at the transition between belt and roller conveyor

6 shows a diagram of the typical signal sequences of two sensors with different switching distances

Fig. 7: a diagram of a data analysis system for the identification of conveyors

8 is a diagram for defining the class affiliation of signal sequences

9 is a diagram of the defined and the measured signal sequences for a sensor

10 is a diagram of the assignment of a signal sequence in two defined classes and

11 shows tables for the class definition of signal sequences in the data analysis system for individual conveyors and for the fuzzy assignment of a measured signal sequence to the defined classes According to the diagram of FIG. 1, a piece goods conveying system may have belt conveyors GU 1 to GU 8 of different lengths, curve conveyors K 1 to K 5, straight roller conveyors GR 1 to GR 3, inclined roller conveyors SR, tapered roller conveyors KR, belt conveyors RF, side positioners SP, pushers PU and rotary tractors DA.

Gem. Fig. 2 sensors S1 to Sn are arranged on a transport object TO, the z. B. transmit a magnetic field and generally constructively existing system elements, such as idlers, deflection and drive drums or the roles shown in FIG. 2 R of a belt conveyor GU, contactless and wear-free capture.

3 shows a position generator PG as a data analysis device and its integration into the system of the method according to the invention for evaluating the signal sequences of sensors for position and position determination when moving transport objects on conveyors F:

Inputs of the position generator PG take over the signal sequences of the sensors S1 to Sn to be analyzed. In the interior of the position generator PG, the process steps claimed according to the invention take place so that information about the position Pos of the transport object TO such as location, identified conveyor Fn, position La of the transport object TO to the direction of movement (angle) and conveying speed v are available at the outputs of the position generator PG.

With the arrangement acc. Fig. 2 with their contactless and wear-free working sensors S1 to Sn are in calibration trips a patterned transport object TO by scanning all constructive assemblies of the conveyor system, such as rollers, frames, covers, diverters, points and similar devices, signal sequences s1 to sn added. accepted.

These signal sequences s1 to sn of all conveyors F, the transitions and branches which are recorded during the calibration runs are stored in the position generator PG as classes, also with blurred boundaries.

4 shows typical signal sequences of belt conveyors GU and roller conveyors. It is irrelevant whether the rollers R, as in a roller conveyor, recognized directly by the sensor field or the elements of a belt conveyor GU (drive and deflection drums, support rollers and support structures for the belt) are detected through the belt. The use of sensors with a hysteresis in the switching point is advantageous.

Decision criterion for the position determination can acc. To the signal sequence s1 of the roller conveyor. the lower figure of Fig. 4 in addition to the pulse widths and the frequency of the signal.

FIG. 5 shows the signal sequence s1 for a transition of a roller conveyor to a belt conveyor GU. In addition to the signal sequences s1 explained in FIG. 4, the transition with its air gap at the distance between the conveyors also provides information for the data analysis.

6 shows the use of sensors S1 and S2 with different depths of detection: both sensors S1 and S2 provide identical signal sequences s1 and s2 for the roller conveyor with its four rollers R shown. However, the sensor S1 with the lower detection depth is not in the Able to detect constructive elements of the conveyor through the belt - in the signal sequence s1 the impulse for the roller R of the belt conveyor GU is missing. By linking the signal sequences s1 and s2 further features for the data analysis are obtained. These additional features can also be through the use be formed by sensors with different physical principles of action.

Fig. 7 shows the scheme of a data analysis system in the form of a multilayer perceptron. On the left side of the scheme, the individual inputs of the network are assigned the sensors S1 to Sn. The weighted connections of the network and the links in the middle and the output layer contain the information for assigning signal sequences from the sensors S1 to Sn to the conveyors F1 to Fm. Also via the combination of several signal sequences s1; s2 of sensors S1; S2 to the conveyors according to FIG. 6, location information can be obtained.

8 illustrates the classical sharp class definition and the overlapping or fuzzy class assignment used in fuzzy systems - where zf is a mapping function. The fuzzy class assignment is also shown in FIG. 10.

The criteria of the classification for the interpretation of the signal sequences s1 to sn of a conveyor system are stored in the position generator PG. When moving transport objects TO, which are equipped with corresponding sensors S1 to Sn, their position and position can be continuously determined by the position generator PG on the basis of the recorded signal sequences s1 to sn without further options on the conveyor system. The fuzzy assignment of a measured signal sequence between two defined classes can be seen from the examples of FIGS. 9 and 10.

