Industrial equipment operation support system

Abstract

The present invention provides an industrial equipment operation support system capable of efficiently supporting operation of industrial equipment and providing operation support appropriate even to non-experts. The industrial equipment operation support system has a server capable of accessing information on a plurality of pieces of industrial equipment. The server has a general agent and equipment agent for generating an answer to a question received through an input means. The general agent receives input of only a part of the information on the plurality of pieces of industrial equipment for all of the plurality of pieces of industrial equipment, and the equipment agent receives input of the information on the plurality of pieces of industrial equipment for only a part of the plurality of pieces of industrial equipment.

Description

Industrial equipment operation support system

[0001] The present invention relates to an operation support system for industrial equipment.

[0002] There is a demand for more efficient systems for supporting the operation of industrial equipment. Furthermore, in anticipation of the future retirement of skilled operators, there is a demand for systems that allow non-skilled operators to provide appropriate operational support.

[0003] Patent Document 1 discloses that a pump device is equipped with an AI function.

[0004] Japanese Patent Application Laid-Open No. 2022-98903

[0005] The AI ​​function described in Patent Document 1 is only used for a single device. For this reason, there has been a demand for a new operation support system that can efficiently manage the operation of multiple pieces of industrial equipment.

[0006] In view of the above circumstances, the present invention was developed as a result of the inventors' research into more effective operational support for industrial equipment. By devising ways to utilize AI, the present invention provides an operational support system for industrial equipment that is efficient and allows even non-experts to provide appropriate operational support.

[0007] An example of an industrial equipment operation support system according to the present invention for solving the above-mentioned problems is as follows.

[0008] An industrial equipment operation support system comprising: a server that can access information relating to a plurality of pieces of industrial equipment; the server having a control agent and equipment agents that generate answers to questions received through an input means; the control agent receiving only a portion of the information relating to the plurality of pieces of industrial equipment for each of the plurality of pieces of industrial equipment; and the equipment agents receiving only information relating to some of the pieces of industrial equipment for each of the plurality of pieces of industrial equipment.

[0009] According to the present invention, it is possible to provide an industrial equipment operation support system that can efficiently support the operation of industrial equipment and that can also provide appropriate operation support even to non-experts.

[0010] Further configurations and effects of the present invention will become apparent from the entire specification below.

[0011] FIG. 1 is an example of a system configuration in an embodiment of the present invention. FIG. 2 is an example of a system configuration in an embodiment of the present invention. FIG. 3 is an example of a system configuration in an embodiment of the present invention. FIG. 4 is an example of a system configuration in an embodiment of the present invention. FIG. 5 is a flowchart illustrating an example of a processing procedure in an embodiment of the present invention. FIG. 6 is a diagram illustrating an example of an interface screen in an embodiment of the present invention. FIG. 7 is a diagram illustrating an example of a detailed configuration in an embodiment of the present invention. FIG. 8 is a diagram explaining an example of an operation on the industrial equipment side in an embodiment of the present invention.

[0012] Hereinafter, an embodiment of the present invention will be described with reference to the drawings as an example.

[0013] 1 shows an example of a system configuration in one embodiment of the present invention. Reference numeral 1 denotes a driving assistance system, and reference numeral 2 denotes a server. A plurality of industrial devices, for example, 90A, 90B, and 90C, are connected to the server via a communication line 202. Operation data 101 is transmitted to the driving assistance system 1 via the communication line 202. Note that the server 2 may also be configured on the cloud.

[0014] The driving assistance system 1 exchanges input 102 and output 105 with the input / output means 5 via a communication line 201. Since one of the features of the present invention is that it enables driving management in an interactive format, 102 mainly takes the form of questions and 105 mainly takes the form of answers.

[0015] A generative AI engine 3 is used to create answers or interpret questions. This may be built into the server 2, or, as shown in Figure 1, a generative AI engine outside the server may be used. In this case, the question sentence 103 is transmitted from the server 2 to the generative AI engine 3 via a communication line 203, and the answer sentence 104 is received from the server 2.

[0016] In this case, there are three communication lines, for example, 210, 202, and 203. This does not mean that these communication lines have the same security measures. However, considering that there are differences in the content of communication, it is desirable that the communication line 202, which handles communication between the industrial equipment and the server, has the highest security level.

