JPH0552786A - Flaw inspection device - Google Patents

Abstract

PURPOSE:To enable a flaw of an object to be inspected to be judged readily and properly based on a constant evaluation standard, a machining control equipment to be specified based on the flaw judgment result, and inspection, repairing, and replacement of that equipment to be instructed properly. CONSTITUTION:The title flaw detection device allows a machining control equipment which causes a flaw to be generated to be specified based on a plurality of output signals from a flaw detection device 3. A flaw judgment mechanism 5 which judges presence or absence of the flaw of an object to be inspected based on a plurality of output signals from the flaw detection device : and the type of the flaw if there is the flaw, a flaw judgment result output mechanism 6 which instructs inspection, repairing, and replacement of a machining control equipment which causes the flaw to be generated based on the flaw judgment result which is obtained by this flaw judgment mechanism 5 or its constitution member are provided. Further, a classification instruction mechanism 7 for instructing distribution of the object to be inspected based on the flaw judgment result which is obtained by the flaw judgment mechanism 5 is provided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is used in a production line for producing an object to be inspected by using a rolling system for producing a rolled steel sheet by using a plurality of rolling mills and other various processing control devices. In connection with the flaw inspection device, in particular, according to the state of flaws such as steel plate or the like to be inspected manufactured by a plurality of processing control equipment, find a specific processing control equipment causing the flaw, and inspect, repair, The present invention relates to a flaw inspection device having a function of giving instructions such as replacement.

[0002]

2. Description of the Related Art In the conventional general flaw inspection method, an inspection worker judges the presence or absence of a flaw and the type of the flaw by visually observing the surface of each subject to be inspected, or The presence or absence of flaws on the object to be inspected and the type of flaws are displayed by displaying the flaw detection signal from the flaw detection device that optically outputs the surface state after converting it into an electrical signal and outputs it. It is determined that the inspection target is to be distributed, and the inspection target object to be inspected is instructed to be inspected, and the inspection, repair, and replacement of a specific processing control device among the plurality of processing control devices that causes a flaw are instructed.

[0003]

However, in the above-described flaw inspection method, it is difficult for an inspection operator having considerable skill to accurately determine the presence or absence of a flaw, let alone identify the type of flaw. Specify the processing control device, and
It is not easy to give instructions for inspection, repair, and replacement.

Moreover, since the experience and intuition of the inspector are largely used, different judgment results may be obtained from time to time, and the judgment result based on a certain evaluation standard cannot be obtained, and the object to be inspected is not obtained. There will be variations in the instructions for sorting, inspection, repair and replacement of specific processing control equipment.

The present invention has been made in view of the above circumstances, and it is possible to quickly and accurately determine a flaw in an object to be inspected on the basis of a certain evaluation standard, and easily perform a machining control device or its configuration based on the flaw determination result. Identifying parts, and inspecting, repairing,
It is an object of the present invention to provide a flaw inspection device that can reliably instruct replacement, can quickly sort the inspection object based on the flaw determination result, and can proceed the work efficiently.

[0006]

First, in order to solve the above-mentioned problems, the invention corresponding to claim 1 detects a plurality of flaws in a state of an object to be inspected manufactured by at least one processing control device. Detected by a flaw detection device consisting of a vessel, in the flaw inspection device that identifies the processing control device or its constituent members that are the cause of the flaw based on a plurality of flaw detection signals from the flaw detection device,

Based on a plurality of flaw detection signals of the flaw detection device, a flaw determination mechanism is provided for determining the presence or absence of a flaw in the object to be inspected and the type of the flaw if there is a flaw, and the flaw determination mechanism determines the flaw type. A defect determination result output mechanism that outputs an instruction to inspect, repair, or replace the processing control device or its constituent members that are the cause of the defect based on the defect determination result. This is a configuration in which a sorting instruction mechanism is provided that issues an instruction to sort the inspection target objects based on the flaw determination result.

The invention according to claims 2 and 3 embodies the flaw determining mechanism. In the invention according to claim 2 thereof, a skilled person in advance has a combination pattern corresponding to a plurality of types of flaws. Is provided with a storage means for storing, and a combination pattern composed of a plurality of flaw detection signals of the flaw detection device and a combination pattern corresponding to a plurality of types of flaws stored in advance in the storage means are compared, and the highest similarity is obtained. A defect pattern recognition means for outputting a defect determination result corresponding to a large stored combination pattern is provided.