Another advantage of the method is the possible position and position determination even when not clearly identifiable signal sequences or signal sections, if z. B. several identical or similar conveyor in one System are present or faults occur. This is made possible by the fact that the position generator PG takes into account in the decision-making in advance clearly recognized conveyor F of the entire conveyor system.

FIG. 9 shows two typical signal sequences s1 of a sensor S1 when moving transport objects TO via two straight-roller conveyors GR in the two upper representations. As a function of the structural arrangement of the rollers R and the transport speeds v of the transport objects TO on the conveyors F, the signal sequences s1 are marked over time 1 or the path through certain pulse widths t11 to t25. The sequence of the pulse widths t11 to t25-starting with rising and falling edges-are typical features for the sensor signal of a conveyor F. The sensor S1 generating the signal sequence s1 in this example is designed to provide a binary output signal, ie , H. Metal is detected or no metallic background is present. The signal sequences s1 are recorded and stored during the calibration runs of a patterned transport object TO.

The typical, generalizing signal sequences s are defined as classes in the data analysis system, in this example by a fuzzy clustering with a fuzzy C-means, as shown in Table 1 of FIG. The number of impulses necessary for the class definition, in the example five, is very variable and depends primarily on the size of the conveyor system and the variety of conveyors F present.

The assignment of a measured, unknown signal sequence with the pulse widths T31 to T35 acc. FIG. 9 by the data analysis system in the position generator PG can be seen in FIG. 10 and in the table 2 of FIG. The measured signal sequence is added in the example with an affiliation of 0.9 Class A and a membership of 0.1 to Class B detected. The sum of all class assignments of a signal sequence is equal to one in fuzzy clustering with fuzzy C-Means. The degree of fuzziness of the class formation process, visible in the overlap of the class boundaries in FIG. 10, can be set variably. In Fig. 10, KZA and KZB are class centers A and B.

Slightly different signal sequences of defined classes can be safely assigned as in this example and signal sequences of different conveyors can be separated cleanly. The transport object TO is thus in this embodiment on the first roller conveyor, which belongs to the class A and is moved with a reduced by about 10% speed v compared to the calibration.

For the recognition of all conveyors F of the system in FIG. 1, the definition of at least ten classes in the data analysis system is required here. For signal sequences of large conveyor systems, fuzzy clustering automatically determines the number of existing classes in the recorded signal sequences. A subsequent reduction of the determined class number is possible and often technically sensible.

For the inventive method, it is irrelevant whether the processing of the signal sequences s1 to sn of the sensors S1 to Sn is carried out in the transport object TO, the signal sequences s1 to sn wirelessly transmitted to a base station and evaluated externally or stored signal sequences of z. B. a data logger after the test runs are evaluated. List of reference and formula signs for patent application

"Method for evaluating the signal sequences of sensors for position and

Orientation when moving transport objects on conveyors "

Dreschleschleuser DA

Conveyor F

Straight roller conveyor GR1 to GR3

Speed, conveying speed V

Belt conveyor GU; GU1 to GU8

Pulse widths t11 to t25, T31 to T35

Taper roller conveyor KR

Classes A; B

Class centers A; B KZA; KZB

Curve conveyor K1 to K5

Location La

Position pos

Position generator PG

Pusher PU

Belt conveyor RF

Inclined roller conveyor SR

Side positioner SP

Sensors S1 to Sn

Signal sequences s; s1 to sn

Transport object TO

Assignment function Zf

Claims

claims

1. Method for evaluating the signal sequences (s1 to sn) of sensors (S 1 to Sn) for position and position determination when moving transport objects (TO) on conveyors (F) of a conveyor system,

1.1. in which the signal foils (s1 to sn) of the sensors (S1 to Sn) detected during calibration runs of the transport objects (TO) are defined as classes (A; B) in accordance with typical signal sequence patterns,

1.2. the conveyors (F) of the conveyor system or transitions between conveyors (F) are assigned to typical classes (A; B),

1.3. the signal sequences (s1 to sn) of the sensors (S1 to Sn) of the moving transport object (TO) are continuously checked as to whether they can be assigned to a class (A; B) comprising a conveyor (F) or a transfer between conveyors (F ) and

1.4. In the case of a possible assignment to a class (A; B) for the moving transport object (TO), parameter and motion scenarios also recorded are assigned location-dependent.

2. The method according to claim 1,

2.1. in which one or more classes (A; B) are assigned to the conveyor (F) or transitions between conveyors (F).

3. The method according to claim 1,

3.1. in the typical signal sequence pattern using technical parameters of the conveyor (F) are created as a comparison pattern.

For this 7 pages drawings.