System and method for preventing defects in work processes

Abstract

This invention provides a system and method for preventing defects in the work process. [Solution] The defect prevention system comprises a detector that detects reactions caused by the subjective perception of line workers in the work process of manufacturing products such as equipment and food; a collection unit that collects information obtained by the detector; an analysis unit that analyzes the information collected and stored by the collection unit; an extraction unit that detects defect indicator reactions from the analysis of the analysis unit; a determination unit that determines the frequency of occurrence of defect indicator reactions using a threshold; a defect indicator database whose components include at least defect indicator reactions and defect information; a conversion unit that converts defect indicator reactions into defect information in the defect indicator database; a manufacturing knowledge database in which countermeasure actions are linked to defect information; and an identification unit that identifies countermeasure actions from defect information in the manufacturing knowledge database.

Description

Technical Field

[0001] Embodiments of the present invention relate to a system and method for preventing defects in a work process.

Background Art

[0002] In an assembly process on a production line in manufacturing or the like, techniques for grasping work man-hours and unexpected operations by photographing and recording the actions of line workers using videos, sensors, etc. have been established. Moreover, the photographed and recorded data is also utilized for factor analysis when defects occur on the line (see, for example, Patent Document 1). In addition, techniques for reporting a defective situation by voice and converting the voice data into text data, and for analyzing the text data to identify defect information have also been disclosed (see, for example, Patent Document 2). However, in the prior art, it is not possible to obtain the "difficulty of work based on the subjective perception of the line worker himself / herself" that cannot be measured by videos or sensors. Also, in a report on a situation where a defect has already occurred, it is not possible to prevent the occurrence of the defect in advance.

[0003]

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0005] The problem to be solved by the present invention is a system and method for preventing defects in a work process.​​​​​​​ The objective is to provide a method for stopping it. [Means for solving the problem]

[0006] To solve the above problems, the defect prevention system of the embodiment is used in the manufacturing of equipment, food, etc. In the manufacturing process of a product, a detector is used to detect reactions caused by the subjective perception of the line worker. , an information collection unit that collects information obtained by the detector, and the information collected and stored by the information collection unit An analysis unit that analyzes information, and an extraction unit that detects signs of malfunction from the analysis of the analysis unit, A determination unit that determines the frequency of occurrence of the aforementioned malfunction indicator reaction using a threshold, and a component that has at least the aforementioned malfunction A malfunction indicator database containing indicator reactions and malfunction information, and the malfunction indicator database The system includes a conversion unit that converts the malfunction indicator reaction into malfunction information, and a unit that converts the malfunction information into malfunction information. A manufacturing knowledge database to which strategic actions are linked, and the said manufacturing knowledge database The system is characterized by comprising a specific unit that identifies countermeasures based on the aforementioned defect information. The line workers perform their work according to the prescribed process control charts, etc., and in doing so, If the line work does not proceed smoothly, the likelihood of malfunctions increases. (Video and sensors) Even if the data obtained makes it appear as though the work is being carried out without any problems, the line workers themselves may feel that the work is being done without any issues. They may be feeling something is wrong with their work. Furthermore, it captures signs of malfunction, such as sounds, that are emitted as a reaction to that perception, and By collecting and analyzing the content and frequency of issues, we can improve work processes before problems occur and prevent problems from happening in the first place. This can be prevented naturally. The malfunction signs and reactions spoken by the line worker and the line worker's own In addition to behavioral tendencies, a defect indicator database linked to process control sheets and work procedures is used. Identify the work to be improved, possible defects, and the corresponding countermeasures (hereinafter referred to as countermeasure actions) using a manufacturing knowledge database where they are associated, and the necessary countermeasures are automatically shown.

Brief Description of the Drawings

[0007] [Figure 1] It is a conceptual schematic diagram of a defect prevention system in the work process according to the first embodiment. [Figure 2] It shows a flowchart of a method for preventing defects in the work process according to the embodiment. [Figure 3] It is a schematic diagram of a defect sign database according to the first embodiment. [Figure 4] It is a schematic diagram of a manufacturing knowledge database according to the first embodiment. [Figure 5] It is a conceptual schematic diagram of a defect prevention system in the work process according to the second embodiment. [Figure 6] It is a conceptual schematic diagram of a defect prevention system in the work process according to the embodiment.