Further, in the invention according to claim 3, it comprises a neurocomputer having an input layer, an intermediate layer and an output layer, and inputs a simulated or actual data pattern corresponding to a flaw to the input layer and A new pattern determining means for determining an optimum weighting coefficient based on the output of the output layer of the neurocomputer and the teacher data pattern corresponding to the flaw is provided, and the weighting coefficient determined by the new pattern determining means and the simulated or actual The storage means for storing the data pattern as the comparison combination pattern corresponding to the flaw and the compared combination patterns of the plurality of flaw detection signals of the flaw detection device and the plurality of comparison combination patterns extracted from the storage means are input. Layers to be compared and combined patterns to be compared obtained from the output layer of the neurocomputer It is provided with a flaw pattern recognition means for outputting the flaw determination result corresponding to the larger the comparative combination patterns of output and comparing the most similar of the output and a plurality of comparative combination pattern.

[0010]

Therefore, in the invention according to claims 1 and 2, by taking the above-mentioned means, the skilled person provides a storage means for storing a combination pattern corresponding to a plurality of kinds of flaws in advance, and detects the flaw. The presence / absence of a flaw in the object to be inspected and the type of flaw are determined from the similarity between the combination pattern consisting of a plurality of flaw detection signals of the device and the combination pattern corresponding to a plurality of types of flaws stored in advance in the storage means. , Based on this determination result, a certain machining control device or its constituent members that are associated in advance with a certain type of flaw are specified, and inspection, repair,
Since an instruction to replace the product is output, the inspector or maintenance person can inspect the processing control equipment or component parts according to the instruction.
It suffices to perform repairs and the like, and even an unskilled person can specify the type of flaw based on a certain evaluation standard and can give an appropriate instruction.

Further, in the invention according to claim 3, the simulated or actual data pattern corresponding to a certain flaw is input to the input layer of the neurocomputer, while the output of the output layer at that time and the teacher data corresponding to the flaw. The optimum weighting factor is determined based on the pattern and the combination pattern for comparison corresponding to the flaw based on the simulated or actual data pattern and the determined weighting factor is stored. A combination pattern for comparison can be added or modified, and a combination pattern for comparison of a plurality of defect detection signals of the defect detection device and a plurality of comparison combination patterns extracted from the storage means are input to the input layer, Output of the compared combination pattern obtained from the output layer of the neural network and the plurality of comparison combination patterns Since the defect determination result corresponding to the comparison combination pattern having the highest similarity is compared with the output, if the comparison combination pattern and the cause of the defect are associated in advance, the cause of the defect occurrence will be accurate. Understand and give appropriate instructions to inspectors or maintenance personnel.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the present invention. In the figure, reference numeral 1 denotes an object to be inspected during or after manufacture, and the object to be inspected 1 is a plurality of processing control devices 2
It shall be manufactured by the control device 2 composed of ₁ , 2 ₂ , .... Denoted at 3 is a flaw detection device comprising a plurality of flaw detectors 3 ₁ , 3 ₂ , ... Which detect the surface state of the inspection object 1. The flaw detection signals from these flaw detectors 3 ₁ , 3 ₂ ,. Is output and sent to the flaw inspection device 4.

The flaw inspection apparatus 4 includes a flaw determination mechanism 5 for determining a flaw based on a plurality of flaw detection signals, and a type of flaw when there is a flaw from the flaw determination result determined by the flaw determination mechanism 5. And a defect determination result output mechanism 6 for inspecting, inspecting, repairing, replacing, etc., the processing control device that has caused the defect, that is, notifying the inspector of the processing control device or control signal name, and the defect determination. Based on the flaw determination result determined by the mechanism 5, there is no flaw, there is a flaw, and a sorting instruction mechanism 7 for instructing the distribution of the destination of the inspected object 1 according to the type of flaw.

Next, FIG. 2 is a diagram showing an example of the structure in which the flaw inspection apparatus according to the present invention is applied to a tandem cold rolling system. In this tandem cold rolling system, a rolling mill 13a in which a coil as an object to be inspected 1 unwound from a coil box 11 is controlled by rolling mill motors 12a to 12e which are driving devices, such as a reduction amount, speed, and tension. ~
After being rolled by 13e to a target plate thickness, it is wound on a take-up reel 14. 15 is a take-up reel 1
4 is a take-up reel motor that controls No. 4.

The coil after winding the reel is taken out from the winding reel 14 and conveyed to the next step such as an inspection line, a weighing line and a binding line. At this time, flaws in the coil width direction are continuously formed. sends the series flaws detection signal when the installed plurality of flaw detectors 3 ₁ to 3 ₅ surface flaws of the respective rolling in the coil from to detect the flaw inspection device 4.

