JP6016767B2 - Method executed by operation data search support device and operation data search support device - Google Patents

Description

The present invention provides an operation data search support device that additionally shows a relationship with another search performed in the past as a search condition when searching for operation data based on an identifier of a product, and supports judgment on a search result. Relates to a method and an apparatus to be executed .

  In the production line of products manufactured through multiple manufacturing processes, if there is a problem with the quality of the process in the process downstream from its own process, the cause of the defect is investigated upstream from the process in which the problem occurred. There are things that must be done. In such a case, it is necessary to search the operation data of the product with the problem from the enormous amount of operation data recorded in its own process, and it takes a lot of time to investigate and search for the cause of the defect. Become. In particular, depending on the manufacturing site, the operation data may not be digitized. In such a case, it is necessary to perform troublesome work such as searching the record book to find the data. It will be very big.

  Therefore, a method has been developed that searches operation data to investigate the cause of defects. For example, Patent Document 1 discloses a technique for automatically acquiring operation knowledge from operation data accumulated in the past. In this technique, all similarities between operation variables obtained by randomly searching stored operation data are extracted, and operation knowledge is created using the extracted similarities. If the operation knowledge created in this way is used, the cause of the defect can be investigated.

JP-A-5-88896

  By the way, the method of Patent Document 1 requires preparations such as classifying operation variables in advance before searching for operation data. However, such a cluster of operational variables requires knowledge of experienced operators. Naturally, there is a possibility that an operator who is young and inexperienced cannot perform appropriate clustering, and may not be able to satisfactorily investigate the cause of failure without obtaining proper operation knowledge. In addition, when operating conditions are changed or new products are manufactured, it may not be possible to make sufficient preparations based on experience, and it may be difficult to investigate the cause of defects as well. is there.

In other words, the method of Patent Document 1 may be beneficial in investigating operation data if the above-described preparation can be performed reliably, but in practice, experience and knowledge are required for preparation. For this reason, it has become a problem when used at manufacturing sites.
The present invention has been made in view of the above-mentioned problems, and it is not necessary for operators to perform complicated operations such as class division themselves, and simple operations such as searching for operation data by specifying an identifier are repeated. It is an object of the present invention to provide an operation data search support method and an operation data search support device that can automatically obtain search conditions that support judgment on search results, and that can perform a search simply and efficiently. To do.

In order to solve the above problems, the present invention takes the following technical means.
That is, a method of operating the data retrieval support apparatus of the present invention is executed is stored operational data bundled with the product identifier and the identifier produced by the production equipment, product identifier is input to search is input If the operation data corresponding to the identifier does not satisfy any of the search conditions registered in the past, the operation data of the input identifier has been searched in the past and has been registered. and operational data of other identifiers neither search does not satisfy the, is output to select the common point of data between two or more operational data between which singled out is extracted as a feature, extracted feature search is converted to the conditions proposed, in addition to the operational data of the input identifier, the converted retrieval condition proposed is characterized Rukoto appears.

Incidentally, preferably, may the displayed search criteria proposal whether registered Luca is possible to select as the search criteria.
Incidentally, it is preferred that, when the operational data of the product currently being manufactured satisfies any one of the registered search condition, the search condition the satisfactory still appears.
Incidentally, it is preferred that, between two or more operational data picked the common features are extracted Runisaishite and clustered using the technique of the decision tree learning cracking line.

In addition, the operation data search support device of the present invention includes an operation data storage unit that stores an identifier of a product manufactured in a manufacturing facility and operation data associated with the identifier, and an input unit that inputs an identifier of a product to be searched. And if the operation data corresponding to the identifier input to the input unit does not satisfy any of the search conditions registered in the past, the operation data of the identifier input to the input unit and the past search And selecting the operation data of another identifier that does not satisfy any of the registered search conditions, and features the common point of data between the selected two or more operation data A feature extraction unit to extract, a search condition plan creation unit that converts the feature extracted by the feature extraction unit into a search condition plan, and an operation data storage unit that stores the operation data of the identifier input to the input unit And displays call al, characterized in that it comprises a display unit for displaying a search condition draft created by the search condition plan creation section as a new search condition.

Preferably, a registration means for registering the search condition plan displayed on the display unit as a new search condition is provided.
Preferably, the display unit is configured to display the satisfied search condition when operation data of a currently manufactured product satisfies any of the search conditions registered by the registration unit. It is good to have.

  Preferably, the feature extraction unit is configured to perform clustering using a decision tree learning technique when extracting common features between the two or more selected operation data. .