Modes for Carrying Out the Invention

[0008] Hereinafter, embodiments will be illustrated while referring to the drawings. In each drawing, the same components are denoted by the same reference numerals, and detailed descriptions thereof are omitted as appropriate. (First Embodiment)

[0009] The first embodiment will be described. FIG. 1 is a conceptual schematic diagram of a defect prevention system in the work process according to this embodiment. The figure includes a line worker W, a detector 1, a primary processing unit 6, and a secondary processing unit 9. The primary processing unit 6 includes a collection unit 2, an analysis unit 3, an extraction unit 4, and a determination unit 5. The secondary processing unit 9 includes a defect sign database 7, a conversion unit 8, and a manufacturing part 6 and a secondary processing part 9. The primary processing part 6 includes a collection part 2, an analysis part 3, an extraction part 4, and a determination part 5. The secondary processing part 9 includes a defect sign database 7, a conversion part 8, and a manufacturing ​​​The knowledge database 10 includes the specific part 11.

[0010] The line worker W shows a reaction due to subjectivity when feeling discomfort in the working process. At this time, the detector 1 detects the reaction due to the subjectivity as input information. The information input into the detector 1 is sent to the primary processing unit 6 and processed according to the procedure described later. In the primary processing unit 6 first, the collection and accumulation of the information sent from the detector 1 by the collection unit 2 are performed Next, the analysis unit 3 analyzes the information accumulated by the collection unit 2. Next, the extraction unit 4 extracts a defect symptom reaction from the information analyzed by the analysis unit 3. After that, the determination unit 5 calculates the occurrence frequency of the defect symptom reaction. When the occurrence frequency exceeds the threshold value in the determination unit 5 the information regarding the defect symptom reaction processed by the primary processing unit 6 is sent to the secondary processing unit 9

[0011] The defect symptom database 7 included in the secondary processing unit 9 is a database including at least the components of the defect symptom reaction and the components of the defect information, and may include other multiple components associated with the defect symptom reaction. The conversion unit 8 has the role of converting the information regarding the defect symptom reaction sent from the primary processing unit 6 into defect information in comparison with the defect symptom database 7. The manufacturing knowledge database 10 is a database including components in which possible defects in the working process and countermeasure methods to be taken (hereinafter referred to as countermeasure actions) are associated with the defect information. In the specific part 11 the above-mentioned defect information sent from the conversion unit 8 is compared with the manufacturing knowledge database 10 to identify the corresponding potential defects

[0012] ​​​​​In the first embodiment, the difficulty of the work is obtained based on the subjective perception of the line worker, and the work process is... To prevent defects from occurring. A work process is a standard procedure for efficiently performing work in a factory or on site. This refers to a process. The work process is carried out by one or more people. This includes factory assembly and processing plants, restaurant kitchens, mail sorting areas, airport inspection areas, and nuclear power plants. It is being implemented in various places such as power plants. In these places, in order to standardize the work, A set process is established, and the work is carried out according to that process. Such a process includes the work To stabilize quality, improve work efficiency, ensure work safety, prevent reliance on specific individuals, etc. There is a purpose. Each work process has specific requirements and objectives at the place where it is carried out. It is determined based on.

[0013] The first embodiment will be described below, with more detailed examples provided.

[0014] Figure 2 is a flowchart of the first embodiment, "Method for preventing defects in the work process." The subjective response input to detector 1 is, for example, speech. The response is any response that can be monitored using sensing technology. That's fine. Detector 1 is a device with a voice perception function, and it detects the voice spoken by line worker W. It is attached to a part that can adequately pick up the signal.

[0015] Line worker W is a worker at Factory A who tightens screws on specific part B according to the work process. Worker W has repeatedly performed the screw tightening operation on specific part B multiple times in the past, You can intuitively determine whether the screws on item B are sufficiently tightened. Intuition is a form of knowledge. The owner, based on past experiences, etc., instantly arrives at a conclusion without engaging in logical thinking such as reasoning or analogy. It is a type of sensory cognitive form that can be expressed.