In this flaw inspection apparatus 4, the flaw determination mechanism 5 performs flaw determination, and the determination result is sent to the flaw determination result output mechanism 6 and the classification instruction mechanism 7, and the classification instruction mechanism 7 is sent.
No defect signal and defect type signal from the coil sorting device 1
Sent to 6. Therefore, the coil sorting device 16 sorts out the coils according to the type of flaws and the flaws, and carries them out to a predetermined place.

On the other hand, the defect determination result output mechanism 6 notifies the inspector of the defect determination result from the defect determination mechanism 5, and
In the case of a flaw determination result with a flaw, the processing control device 2 ₁ or 2 ₂ , 2 ₃ , ... Of the control device 2 requiring inspection, repair, or replacement is specified from the type of the flaw, or the control signal name is specified. Notify the inspector. In the case of the present embodiment, for example, in the case of the defect type A, it is necessary to replace the rolls of the rolling mill 13a.
When the defect type is B, the inspector is notified that the rolling mill motor 12b that drives the rolling mill 13b needs to be inspected.

Next, the classification instructing mechanism 7 links the coil based on the flaw judgment result from the flaw judging mechanism 5, whether there is a flaw, there is a flaw, and if there is a flaw, the inspected object is in accordance with the kind of the flaw. An instruction to distribute the destination of the coil of the object 1 is output. In the case of the present embodiment, for example, if the coil has no flaw, the route 1 is designated as the destination for carrying to the next step, and if the coil is the type A flaw, the route 2 is designated as the destination for temporary placement in the stockyard.

Next, FIG. 3 is a diagram showing a detailed structure of the flaw determination mechanism 5. The flaw determination mechanism 5 is a setting mechanism for storing a plurality of combination patterns in a storage mechanism described later by an expert, and a change / correction for adding, deleting, or changing the combination pattern already stored in the storage mechanism by the setting mechanism. A mechanism for instructing maintenance personnel on a desired output pattern for the mechanism and the defect determination result, that is, a teacher data input mechanism 51 having a setting, correction, and learning instruction function, and a plurality of defect detections sent from the defect detection device 3. A storage mechanism 52 that stores the latest new combined pattern by inputting signals in time series, and a combination pattern that is set by the setting mechanism, and a combination pattern that is input from the flaw detection device 3. The plurality of combination patterns of the storage mechanism 52 set by the setting function of the teacher data input mechanism 51 are compared and already stored. Scratch pattern recognition mechanism 5 for determining the degree of similarity flaws with combination patterns are
3 and 3.

The flaw pattern recognition mechanism 53 uses a neurocomputer as shown in FIG. 4, and specifically, an input layer 531 having a plurality of input elements 531 _{1 to} 531 _n (not shown) and the input element 531. _{1 to} 531 _n and a plurality of intermediate elements 532 ₁ to be combined with appropriate weighting factors
An intermediate layer 532 having 532 _n (not shown) and an output layer 533 having output elements 533 _{1 to} 533 _n coupled to the intermediate elements 532 _{1 to} 532 _n with appropriate weighting factors.
And a new pattern determining means for creating new and changed combination patterns, and a flaw pattern recognizing means for making a flaw determination based on the flaw detection signal from the flaw detection device 3.

This new pattern determining means inputs a teacher data pattern corresponding to a certain flaw from the teacher data input mechanism 51, while storing a simulated or actual data pattern corresponding to the flaw from the flaw detecting device 3 in the storage mechanism 52 and. It is input to the input layer of the neuro computer, and the learning mechanism 511 is used to sequentially change the weighting factors to determine the optimum weighting factor. Then, the optimum weighting coefficient is stored in the storage mechanism 52 in correspondence with the simulated or actual data pattern to form a new combination pattern. Such a series of processing is performed on the required types of flaws and the combination pattern corresponding to each flaw is stored in the storage mechanism 52.

On the other hand, the flaw pattern recognition means is a combination pattern corresponding to a plurality of flaws stored in the storage mechanism 52 and a combination pattern consisting of a plurality of flaw detection signals actually sent from the flaw detection device 3. The pattern is input to the input layer of the neuro computer, both outputs obtained from the output layer 533 are compared while applying the same weighting coefficient, and the combination pattern stored in the storage mechanism 52 having the highest degree of similarity is determined to be the defect pattern. To do. Although there is only one intermediate layer 532 in FIG. 4, a plurality of intermediate layers can be provided if necessary.