[0017] This is because the communication line directly transmits data on the operating status of the equipment, including management data directly related to company management, such as production status and production busyness within the factory. For this reason, it is desirable to use a dedicated line that is not connected to the outside world, or to apply dedicated encryption, etc.

[0018] Of course, this does not mean that the security level of other communication lines can be low, but when considering the balance with costs, the range of security measures used for normal Internet lines is practically appropriate.

[0019] It should be noted that throughout this specification, the multiple pieces of industrial equipment do not necessarily have to be the same type of industrial equipment, such as compressors. The present invention is also applicable to a mixture of different types of industrial equipment, such as an air compressor 90A, an inkjet printer 90B, and a pump 90C. While a mixture of such multiple types of industrial equipment traditionally required skilled personnel for operational management, the present invention is characterized by and has a major advantage in that it enables even non-skilled personnel to efficiently manage operations, even in such an environment.

[0020] Therefore, the target industrial equipment is not particularly limited, but it can be applied to a wide range of industrial equipment, such as air compressors, inkjet printers, pumps, air purifying equipment, transport equipment such as hoists, power receiving and transforming equipment, etc. Furthermore, this operation support system is extremely well suited for application to remote monitoring services, maintenance services, etc.

[0021] FIG. 2 is a diagram showing a more detailed configuration example of the present invention based on FIG.

[0022] The communication between the server 2 and the input / output means 5 is shown in a bidirectional format as 110. The communication between the industrial devices 90A, 90B, 90C and the server 2 is shown in a bidirectional format as 111. However, this does not exclude the case of one-way communication from the industrial devices to the server 2.

[0023] Two-way communication 112 is established between the server 2 and the generation AI engine 3.

[0024] 4 is a voice synthesis engine. In the present invention, the use of a voice synthesis engine is not essential, and questions and answers can also be entered and answered by inputting and outputting text information. However, in consideration of the convenience of the operations manager, allowing answers by voice reduces the impression that the operations manager has of monotonous work, contributing to an improvement in the working environment of operations managers and leading to increased motivation for work among inexperienced operations managers. Examples of operations managers include users of industrial machinery or employees of companies that maintain industrial machinery.

[0025] The speech synthesis engine 4 is bidirectionally connected to the server 2 via a communication line 203 and bidirectional communication 113. In FIG. 2, the generative AI engine 3 and speech synthesis engine 4 are shown as being located outside the server 2, i.e., on a separate server or cloud. This is because the generative AI engine and speech synthesis engine are intended to utilize the resources of external service providers that are more focused on these services. This makes it possible to more efficiently utilize development resources for the industrial equipment operation assistance system.

[0026] Of course, this does not exclude the possibility of implementing them on the same server. Figure 4 shows an example in which the generation AI engine 3 and the speech synthesis engine 4 are configured on the same server 2. Alternatively, only one of them, for example, only the speech synthesis engine 4, may be configured on the same server 2.

[0027] Returning to FIG. 2, the description will continue.

[0028] The CPU 10 is, for example, a CPU of a server, which also includes a virtualized CPU or computing means.

[0029] The CPU 10 executes a control agent 12 and device agents 13. A major feature of the present invention is that the agents are divided into at least two stages as described above.

[0030] The inventors of this application, who have studied the application of AI to the operation management of industrial equipment, have found that when a single agent is used to dialogue with and respond to operations managers and to grasp the status of industrial equipment, the following problem occurs: The agent has to process a huge number of objects, and the content of the operations manager's questions gets buried in the huge amount of data on operation information from the industrial equipment, making it impossible to provide an appropriate response or response.

[0031] Absolute reliability is required when applying AI to operation assistance systems for industrial equipment. Hallucination, a phenomenon currently recognized as a major issue with generative AI (where an AI responds as if something that is not true were true), could potentially lead to a major accident if it occurs in an operation assistance system for industrial equipment.

[0032] The inventors of the present application have intensively researched a method for reliably avoiding hallucination when using AI in an operation support system for industrial equipment. As a result, they have come to the realization that, in order to be able to subdivide the dialogue with the operation manager and limit the content of the dialogue, when using AI in an operation support system for industrial equipment, it is important to divide the AI ​​into multiple parts.