[0016] At the current level of technology, artificial intelligence that makes decisions based on quantitative data according to an algorithm Stems cannot be judged intuitively. Quantitative data includes, for example, dimensions and weight. In addition to things that can be expressed by specific numerical values ​​such as temperature and pressure, there are also videos, recordings, photography and measurement, and cameras. This includes numerical values ​​obtained from sensors, etc. From this point onward, the system makes decisions based on quantitative data. The method using TEM is referred to as a "decision-making method based on quantitative data."

[0017] The following explains the shortcomings of the judgment method based on quantitative data. An artificial system makes decisions based on videos and sensed records taken on behalf of contractor W. Let's assume that Tem α is introduced. Note that the artificial system α is a determination method based on quantitative data. It is assumed to be equipped with.

[0018] Firstly, there are minor defects in the product (specific component B and) that cannot be identified by artificial system α. Using the example of a case involving a computer, the shortcomings of a judgment method based on quantitative data will be explained. Specific parts B and screws may vary in quality during the manufacturing process, even if they are the same product. Each item has individual differences (for example, wear in certain areas, variations in materials, changes in processing conditions, etc.). There are individual differences, however, that cannot be identified from videos, recordings, photography, measurements, cameras, sensors, etc. Although minor, it can sometimes cause malfunctions. Accuracy depends on the performance of the video, recording, shooting, measurement, camera, sensor, etc. If there are individual differences that cannot be identified through recording, photography, measurement, cameras, sensors, etc., Even if a defect occurs in the aforementioned product, it cannot be detected. The aforementioned individual differences in the population system α To identify this, use higher-performance video, recording, shooting, measurement, cameras, sensors, etc. It is necessary. However, due to factors such as high cost, it is not widely used in factories and on-site. difficult.

[0019] Next, we will use the characteristics of artificial system α as an example to illustrate the shortcomings of judgment methods based on quantitative data. Let me explain. In the work process, we determine whether specific part B and the screw are tightened. To this end, artificial system α uses a weight sensor. The weight sensor measures the "weight" of an object. It possesses specific characteristics. Here, characteristics refer to the unique functions and capabilities of the camera, sensor, etc. In this case, by using a weight sensor, the measured value (i.e., quantitative data) can be used to identify specific part B. It is possible to determine whether the screw is assembled. Specifically, the method involves determining whether a particular part is assembled. The weight is set based on the weight of B and the screws when they are correctly assembled. Then, the actual assembly is performed. The weight of the attached part is measured and compared to the reference weight. If the weight is less than the minimum weight, there is a possibility that there is a problem with the assembly of specific part B and the screw. It can be determined that it exists. However, this method only compares the measured value of "weight". Therefore, it is not possible to determine the torque level (i.e., whether the screw is sufficiently tightened). No. In order to determine the torque, for example, a device with characteristics for measuring torque values ​​is used. A digital torque wrench or similar tool will be required. Thus, a judgment method based on quantitative data is... This is a determination method that depends on the characteristics of cameras and sensors, and has the defects described above. do.

[0020] On the other hand, line worker W, who performs tasks while making decisions based on intuition, makes decisions based on quantitative data. They may also be feeling uneasy about aspects that cannot be grasped from the outset.

[0021] For example, regarding the individual differences mentioned above, if the line worker W is the one who is having difficulty inserting the screw into a specific part B, It may be possible to identify the problem by noticing something unusual, such as a specific part. In the process of tightening screw B, line worker W initially tightened the screw without noticing anything unusual. Let's assume that this was being done. However, at a certain point, line worker W found that the screws were harder to insert than before. I begin to feel a sense of unease. This unease is related to video, recording, shooting, measurement, cameras and sensors. The cause is a minor defect that cannot be detected by such methods (for example, a defect in a specific part B). (Assembly line work) Person W carefully (carefully / with attention / with precision) kept the screwdriver in the correct orientation and tightened the screw. The quality of the torque is ensured by performing the work. If the work continues as is, the line There is a possibility that worker W may lose concentration and drive the screws in at an angle, resulting in a defective product. After making several utterances such as "it doesn't fit" (i.e., "subjective reactions" mediated by speech) Yes, they are.