According to the method and the operation data search support device executed by the operation data search support device of the present invention, it is not necessary for the operator to perform a complicated operation such as class classification, and the operation data is searched by designating an identifier. By simply repeating such simple operations as described above, it is possible to automatically obtain the search conditions that support the judgment on the search results, and the search can be performed easily and efficiently.

It is the figure which showed the structure of the operation data search assistance apparatus of this embodiment. It is a figure which shows the screen of a display part when operation data is registered when it searches by inputting an identifier. It is a figure which shows the screen of a display part when operation data is not registered when it searches by inputting an identifier. It is the figure which showed the flow of the signal processing in an operation data search assistance apparatus. It is the figure which showed STEP4 of FIG. 4 in detail. It is a figure which shows the example of the lot number memorize | stored in the operation data memory | storage part, and operation data. It is a figure which shows the registered search conditions. It is the data memorize | stored in the lot number memory | storage part, Comprising: It is a figure which shows the tied condition of a lot number and a search condition. It is the figure which showed the operation data clustered by the feature extraction part. It is the figure which showed the procedure of the clustering performed in a feature extraction part. It is a figure which shows the data of the lot number memory | storage part when operation data is newly added. It is a figure which shows the screen of the display part which displays the operation condition under manufacture in manufacturing equipment, and the search conditions which this operation data is satisfied.

Hereinafter, a method executed by the operation data search support device of the present invention , that is, an embodiment of the operation data search support method and device will be described in detail with reference to the drawings.
FIG. 1 schematically shows a configuration of an operation data search support apparatus 1 according to the present invention.
As shown in FIG. 1, the operation data search support device 1 of the present embodiment is such that an operator (searcher S) of a manufacturing facility P in which the operation data search support device 1 is provided has some quality defect in a lower process than himself. When it occurs, it is used to search operation data in which a quality defect has occurred from a huge record of operation data, and to investigate the cause of the quality defect from the retrieved operation data.

  Specifically, when the lot number (identifier) of a product to be searched for is input using the input unit 2 such as a mouse, the operation data search support apparatus 1 of the present invention inputs operation data associated with the lot number. Is displayed on the display unit 3 as well as a search condition that is beneficial when investigating the cause of the quality defect. The operation data search support device 1 makes it easy to find the cause of the quality defect by using the search conditions displayed in this way.

Such a search is actually performed using a computer having a monitor and a keyboard.
For example, a lot number is input into the input field 10 at the upper left on the monitor screen of the computer illustrated in the upper part of FIG. Click to perform a search. The result of the search performed in this way is as illustrated in the lower side of FIG. That is, the operation data corresponding to the lot number is called out from the operation data stored in advance in a storage device such as a hard disk drive (HDD) of the computer and displayed on the monitor screen. At this time, not only the operation data but also the above-described search conditions are displayed on the monitor screen.

  On the other hand, as shown in FIG. 3, when the lot number (operation data linked to the lot number) input by the operator does not satisfy any search condition, in other words, a new quality defect that has not been searched until now. When there is no search condition, the above-mentioned “operation data obtained by another search” does not exist, and therefore the search condition is not displayed. In such a case, “suggested search conditions” as illustrated on the monitor screen of FIG. 3 is displayed.

  This “search condition plan” is similar to the above-described search condition, and includes operation data associated with the input lot number and operation data that does not satisfy any search condition in the searched operation data. It is a search condition candidate created in between. If the operator who operates the operation data search support device 1 recognizes similarity between these operation data, in other words, if these operation data are considered to belong to the same defect, the search condition proposal is searched. By officially registering as a condition, it is possible to identify by the registered search condition when the next similar search is performed.

As described above, the operation data search support device 1 according to the present invention for searching is specifically configured by combining the following parts.
That is, as shown in FIG. 1, the operation data search support device 1 has an input unit 2 for inputting a lot number of a product to be searched. The input unit 2 corresponds to a computer keyboard, mouse, microphone, and the like.

Further, the operation data search support device 1 has been searched in the past with an operation data storage unit 4 that stores a lot number (identifier) of a product manufactured in the manufacturing facility P and operation data associated with the lot number. A lot number storage unit 5 (identifier storage unit) that stores the lot number of the product and a search condition storage unit 6 that stores search conditions registered in the past.
Furthermore, in the operation data search support device 1, if the operation data corresponding to the lot number input to the input unit 2 described above does not satisfy any of the search conditions registered in the past, the operation data search support device 1 is input to the input unit 2. The selected lot number operation data and the operation data of other lot numbers that have been searched in the past and do not satisfy any of the registered search conditions are selected, and two or more selected A feature extraction unit is provided that extracts the common points of the data among the operation data as features.