Next, the operation of the apparatus configured as described above will be described. First, a skilled person inputs a combination pattern of flaws with the inspection object 1 expected in advance to the teacher data input mechanism 51, while a device corresponding to the flaw detection device 3 performs a simulated inspection target having the same type of flaw. A combination pattern from an object or an actual inspected object or a combination pattern of defects of the same type created from past experience without the simulated inspection object is input to the storage mechanism 52 and the defect pattern recognition mechanism 53. And the learning mechanism 51
1 so as to obtain the optimum output from the output layer 533 of the flaw pattern recognition mechanism 53.
The learning is performed while changing the weighting factor of 2-533, and the weighting factor which is the optimum output value is stored in the storage mechanism 52 in association with the combination pattern.

The series of processes as described above are combined into patterns for each defect and stored in the storage mechanism 52. It should be noted that each of these defects is associated with each of the processing control devices 2 ₁ and 2 that constitute the control device 2.
_2. It is assumed that it is understood in advance that it will occur based on the abnormalities of ₂ , ... and other causes.

After the combination patterns of a plurality of types of flaws are stored in the storage mechanism 52 as described above, the inspected object 1 is manufactured by using each processing control device 2 ₁ , 2 ₂ , ... Of the control device 2. but the inspected object 1 in the manufacturing process the inspected object 1 flaw state a plurality of flaw detectors 3 _1, 3 _2, detected by ..., these flaws detectors 3 _1, 3 _2, ... more The combined pattern composed of the flaw detection signals of No. 2 is input to the flaw pattern recognition mechanism 53 of the flaw determination mechanism 5. At this time, the combination pattern corresponding to each defect is sequentially input from the storage mechanism 52 to the defect pattern recognition mechanism 53, and a weighting factor corresponding to this combination pattern is applied.

Then, in the output layer 533 of this flaw pattern recognition mechanism 53, the output and storage mechanism 52 relating to the combination pattern composed of a plurality of flaw detection signals of the flaw detection device 3 are described.
From the output related to a plurality of combination patterns sequentially read from, from the similarity of those patterns, for example, if there is no flaw, if there is a flaw, the flaw pattern (the type of flaw) with the highest similarity is determined, and the determination is made. The result is sent to the flaw determination result output mechanism 6 and the classification instruction mechanism 7.

Therefore, in the flaw determination result output mechanism 6,
When the flaw determination result indicating the type of the flaw is received from the flaw determination mechanism 5 when there is no flaw or the flaw is present, as shown in FIG. 5, the flaw determination result and the inspected object 1 are linked (the inspected object). After the flaw marking in 1) is performed (ST1), a linking process with the treatment method for the flaw type is performed (ST2).
Further, the defect determination result and the treatment method according to the type of the defect are displayed on the output device to notify the inspector or maintenance staff (ST.
3).

Therefore, if the inspector or maintenance person needs to inspect, repair, or replace the processing control equipment, such as the motor and rolling mill, from the displayed contents, the inspector or the maintenance staff will follow the instructions. Inspect or repair,
Or perform replacement work.

On the other hand, in the sorting instruction mechanism 7, the flaw determination mechanism 5
When the flaw determination result indicating the type of the flaw is received in the case where there is no flaw or the flaw exists, after the flaw determination result and the inspected object 1 are linked as shown in FIG. 6 (ST11), The process of associating the flaw determination result with the destination of the inspection object 1 is performed (ST12). Then, the destination instruction signal obtained in step ST12 is sent to the sorting device 8. Here, in the case of a defective coil, the sorting device 8 further conveys it to a downstream process based on the destination instruction signal,
In the case of a coil with a flaw, it can be transported to a temporary storage place by classifying it according to the flaw type.

Therefore, according to the configuration of the above embodiment, if the knowledge and experience of the skilled person is utilized in advance, the combination pattern of the weighting factors corresponding to various defects is stored by using the neuro computer. Based on the similarity between the combination pattern composed of a plurality of defect detection signals of the defect detection device 3 and the stored plurality of combination patterns, the defect can be comprehensively and accurately determined under a certain evaluation standard. Further, since the pattern recognition method using the neuro computer is adopted, it is possible to perform the flaw inspection promptly from the input of the flaw detection signal to the output of the flaw determination without the need for programming as in the past.

If a neuro computer is used,
It is possible to easily modify, add, and change the defect determination result while making the most of the learning function, and to enhance the expandability. Moreover, the defect determination result and the processing control device 2 ₁ ,
If the identification of 2 ₂ , ... or its constituent members and the cause of the flaw are clarified, the machining control equipment 2 ₁ , 2 ₂ , ... or its constituent members can be easily identified, inspected, from the flaw judgment result.
Instructions for repair and replacement can be given, and eventually the source of defects can be removed accurately and promptly.