[0022] The feeling of unease experienced by line worker W is a subjective one obtained through the animal's sensory organs. There can be individual differences in the accuracy of the judgment. Since the feeling of discomfort is subjective, line worker W When something feels off during the work process, whether consciously or unconsciously, it triggers a subjective reaction. It is expected that this will be demonstrated.

[0023] In process S1, line worker W performs a task of tightening screws on a specific part B, and there is a sense of discomfort. Let's say that the line worker W is trying to memorize something. In this case, as a reaction caused by subjectivity, for example, he might make a sound. Speaking provides input to detector 1. The sound input to detector 1 is converted from analog to It is converted into digital information and sent to the primary processing unit 6, for example, via a local communication environment. (Note: A remote communication environment using the internet is also acceptable.)

[0024] In the primary processing unit 6, the collection unit 2 records (collects and stores) digital information, and the analysis unit 3 Therefore, it has a step S2 in which the recorded digital information is analyzed. The analysis in step S2 is It plays a role in the system recognizing the input audio data as language.

[0025] In step S3, some of the data recognized by language recognition in step S2 is processed by the extraction unit 4. These are extracted as indicator words. Indicative words for malfunctions include objective terms such as "hot" and "hard." In addition to spoken words whose meaning can be understood, it also includes slang, interjections, and other similar expressions. Furthermore, the words indicating malfunctions vary greatly depending on the line worker W's catchphrases, native language, and experience. There is a variation. In addition, the primary processing unit 6 detects the malfunction indicator word LLM (Large You may use technologies such as Language Models (Language Models) or RAG (Retrieval-Augmented Generation).

[0026] The defect indicator words extracted in process S3 are used for calculation of their frequency of occurrence in the judgment unit 5. In step S4, if the frequency of occurrence of a word indicating a malfunction exceeds a certain threshold, it indicates that there is a problem in the work process. The judgment is made that there is a problem. The presence of a problem means that there is a possibility that a problem is occurring. Yes, there is. For example, line worker W, who does not speak in particular when he does not feel anything unusual, temporarily If the user utters the word "doesn't fit" five or more times during the work process, In step S4, a problem is identified. If a problem is identified in step S4, a secondary process is performed. Access is made to the malfunction indicator database 7 contained within section 9.

[0027] Information related to the defect symptom word processed in the primary processing unit 6 indicates a defect in process S5. The system compares the results with the symptom database 7 and converts them into defect information in the conversion unit 8. The candidate database 7 consists of at least two components: "indication of malfunction" and "malfunction information". This is a database containing elements. Figure 3 is a schematic representation of the defect indicator database 7 according to this embodiment. This is a diagram. In the same diagram, the component, the malfunction indicator reaction, is referred to as the "malfunction indicator word". This indicates that malfunction signs and reactions include, for example, voice, facial expressions, posture, body temperature, sweating, and skin electrical activity. These include motion, blood pressure, and electroencephalogram, but it is necessary to distinguish between them. In this embodiment, line worker W Since this method captures signs of malfunction from speech, these signs of malfunction will be newly identified. The term "defect indicator word" is defined and shown as a component. As shown in the figure, the defect in this embodiment The indicator database 7 organizes data into rows (horizontal) and columns (vertical). In Figure 3, each column represents a component, and each row connected to each component represents an individual event. This represents the malfunction indicator database 7, which may also be in the form of a tree diagram or other formats. It doesn't fit. In the same diagram, line worker W, who is tightening the screw on specific part B, says, "It doesn't fit." If the malfunction symptom word " " is extracted, the conversion unit 8 will convert it to "does not fit". Compare the Word document with the defect indicator database 7 shown in Figure 3. The suspect database 7 already has the malfunction symptom word "doesn't fit" registered. The conversion unit 8 displays the malfunction information: "If specific component B malfunctions, product C will stop operating." It is possible to perform a conversion to [this format].