In addition, the operation data search support device 1 is provided with a search condition plan creation unit that converts the feature extracted by the feature extraction unit into a search condition plan.
The operation data storage unit 4, the lot number storage unit 5, the search condition storage unit 6, the feature extraction unit, and the search condition plan creation unit are realized by a program stored in the main body unit 12 of the computer.

  Further, the operation data search support apparatus 1 is provided with a display unit 3 for displaying a new search condition created by the search condition plan creation unit as a search condition plan. The display unit 3 corresponds to a computer monitor. The monitor screen, which is the display unit 3, is provided with a registration button 9 (registration means) for registering the search condition plan displayed on the display unit 3 as a new search condition. The search conditions are sequentially stored in the search condition storage unit 6 described above.

Next, each member which comprises the operation data search assistance apparatus 1 of this invention is demonstrated.
First, the manufacturing facility P provided with the operation data search support device 1 of the present invention will be described. The manufacturing equipment P in which the operation data search support device 1 is provided includes various equipment such as equipment related to metal production such as rolling, extrusion, forging, and casting, and equipment related to machining such as bending and drawing. Hereinafter, the operation data search support device 1 of the present embodiment will be described by giving an example of what is provided in a steel plate thick plate facility (hot rolling facility).

  The input unit 2 is a part for an operator (searcher S) to input a lot number of a product to be searched into the operation data search support apparatus 1, and is composed of the above-described computer keyboard and mouse. Specifically, the product lot number is input to the lot number input field 10 provided on the monitor screen using the input unit 2 and the search button 11 provided on the same monitor screen is clicked. The input lot number is sent to the operation data storage unit 4, and the operation data associated with the input lot number is output from the operation data storage unit 4.

The operation data storage unit 4 is a part that stores a lot number of a product and operation data associated with the lot number.
The operation data stored in the operation data storage unit 4 includes data indicating the operating state sent from the manufacturing equipment P described above, and the product name, destination, and product dimensions directly input via the input unit 2. Product data such as product material or tempering number. The operation data illustrated in FIG. 6 includes items such as “product width”, “product thickness”, “component composition classification”, and “heating temperature error”.

These operation data are stored for each lot number in the operation data storage unit 4 described above.
The search condition storage unit 6 is a part for storing search conditions registered in the past by the searcher S. FIG. 7 shows an example of the search condition registered in the search condition storage unit 6.
The search condition storage unit 6 stores all the search conditions registered in the past when the searcher S presses the registration button 9 on the monitor screen. One search condition ID is assigned to each of these search conditions, and the search condition ID and the search condition are stored in a one-to-one data state.

As shown in FIG. 8, the lot number storage unit 5 stores a combination of a lot number and the above-described search condition ID, and a lot number that has been searched in the past and a search obtained by the search. All condition IDs are stored.
For example, looking at the products with lot numbers “1” and “25” in FIG. 8, it can be seen that a search has already been performed in the past. Then, it can be seen that the operation data obtained by this past search satisfies the search condition registered with the search condition ID of “1” in the past. Further, looking at the products having the lot numbers “9”, “13”, and “16”, it can be seen that the search has been performed in the past and the search condition ID of “2” is assigned. However, even if a product has been searched in the past, if none of the registered search conditions is satisfied, there is no product corresponding to the registered search condition. In this case, the search condition ID is “−1”, and it can be determined that the search condition is not registered. That is, the lot number storage unit 5 stores a lot number of a product searched in the past and a registration state of a search condition for the product.

As described above, when the operation data output from the operation data storage unit 4 corresponding to the lot number input to the input unit 2 does not satisfy any of the search conditions registered in the search condition storage unit 6 in the past. In the operation data search support device 1, a new search condition plan is created.
In other words, the operation data search support device 1 of the present invention is configured to extract features common to operation data that are necessary when creating a new search condition plan. Specifically, this feature extraction is performed by a feature extraction unit. The feature extraction unit includes operation data of the lot number input to the input unit 2, operation data of other lot numbers that have been searched in the past and that do not satisfy any of the registered search conditions, Is selected, and a common point of data is extracted as a feature between two or more selected operation data.