[0033] Therefore, in this invention, by dividing the AI ​​into multiple AIs, namely, a control agent and equipment agents, hallucination can be avoided and a highly reliable industrial equipment operation support system can be realized.

[0034] At the same time, this configuration also clarifies the procedure for the operation manager to ask questions to the operation support system, making it possible for even non-expert operation managers to handle the system.

[0035] An example of the configuration at this time can be expressed as follows, for example.

[0036] An industrial equipment operation support system comprising: a server that can access information relating to a plurality of pieces of industrial equipment; the server having a control agent and equipment agents that generate answers to questions received through an input means; the control agent having input thereto only a portion of the information relating to the plurality of pieces of industrial equipment for each of the plurality of pieces of industrial equipment; and the equipment agents having input thereto only information relating to some of the pieces of equipment for each of the plurality of pieces of industrial equipment.

[0037] Alternatively, it can be expressed as follows:

[0038] An industrial equipment operation support system comprising: a server that can access information relating to a plurality of pieces of industrial equipment; the server having a control agent and equipment agents that generate answers to questions received through an input means; the control agent generating answers that identify which of the plurality of pieces of industrial equipment meets the conditions of the questions asked; and the equipment agents generating answers relating to the corresponding individual pieces of equipment.

[0039] An example of the specific dialogue between the operation manager and the industrial equipment operation support system is as follows: These may also be called prompts.

[0040] The following is an example of an operation support system that manages multiple compressors as industrial equipment. Naturally, the response text will differ depending on the connected device.

[0041] (Question 1) Who are you?

[0042] This is an example of a question for identifying the dialogue target. By identifying and confirming the dialogue target, this question serves as a fail-safe measure to avoid sending erroneous instructions or questions to another system.

[0043] (Answer 1) I am an operation support system that manages compressors.

[0044] In this case, it is desirable to give the driver assistance system a more specific name. This is because it makes it easier for the driver management to recognize the system, and it is expected that the driver management will feel a sense of familiarity with the system and become more enthusiastic about their work. For example, a name such as "Aias" (meaning "I" or "we") would be appropriate.

[0045] (Question 2) Please tell me the operating status of the compressor.

[0046] In this case, if multiple types of industrial equipment are connected, a question such as "Please tell me the operating status of the industrial equipment" can be used. Alternatively, if you want to know the status of a specific type of industrial equipment, you can use a question that includes the name of the specific industrial equipment, as in Question 2.

[0047] (Answer 2) There is a malfunction in Unit 1. There are signs of an abnormal discharge temperature in Unit 2. Unit 3 is operating normally.

[0048] In this way, the control agent is specialized in receiving and answering questions about the general situation. This eliminates ambiguous interpretations of the questions and enables accurate answers. In the above example, the devices are numbered 1, 2, and 3, but to ensure that the target devices are identified later on-site, the device may be specified in more detail, such as compressor serial number U1234567.

[0049] (Question 3) I want to talk to Unit 2.

[0050] Here, by inputting the question, the dialogue shifts from the control agent to the appliance agent. In other words, the conversation target is handed over from the control agent to the appliance agent.

[0051] (Answer 3) I am compressor number 2.

[0052] The control agent declares that the dialogue has been transferred to the device agent.

[0053] (Question 4) What is the cause of the abnormal discharge temperature?

[0054] (Answer 4) This is caused by factors such as a rise in ambient temperature.

[0055] The equipment agent 13 examines the cause of the abnormality or the signs of such an abnormality from the operation data of the equipment, etc., and responds to the operation manager. In this case, it is necessary to install an equipment diagnostic engine 14 on the server 2 or in cooperation with an external service. This equipment diagnostic engine uses operation information from industrial equipment and devices to output diagnostic information on failures, alarms, and their signs, and provides this diagnostic information to the equipment agent 13 or the control agent 12.

[0056] Furthermore, the device diagnosis engine 14 may have a predictive diagnostic engine or the like installed inside or outside it, and may perform device diagnosis in cooperation with it. Alternatively, the device diagnosis engine itself may be configured to determine failures, alarms, and their predictive signs.

[0057] (Question 5) Please tell me how to deal with rising ambient temperatures.

[0058] Since this is a sentence asking about countermeasures, it would be better to be more direct and say, "Please tell me the countermeasures," or "Please tell me the solution." One of the advantages of using AI is that it can respond to ambiguous expressions.