[0028] Once the conversion unit 8 converts the information into defect information, the secondary processing unit 9 processes the manufacturing knowledge database Access to -10 is made. The manufacturing knowledge database 10 is used at each factory and site. Regarding effective countermeasures (countermeasure actions) for accidents and troubles that have been accumulated through accumulated know-how This information is linked to defect information and compiled into a database. Process S6 is a specific part Based on the defect information, 11 performs a search of the manufacturing knowledge database 10, This identifies the appropriate countermeasures and actions.

[0029] Figure 4 is a schematic diagram of the manufacturing knowledge database 10 according to this embodiment. The manufacturing knowledge database 10 is represented by rows and columns, with "parts," "processes," and "work operations." It includes the components of "content", "defect information", and "countermeasure action". For example, conversion unit 8 The conversion revealed a malfunction: "If specific component B malfunctions, product C will stop operating." Let's assume that the following result was obtained. In this case, access to the manufacturing knowledge database 10 is performed. Based on this defect information, the specific unit 8 performs a search in the manufacturing knowledge database 10. As a countermeasure action for the defect information, the search result yielded "change the material of specific part B". It can be done.

[0030] Through this series of processes, the difficulty level of the work, based on the subjective opinion of line worker W, is obtained. By implementing appropriate countermeasures, we can prevent defects in the work process. It is possible.

[0031] Here, we will also explain the component "Awareness (Line Worker W)" shown in Figure 3. In the diagram, "Awareness (Line Worker W)" refers to the line worker's awareness during any given work process. When W showed signs of a malfunction, line worker W himself noticed (i.e., the work was different) This represents subjective feelings and perceptions such as experiencing a sense of harmony. This embodiment is for line worker W himself. When an awareness is gained, the malfunction signs and reactions that are emitted as a response to that awareness are captured, The next step is to identify the corrective action. Therefore, the process of this embodiment is line production Even if we can infer what contractor W noticed from the information linked to the signs of a malfunction, the actual line It is not possible to directly know what worker W noticed. When a malfunction sign was detected, In order to accurately understand what line worker W specifically noticed, line worker W's book It is necessary to confirm this directly with a person. Through repeated verification of this process, "indications of a malfunction" can be detected. " and "defect information" are linked to "notice (line worker W)", and defect indication data Base 7 will be constructed.

[0032] To build the malfunction indicator database 7, past data (for example, the behavioral trends of line worker W) is needed. The collection, accumulation, and organization of such data are necessary, and because it involves elements that depend on human cognition, it is uncertain. It is real. Here, let's explain uncertainty with a specific example. Line worker W is identified Let's say that when checking the operation of component B, it uttered the malfunctioning word "hot". In total, line worker W noticed that "the ambient temperature is hot," "part B is hot," Several interpretations can be given, such as "part B is thick." However, there are multiple interpretations. Among these, it is impossible to determine which one is being referred to solely from the malfunction symptom words. However, this introduces uncertainty into the results of the conversion unit 8's conversion into defect information.

[0033] While it is impossible to completely eliminate such uncertainties, it is possible to narrow the scope of those uncertainties. It is possible. One way to narrow the range of uncertainty is when building the defect indicator database 7. The goal is to increase the number of components included (i.e., information linked to malfunction indicators).

[0034] For example, a line worker W, who is checking the operation of a specific part B, might utter the word "hot," which is a sign of a malfunction. Let's assume that a suspected word was uttered. In response to this suspected word, the suspected word database 7 "Normal environment: The temperature at Factory A is a constant 5°C", "Current environment: The current temperature at Factory A is 5°C" Let's assume that the two components, , and are linked. In this case, the disability is "hot". It is not appropriate to convert the symptom keyword into malfunction information such as "the ambient temperature is hot." Because, even though the temperature hasn't changed, the current situation is "a" compared to the past. The increased frequency of the utterance "tsui" is unlikely to be due to changes in temperature. Therefore, we can exclude the option (interpretation) that "the ambient temperature is hot." As mentioned above, the conversion into defect information can be performed using natural language processing, etc. Collecting information linked to malfunction signs and reactions and increasing the number of components is an overreaction when performing conversion. This reduces the number of options (which correspond to multiple interpretations or noise) and narrows the uncertainty. To build a more accurate defect information database 7, it is necessary to link defect symptoms to reactions. Information gathering and organization of constituent elements are necessary.