In this feature extraction unit, data is processed according to the flow shown in FIGS. 4 and 5, and common points of operation data are extracted as features (details of the processing will be described later).
As a result, when the operation data output from the operation data storage unit 4 satisfies the search condition registered in the search condition storage unit 6, the search condition satisfied by the operation data is displayed as “hit search condition”. Part 3 is displayed.

  On the other hand, if the operation data output from the operation data storage unit 4 does not satisfy any of the search conditions registered in the search condition storage unit 6, operation data that does not satisfy any of the search conditions registered in the past, In other words, since it can be determined that a new quality defect that has never been searched for has been searched, a “search condition plan” is created. That is, in the feature extraction unit, all lot numbers having the search condition ID “−1” are selected from the lot numbers stored in the lot number storage unit 5. The lot number with the search condition ID “−1” corresponds to a product that has been searched in the past but does not satisfy any of the registered search conditions.

When the lot number having the search condition ID “−1” is selected in this way, the operation data corresponding to the selected lot number is output from the operation data storage unit 4 and input (designated) in the current search. The operation data corresponding to the lot number) is output from the operation data storage unit 4. And all the output operation data are distributed to A group (Gr.A).
On the other hand, all the lot numbers stored in the lot number storage unit 5 described above have been searched in the past. In other words, among the operation data stored in the operation data storage unit 4, the data for which the lot number is not stored in the lot number storage unit 5 has not been searched in the past, and there is no quality defect. It can be considered as a “normal product”. Therefore, in the feature extraction unit, among the operation data stored in the operation data storage unit 4, data whose lot number is not stored in the lot number storage unit 5 is extracted, and all the extracted operation data is stored in the B group (Gr .B).

  When two groups of operation data are obtained in this way, rules for classifying (separating) both data groups, specifically, variables (operation variables) constituting the operation data and their threshold values are extracted. For this extraction, an information amount (entropy) calculation is used, and a technique such as “C” used for creating a decision tree or the like can be used. That is, if the operation data is divided into group A and group B as shown in FIG. 9, for each data item such as “product width”, “product thickness”, “component composition category”, “heating temperature error”, A threshold that can be classified is determined so that the data of group A and the data of group B have the fewest exceptions. If the threshold value determined in this way is used, for example, it is possible to sort such that group A is larger than the threshold and group B is smaller. The threshold value determined in this way is sent to the search condition proposal creating unit.

  The search condition plan creation unit is a part that combines the features extracted by the feature extraction unit and changes the combined features into a search condition plan. Specifically, the search condition draft creation unit obtains a conditional expression that classifies the A group with respect to the B group using the threshold value determined for each data item (operation variable) by the feature extraction unit. . Then, all the conditional expressions obtained for each data item are combined, and a search condition proposal corresponding to the set product of all the conditional expressions is created. The search condition plan created by the search condition creation unit is displayed on the display unit 3.

The display unit 3 is a part that displays the operation data output from the operation data storage unit 4, the search condition plan created by the search condition creation unit, the search condition stored in the search condition storage unit 6, and the like.
Specifically, when the operation data output from the operation data storage unit 4 satisfies the search condition registered in the search condition storage unit 6, the display unit 3 registers the registered search that the operation data satisfies. The condition is displayed on the display unit 3 as a “hit search condition”. If the operation data does not satisfy any of the search conditions registered in the search condition storage unit 6, the search condition created by the search condition creation unit is displayed as a “search condition plan”.

  Further, on the monitor screen which is the display unit 3, registration means (registration button 9) for registering a search condition plan as a new search condition in the search condition storage unit 6, and the lot number currently searched for in the lot number storage unit 5 And a lot number registration button 13 to be registered in. That is, when the searcher S determines that it is better to register the proposed search condition displayed on the display unit 3, the searcher S clicks the registration button 9 and stores the new search condition in the search condition storage unit 6. Register as In this way, when a search for a similar defect occurs thereafter, it is possible to handle it using the newly registered search condition.

Further, when the searcher S determines that it is better to register the lot number and the search result (search condition ID) searched this time, the lot number registration button 13 is clicked, and the lot number and the search ID are set to the lot number. Store in the number storage unit 5. In this way, it is possible to create a proposed search condition using the stored lot number search data.
Next, a procedure for performing a search using the operation data search support device 1 described above, in other words, the operation data search support method of the present invention will be described.

The flowcharts shown in FIGS. 4 and 5 show a flow of a series of search procedures.
As shown in FIG. 4, the operator first inputs the lot number in the lot number input field 10 (the black portion shown in the upper side of FIGS. 2 and 3) on the monitor screen (STPE1).
Then, operation data corresponding to the input lot number is output (extracted) from the operation data storage unit 4 (STEP 2).