[0059] (Answer 5) Please ventilate the room so that the ambient temperature remains below 45 degrees.

[0060] This provides a specific response. At this time, the driving assistance system 1 has instruction manuals, operation manuals, etc. stored in a database 17. For example, this may take the form of a manual database 18. The equipment agent refers to this manual database, considers and presents a corresponding solution.

[0061] As described above, the process is to first grasp the general situation through dialogue with the control agent, and then to discuss the detailed situation and solutions through dialogue with the device agents.

[0062] In this way, repeated dialogue makes it possible to understand the operating status of equipment and how to respond when an abnormality occurs. Maintenance information can also be included in the scope of understanding.

[0063] Then, based on the failures and warnings and their predictive judgments made by the equipment diagnosis engine 14, the equipment agent 13 prepares the contents of a probable cause or a countermeasure, or both, to be sent to the operation manager.

[0064] In this case, in the present invention, since the equipment agent is specialized in grasping the status of the equipment and deriving solutions, it is possible to preemptively eliminate questions and learning that are unrelated to the operation management of industrial equipment.

[0065] For example, the control agent may declare the end of the conversation by saying, "Such questions are outside the scope of this system," or the equipment agent may declare that it will not respond by saying, "This system cannot answer such questions," or "Such questions are unrelated to this industrial equipment."

[0066] This also eliminates the risk of incorrect information being learned by mistake, making it possible to ensure the reliability of the driving assistance system.

[0067] In this case, the basis for formulating the answer plan is based on pre-verified texts and descriptions in the database 17, such as the manual database 18. In addition, the supplementary learning case database 19 stores only correct case studies for the operation and management of industrial equipment, eliminating the need to learn random information.

[0068] This makes it possible to avoid the occurrence of hallucination in principle, because incorrect or ambiguous information that causes hallucination is not accumulated in the first place.

[0069] In this way, the industrial equipment operation support system of the present invention can provide an industrial equipment operation support system that utilizes AI, is highly reliable, avoids the occurrence of hallucination, is easy to use for operation managers, and does not require expertise.

[0070] Even if all industrial equipment is operating normally, by changing the question text, it is possible to grasp the lifespan of the equipment and the time until the next maintenance work through questions and answers about the total operating time, for example. Also, for industrial equipment that has consumable parts that have a finite lifespan and need to be replaced at a certain interval, questions and answers about the remaining lifespan and the time until replacement make it possible to plan maintenance and part replacement. In other words, maintenance information can also be provided.

[0071] Returning to FIG. 2 again, the description will be continued.

[0072] The learning case database 19 can be constructed by inputting information from an external source, or in the example of Figure 2, a machine learning engine 15 can be installed in the server 2 so that appropriate response cases can be added to the learning case database. This makes operation management easier the longer the operation continues, making it possible to prevent malfunctions before they occur.

[0073] 2 also shows a memory 16. Data being processed by the CPU 10, and various programs such as the control agent 12 and device agents 13 executed by the CPU 10 are actually loaded into this memory 16 and executed by the CPU 10.

[0074] The server 2 utilizes the generation AI engine 3 as necessary when interpreting input between the control agent 12 and the input / output means and creating output statements. Similarly, the equipment agents 13 utilize the generation AI engine 3 as necessary when interpreting input between the input / output means and creating output statements. In this case, the generation AI engine is used solely for interpreting sentences and creating natural sentences. The identification of the target model of dialogue by the control agent 12, and the presentation of problems, probable causes, solutions, etc. by the equipment agents 13, which require precision, are performed solely by a dedicated system of this operation management support system.

[0075] The use of the generative AI engine 3 is strictly limited to the purpose of natural dialogue. In this way, by clearly separating the AI ​​system and the content of use between areas that require precision and others, it is possible to prevent halcyonosis and ensure reliability in the operation support system for industrial equipment.

[0076] The dialogue between the input / output means and the server 2 can also be based on character information, such as keyboard input by the operation manager and character display on the output screen.