[0035] Next, in the malfunction indicator database 7, there is "Characteristics: Line worker W is a skilled worker", and "Trend: Line Let's consider a case where the two components are included: "Worker W is taciturn" and "Worker W is taciturn." If the training unit utters the malfunction indicator word "hot," then the malfunction indicator word associated with that will The possibility of accurately identifying potential defects is greater than if the two aforementioned components were absent. The cost will also increase. The conversion performed by the conversion unit 8 may use, for example, natural language processing. In addition, the transformation unit 8 operates based on mathematical and statistical methods, learning patterns from the data. It is possible to perform predictions and transformations. In other words, increasing the number of components allows for weighting of information, etc. This improves the accuracy of the conversion. Information weighting refers to the importance of different pieces of information. This involves assigning different "weights" to each item based on its degree, priority, or reliability. For example, both the "talkative inexperienced worker" and the "taciturn skilled worker" can be described as "hot." Let's assume that both parties uttered the same words indicating a malfunction. If we consider the weighting of the information in their statements, for example... When comparing the two, the response "It's hot" from the "usually taciturn skilled worker" is more reliable. Therefore, a malfunction symptom reaction from a "normally taciturn skilled worker" is not the same as a reaction from a "normally talkative worker" The information can be given more weight than the malfunction indicators and reactions of "inexperienced workers." The weighting of these values ​​is useful for setting thresholds, etc., and therefore leads to improved system accuracy. In this way, To build a more accurate defect information database 7, information linked to defect symptoms and reactions is needed. Information gathering and organization of constituent elements are necessary.

[0036] This embodiment uses a defect indicator database 7 and a manufacturing knowledge database 10. Next, we will explain the differences between the two. As mentioned earlier, the defect indicator database 7 is uncertain. It has the characteristic of changing the accuracy of the conversion depending on the amount of data collected, stored, and organized. On the other hand, the manufacturing knowledge database 10 appears to be routinely used in each factory and on-site. This is a database of established, known information and is not directly related to malfunction symptoms or reactions. Because there is no relationship, there is no uncertainty. This embodiment improves the accuracy of the conversion while having uncertainty. A database of defect signs 7 that can be improved, and manufacturing narration including established known information. By using the data database 10 in conjunction with the system, defects in the work process can be prevented. (Second Embodiment)

[0037] A second embodiment will now be described. Note that points common to the first embodiment will be omitted from the explanation. Figure 5 is a flowchart of a method for preventing defects in the work process according to this embodiment. Yes. Similar to the first embodiment, it includes a line worker W, a detector 1, a primary processing unit 6, and a secondary processing unit. In addition to the processing unit 9, it has a display unit 12. The display unit 12 communicates with the secondary processing unit 9 via local communication. Assume they are connected. The display unit 12, according to the flowchart in Figure 2, in process S6 After identifying the countermeasure action, the secondary processing unit 9 provides information regarding the countermeasure action. Receive. Information regarding countermeasures includes calculation results obtained in the primary processing unit 6 and manufacturing This may also include secondary information such as components linked to Knowledge Database 10. For example, in the manufacturing knowledge database 10 in Figure 4, the corrective actions are linked. The components are parts, processes, work details, and defect information. The display unit 12 is two The information regarding the countermeasure action obtained from the next processing unit is processed into human-recognizable characters or language, etc. The system converts the data into a format and displays it on a display medium. In the second embodiment, the display medium This is a liquid crystal display. "Display" means to show or indicate clearly. Therefore, people As long as the audience can perceive the context, the display medium can take any form (e.g., audio commentary). No. Furthermore, if the display unit 12 is a display medium with input functionality, such as a touch panel terminal... In addition, line worker W noticed his own observations (i.e., subjective feelings and perceptions such as discomfort with the work content) It is acceptable for the information to be input and recorded in the display unit 12. The display unit 12 is for line workers W, etc. It has the role of informing information regarding the aforementioned countermeasures. The display unit 12 provides information about the A factory. The parties involved, including worker W, improved their work procedures before the malfunction occurred and prevented the malfunction from happening in the first place. It can be prevented. (Third embodiment)