Next, whether the output operation data satisfies the search condition registered (stored) in the search condition storage unit 6, in other words, can be extracted with the search condition (registered search condition) registered in the search condition storage unit 6. Is determined (STEP 3).
As a result of the determination, when the operation data to be output satisfies any of the registered search conditions, the satisfied registered search conditions are displayed on the monitor screen as the display unit 3 (STEP 6).

On the other hand, if the operation data to be output does not satisfy any of the registered search conditions as a result of the determination, the feature extraction unit operates the operation data of the lot number searched (designated) in the past and the operation searched this time. Features common to the data are extracted (STEP 4). This feature extraction is carried out in detail according to the procedure shown in FIG.
As shown in FIG. 5, the feature extraction unit first selects a lot number stored in the lot number storage unit 5 that does not hit the existing search condition, in other words, a search condition that has not been registered. . Specifically, a search condition ID is assigned to any search condition in the search conditions recorded in the lot number storage unit 5. Therefore, when selecting an unregistered lot number, only the lot number of “−1” whose search condition ID indicates unregistered is extracted (STEP 4-1).

Next, the operation data corresponding to the extracted lot number and the operation data corresponding to the lot number input this time are output from the operation data storage unit 4. The operation data output in this way is defined as a group A indicating a “search history present” group (STEP 4-2).
In addition, the operation data of the lot number that is not registered in the lot number storage unit 5 from the operation data storage unit 4, in other words, that has not been searched in the past (it is considered to be normal operation data without defects). Stuff). The operation data extracted in this way is defined as a group B indicating a “no search history” group (STEP 4-3). That is, the group A and the group B defined in this way are classified as shown in FIG. 9, for example.

Data items (operation variables) for classifying (separating) the operation data of group A and group B extracted in this way and threshold values thereof are extracted. For this extraction, for example, an information amount (entropy) calculation is used, and a technique such as “C”, which is used to create a decision tree, is used. By using such a method, a threshold value is obtained for each data item, and a conditional expression (rule) for extracting the data of group A and the data of group B without exception using the threshold value is obtained. The conditional expression thus obtained is sent to the search condition creating unit as the above-described feature (STEP 4-4).

The search condition creation unit converts the sent feature into a search condition. For example, as shown in FIG. 10, the result of using a technique such as “C” utilized for creating a decision tree is created in the form of a SQL conditional phrase (WHEHRE phrase), which is a language for database access. To do. Then, the sent feature can be converted into a search condition (STEP 5).
The search conditions created in this way are displayed on the lower side of the monitor screen shown in FIG. 3 as “draft search conditions” (STEP 6).

  The monitor screen also displays operation data corresponding to the lot number input this time and other operation data that satisfies the proposed search condition. Further, when displaying the search condition proposal, the ratio of the number of cases satisfying the search condition proposal and the number of noises (“2/26” in the figure is the ratio and “0” is the number of noises) are also presented. This ratio indicates how much operation data among all the operation data satisfies the proposed search condition. The number of noises indicates how many lot numbers extracted in the group A by the feature extraction unit do not satisfy the proposed search condition.

  If the operator who sees the search condition plan displayed on the monitor screen in FIG. 3 described above may register the presented search condition plan, a registration button 9 provided on the lower side of the monitor screen is displayed. Click. Then, the proposed search condition is registered in the search condition storage unit 6 as a new search condition. Also, when the lot number used for the current search is registered as a searched lot number, the lot number registration button 13 provided at the lower side of the screen is also clicked. Then, the lot number used for the current search is registered in the lot number storage unit 5 (STEP 7).

By providing such a lot number registration button 9, it becomes possible to suppress the registration of exceptional operation data, and "operation data that is not desired to be used for creating search conditions is also registered". It becomes possible to deal with the disadvantages of automation.
The monitor screen shown in FIG. 3 is provided with an “update screen button”. When this “update screen button” is clicked, each time operation data is newly generated, the newly generated operation data is compared with the search condition stored in the search condition storage unit 6.

  Specifically, as shown in FIG. 11, when a search is performed on the operation data of the lot numbers “1” to “31”, the lot numbers “32” and “33” are newly added. Let us consider a case in which a product (rolled) is manufactured. When such a lot number “32” or “33” (operation data linked to this lot number) is newly generated, it corresponds to the lot number “32” or “33” as shown in FIG. Operation data is displayed as a list on the monitor screen as an automatic search result. When this list display is clicked, a list of operation data is displayed as in the case of manual search. If such an automatic search result is displayed, a search can be performed while the operation is performed, and a quality defect can be quickly dealt with.