Further, when the flaw determination result is that there is no flaw or there is a flaw, the type of flaw is shown. Therefore, based on this flaw determination result, classification is made for each inspected object 1 of the same flaw type and the same standard. Therefore, the sorting work can be systematically and quickly performed.

In the above embodiment, the steel sheet rolling equipment is taken as an example, but it can be easily applied to various processing / manufacturing equipment. Further, although the comparison combination pattern is created using the neurocomputer and stored in the storage mechanism 52, the combination pattern is directly stored in the storage mechanism 52 based on past knowledge and experience without using the neurocomputer.
The pattern comparison may be performed by storing. Besides, the present invention can be variously modified and implemented without departing from the scope of the invention.

[0034]

As described above, according to the present invention, a flaw of an object to be inspected can be promptly and accurately determined based on a certain evaluation standard, and a processing control device or its configuration can be easily obtained from the flaw determination result. The member can be specified, and inspection, repair, and replacement can be instructed reliably. Furthermore, the inspected object can be quickly sorted based on the defect determination result, and the work can be efficiently performed.

[Brief description of drawings]

FIG. 1 is a functional block diagram of a flaw inspection apparatus according to the present invention.

FIG. 2 is a diagram in which the device of the present invention is applied to a tandem cold rolling system.

FIG. 3 is a configuration diagram of the flaw determination mechanism of FIG.

FIG. 4 is a specific configuration diagram of the flaw determination mechanism of FIG.

FIG. 5 is a diagram for explaining the operation of the flaw determination result output mechanism of FIG.

FIG. 6 is a diagram for explaining the operation of the sorting instruction mechanism of FIG.

[Explanation of symbols]

1 ... Object to be inspected, 2 ... Control device, 3 ... Defect detection device, 4
Defect inspection device, 5 Defect determination mechanism, 6 Defect determination result output mechanism, 7 Classification instruction mechanism, 51 Teaching data input mechanism, 5
2 ... memory mechanism, 53 ... flaw pattern recognition mechanism, 511 ... learning mechanism, 531 ... input layer, 532 ... intermediate layer, 533 ... output layer.

Claims (3)

[Claims] 1. A flaw detection device including a plurality of flaw detectors detects a flaw state of an object to be inspected manufactured by at least one or more processing control devices, and outputs a plurality of output signals from the flaw detection device. In a flaw inspection device that specifies the processing control device or its constituent members that are the cause of the flaw based on the above, in the presence or absence of the flaw and the flaw of the inspection object based on a plurality of output signals from the flaw detection device. In the case of, a flaw determination mechanism for determining the type of flaw, and the processing control device or its constituent member that is the cause of the flaw based on the flaw determination result determined by this flaw determination mechanism, and A defect determination result output mechanism for instructing inspection, repair, and replacement of the device and member, and a sorting instruction mechanism for instructing the distribution of the inspected objects based on the defect determination result determined by the defect determination mechanism. Flaw inspection apparatus characterized by comprising a. 2. A defect determination mechanism, a storage means for storing in advance a combination pattern for comparison corresponding to a plurality of types of defects, a combination pattern for comparison of a plurality of output signals from the defect detection device, and the plurality of types. The flaw inspection apparatus according to claim 1, further comprising: a flaw pattern recognition unit that compares a flaw combination comparison pattern corresponding to a flaw and outputs a flaw determination result corresponding to the comparison combination pattern having the highest degree of similarity. 3. The flaw determination mechanism is composed of a neuro computer having an input layer, an intermediate layer and an output layer, and inputs a simulated or actual data pattern corresponding to a flaw to the input layer and outputs the neuro computer. New pattern determining means for determining an optimum weighting coefficient based on the layer output and the teacher data pattern corresponding to the defect, and the weighting coefficient and the simulated or actual data pattern determined by the new pattern determining means for the defect. Storage means for storing as a comparison combination pattern corresponding to, a comparison combination pattern of a plurality of output signals from the flaw detection device, and a plurality of comparison combination patterns extracted from the storage means are input to the input layer. The output of the compared combination pattern obtained from the output layer of the neurocomputer and the output of the composite pattern. The flaw inspection apparatus according to claim 1, further comprising: a flaw pattern recognition unit that compares the output of a plurality of comparison combination patterns and outputs a flaw determination result corresponding to the comparison combination pattern having the highest similarity.