[0038] A third embodiment will be described. In the first embodiment, the determination unit 5 in the primary processing unit 6 In this process, when the frequency of occurrence of a word indicating a malfunction exceeds a threshold, the primary processing unit 6 processes it. Information regarding the malfunction indicator word is sent to the secondary processing unit 9. In the flowchart in Figure 2... Step S4 is the step in which the determination unit 5 calculates a threshold for the frequency of occurrence of the defect indicator word. In the third embodiment, the subsequent flow is selected based on the threshold determination in step S4. This section describes the situation in which a branch route is selected and followed. Process S4 shows signs of failure. There are branching paths for Yes and No depending on the frequency of word occurrence. In the case of a "Yes" branch route, the process proceeds sequentially to step S5 according to the flowchart. The Yes branch route leading to processes S5 and S6 is the same as the flow described in the first embodiment. This is the process.

[0039] On the other hand, if the frequency of occurrence does not exceed the threshold, the No branch route is selected. No branch route So, line worker W is experiencing difficulties with the work (i.e., feeling something is wrong). Because the probability is low, the situation is "the work is proceeding without any problems." Therefore, the branch route for No is If selected, the flow ends without any subsequent steps performed in the secondary processing unit 9 after step S4. do.

[0040] Furthermore, information regarding the frequency of occurrence of error indicator words is stored in collection unit 2 (i.e., collected and stored). Furthermore, it may be used as data when rebuilding (organizing) the malfunction indicator database 7. stomach.

[0041] Figure 6 is a conceptual schematic diagram of a defect prevention system in the work process according to this embodiment. As shown in the figure, the display unit 12 is connected to the primary processing unit 6 via a local communication environment. This is where it differs from Figure 5. As a result, the primary processing unit 6 directly communicates with the display unit 12. This enables the exchange of information. Here, in process S4, the branch route No is selected. Let's consider the case where the information obtained by the primary processing unit 6 is sent to the display unit 12. The method of transmitting information to unit 12 is either "from primary processing unit 6 to display unit 12" or "from primary processing unit 6 to display unit 12". There are two ways of transmission: via the secondary processing unit 9 to the display unit 12. Since information can be directly transferred to the display unit 12, the processing time is reduced because it goes through the secondary processing unit 9. This is shorter than the case where it goes through the secondary processing unit 9. On the other hand, the transmission method when going through the secondary processing unit 9 is Because an extra step is involved, the processing time is increased. Display unit 1 after step S4. It doesn't matter which transmission path is used, as long as the information can be sent to point 2.

[0042] In this embodiment, if the Yes branch route is selected, the contents described in the second embodiment will be as described. Information regarding corrective actions will be displayed using a similar procedure. On the other hand, if the branch route "No" is selected... When this happens, the signal follows the path from the primary processing unit 6 to the display unit 12, and the display is then shown in the display unit 12. The speed at which information is reflected in the display unit 12 is determined by the Yes branch route via the secondary processing unit 9. It was also quick. Therefore, the parties involved, including line worker W at Factory A, were able to perform their work before the malfunction occurred. This can improve the condition and prevent malfunctions from occurring.