  In the operation data search support apparatus 1 described above, when a lot number is input to the input unit 2 and the search button 11 is pressed, not only the operation data corresponding to the lot number, but also other lot numbers searched in the past by this operation data. It becomes possible to judge from the search conditions what kind of relationship it is with the operation data of the user, and it becomes possible to assist the searcher to investigate the cause of the defect by using the operation data as the search result.

In addition, when there is no registered search condition corresponding to the operation data of the lot number input to the input unit 2, the operation data that has been already searched and the search condition is not registered During this period, a search condition draft that is a candidate for the search condition is created, and the created search condition draft can be displayed on the display unit 3. When the search condition plan displayed on the display unit 3 in this way is reasonable from the operator's perspective, the operator can register the search condition plan as a new search condition by pressing the registration button 9. It becomes.

  That is, in the operation data search support device 1 described above, since the search condition is automatically created from the search of the operation data that is repeatedly performed, the operator does not need to perform a complicated operation such as class, and lots By simply repeating a simple operation of searching for operation data by specifying a number (identifier), search conditions that support judgment on search results can be automatically obtained. In addition, regardless of the amount of experience of the operator, it becomes possible to perform effective filtering and checking using operation factors that are likely to cause problems before the big trouble occurs on the operation data generated in daily operations. As a result, the search can be performed easily and efficiently.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. In particular, in the embodiment disclosed this time, matters that are not explicitly disclosed, for example, operating conditions and operating conditions, various parameters, dimensions, weights, volumes, and the like of a component deviate from a range that a person skilled in the art normally performs. Instead, values that can be easily assumed by those skilled in the art are employed.

DESCRIPTION OF SYMBOLS 1 Operation data search support apparatus 2 Input part 3 Display part 4 Operation data memory | storage part 5 Lot number memory | storage part (identifier memory | storage part)
6 Search condition storage section 9 Registration button (registration means)
10 Input field 11 Search button 12 Main body 13 Lot number registration button P Manufacturing equipment S Searcher

Claims (8)

Operational data bundled are stored in the product identifier and the identifier produced by the production facility,
Product identifier is input to search for,
When the operation data corresponding to the input identifier does not satisfy any of the search conditions registered in the past, the operation data of the input identifier has been searched in the past, and the and operation data of other identifiers that none of the registered search condition is not satisfied, are out to choose,
Common point of data between two or more operational data between which singled out is extracted as a feature,
How extracted features are converted to search conditions proposed in addition to the operational data of the input identifier, operation data retrieval support apparatus executes, wherein Rukoto the converted search proposal appears. How to perform the operation data retrieval support apparatus of claim 1, wherein the displayed search criteria proposal whether registered Luca is possible to select as the search criteria. If the operational data of the product currently being manufactured satisfies any one of the registered search condition, operation data according to claim 1 or 2 search condition the satisfactory characterized Rukoto appears A method executed by the search support apparatus . Between two or more operational data picked the common features are extracted Runisaishite, clustering using techniques decision tree learning according to any one of claims 1 to 3, wherein the dividing lines A method executed by the operation data search support device . An operation data storage unit that stores an identifier of a product manufactured in the manufacturing facility and operation data associated with the identifier;
An input unit for inputting an identifier of a product to be searched;
If the operation data corresponding to the identifier input to the input unit does not satisfy any of the search conditions registered in the past, the operation data of the identifier input to the input unit and that the search has been performed in the past The operation data having another identifier that does not satisfy any of the registered search conditions is selected, and the common point of the data is extracted as a feature between the two or more selected operation data. A feature extraction unit;
A search condition proposal creating section for converting the feature extracted by the feature extraction section into a search condition proposal;
A display unit that displays the operation data of the identifier input to the input unit from the operation data storage unit and displays it, and displays the search condition plan created by the search condition plan creation unit as a new search condition,
An operation data search support device characterized by comprising:   The operation data search support apparatus according to claim 5, further comprising a registration unit that registers the search condition plan displayed on the display unit as a new search condition.   The display unit is configured to display the satisfied search condition when operation data of a currently manufactured product satisfies any of the search conditions registered by the registration unit. The operation data search support device according to claim 6.   The feature extraction unit is configured to perform clustering using a decision tree learning technique when extracting a common feature between the two or more selected operation data. The operation data search support device according to any one of 5 to 7.