[0043] The branch route for "No" is when the frequency of malfunction indicator reactions does not exceed the threshold and "work is performed without problems." This is equivalent to being in a state where "it is in a certain condition." Therefore, by appropriately changing the threshold setting, for example, Whether the product containing specific part B assembled by worker W is defective, good, or excellent, etc. It is also possible to support product sorting. In this process, each line worker is given the malfunction indicator reaction described in this embodiment. By performing monitoring and threshold determination, we can prevent defects from occurring and improve the precision of product selection. It is also possible to improve the degree. (Fourth embodiment)

[0044] A fourth embodiment will be described. In the previous embodiments, the defect information database 7 We have been explaining that the malfunction indicator reaction, which is part of the components, is in the form of sound. However, the number of components can be increased or decreased as needed, and malfunction indicators are limited to audio. This is not the case. Therefore, in this embodiment, the detector 1 shown in Figure 1 detects reactions caused by subjectivity. To detect any abnormalities, blood pressure, heart rate, brain waves, and changes in eye gaze direction are monitored. These will be assigned to each component as potential reactions. In this way, there are multiple types of malfunction indicator reactions. By simultaneously monitoring different types of data, the conversion accuracy of the defect information database 7 can be improved. This is possible. The malfunction indicators and reactions described in this embodiment do not depend solely on human language, In the work processes of police dogs and other animals that cannot communicate verbally (for example, olfactory inspections at airports) Even if this is the case, the series of flows shown in Figure 2 can be applied.

[0045] Furthermore, the present invention is not limited to the embodiments described above, and may deviate from the spirit of the present invention. Of course, various modifications can be made within the limits of the present invention. Although these embodiments have been described, they are presented as examples only and do not limit the scope of the invention. It is not intended that these novel embodiments may be implemented in various other forms. It is possible to make various omissions, substitutions, and modifications without departing from the spirit of the invention. This is possible. These embodiments and their variations are included in the scope and gist of the invention, and are patentable. The invention described in the claims and its equivalents are included within the scope of the claims. [Explanation of symbols]

[0046] W...Line worker 1. Detector 2. Collection Department 3...Analysis Department 4...Extraction part 5... Judgment section 6. Primary Processing Unit 7. Malfunction Indication Database 8. Conversion section 9. Secondary Processing Unit 10. Manufacturing Knowledge Database 11...Specific section 12...Display section α... Artificial System A... Factory B...Specific parts

Claims

1. A detector that detects reactions caused by the subjective nature of the worker, A collection unit that collects information obtained from the detector, The analysis unit analyzes the information collected and stored in the aforementioned collection unit, An extraction unit that detects signs of malfunction from the analysis of the aforementioned analysis unit, A determination unit that determines the frequency of occurrence of the aforementioned malfunction indicator reaction using a threshold, The components include at least the malfunction indicator reaction and the malfunction indicator database which includes malfunction information. 、 In the aforementioned malfunction indicator database, the malfunction indicator response is converted into the malfunction information. Replacement part, A manufacturing knowledge database in which corrective actions are linked to the aforementioned defect information, In the aforementioned manufacturing knowledge database, the corrective actions are identified from the defect information. Teibu and, A system for preventing defects in work processes that includes the following features.

2. If the frequency of the malfunction indication reaction exceeds a threshold in the determination unit, the countermeasure action is taken. A display unit that can display and input information related to the system, A system for preventing defects in a work process according to claim 1, comprising:

3. Claim 1 includes a plurality of components belonging to the malfunction indicator response in the malfunction indicator database. Alternatively, a system for preventing defects in the work process described in claim 2.

4. A step in the work process to detect reactions caused by the worker's subjective opinion using a detector, The steps include: collecting information obtained by the detector into the collection unit; The information collected and stored in the collection unit is analyzed in the analysis unit, and the malfunction indicator reaction is extracted in the extraction unit. The extraction step, The determination unit determines the frequency of occurrence of the malfunction indicator reaction using a threshold value, When the frequency of the aforementioned malfunction indicators exceeds a threshold, access the malfunction indicator database. The conversion unit converts the information into malfunction information, Steps in which the specified unit accesses the manufacturing knowledge database and identifies corrective actions. and, A method for preventing defects in a work process that includes [a specific feature / feature].

5. After identifying the corrective action, the display unit displays information related to the said corrective action. Top, A method for preventing defects in a work process according to claim 4, comprising:

6. If the frequency of the aforementioned malfunction indicator does not exceed the threshold, the information is sent to the display unit and the operation is not performed. Steps to display that something is being done without a title, A method for preventing defects in a work process according to claim 4, comprising:

Images