[0077] As an example, in order to further foster motivation among driving managers and ensure a clear understanding of the situation, the example in Fig. 2 is provided with a text-to-speech engine 11. This allows, for example, a driving manager to speak into a microphone, which the text-to-speech engine 11 converts into text information, allowing a dialogue with a control agent or an equipment agent to begin. This opens the door for people with physical disabilities who have difficulty inputting text information to become driving managers, thereby promoting the respect and utilization of human resources.

[0078] Furthermore, when outputting, in addition to displaying the answer as text information on the display, it is also possible to convert it into audio information through the text-to-speech engine 11 and output the audio from the speaker in the input / output means 5. This also helps to foster motivation in the operation manager, and by changing the tone and volume of the voice depending on the content, it is also possible to add additional information such as the urgency of the response.

[0079] Furthermore, for example, when using many different types of industrial equipment, by using different tones of voice for each piece of industrial equipment, such as high or low tones, or a male or female voice, the operations manager can more intuitively grasp the target. In this case, the target equipment group can be grasped using both text information and audio information, providing the effect of a double-check function.

[0080] Of course, when a plurality of compressors are targeted, the audio may be different for each of the compressors.

[0081] The target of the different voices can be the control agent 12 and the appliance agents 13. For example, different tones can be used for the control agent 12 and the appliance agents 13.

[0082] The conversion of speech to text or text to speech in the character-to-speech engine can be performed by a dedicated system on the server 2. However, from the perspective of more natural speech, as shown in Figure 2, by linking with the speech synthesis engine 4 and forming a waveform of the speech output, it becomes possible to make the speech sound more similar to a conversation with a person without being aware of AI.

[0083] Part of the above explanation can be expressed in other words as follows:

[0084] <No. 1> An industrial equipment operation support system comprising: a server that can access information relating to a plurality of pieces of industrial equipment; the server having a control agent and equipment agents that generate answers to questions received through an input means; the control agent having input thereto only a portion of the information relating to the plurality of pieces of industrial equipment for each of the plurality of pieces of industrial equipment; and the equipment agents having input thereto only information relating to some of the pieces of equipment for each of the plurality of pieces of industrial equipment.

[0085] <Item 2> The operation support system for industrial equipment according to item <Item 1>, wherein the input of information to one or both of the control agent and the equipment agents is a prompt.

[0086] <No. 3> In the industrial equipment operation support system described in <No. 1>, when the server receives a request for dialogue with a specific equipment agent from the input means, it is possible to start a dialogue between the equipment agent and an operation manager.

[0087] <No. 4> In the industrial equipment operation support system described in <No. 3>, when the server receives an inquiry from an operation manager to check the status of the specific equipment agent, it causes the equipment agent to create a response regarding the operation status of the corresponding industrial equipment and responds to the operation manager.

[0088] <Item 5> The operation support system for industrial equipment according to item <Item 4>, wherein the operation status includes a sign of a failure or an alarm, or maintenance information.

[0089] <No. 6> In the industrial equipment operation support system described in <No. 1>, the server receives questions about countermeasures from the operation manager, causes the equipment agent to prepare a response about the estimated cause or a response about countermeasures, and responds to the operation manager.

[0090] <No. 7> In the industrial equipment operation support system described in <No. 1>, the server has an equipment diagnostic engine, and the equipment diagnostic engine uses operation information from the industrial equipment to output diagnostic information on failures, alarms, and their precursors, and inputs the diagnostic information to the control agent or the equipment agent.

[0091] <No. 8> In the industrial equipment operation support system described in <No. 7>, the server has an instruction manual database, and the equipment agent creates the contents of the estimated cause or countermeasures to be sent to the operation manager based on the contents of the instruction manual database and the judgment of failures, alarms, and their precursors by the equipment diagnosis engine.

[0092] <No. 9> In the industrial equipment operation support system described in any of <No. 1> to <No. 8>, the server works in conjunction with an internal or external generation AI engine, and interprets questions between the control agent or the equipment agent and the operation manager, and creates answers to the operation manager.

[0093] <No. 10> In the industrial equipment operation support system described in <No. 4>, the server works in conjunction with an internal or external speech synthesis engine to convert into voice responses from the control agent or the equipment agent to the operation manager.

[0094] <Item 11> The industrial equipment operation support system according to <Item 8>, wherein the server has a machine learning engine and accumulates corresponding cases in a learning case database.

[0095] <Item 12> The industrial equipment operation support system according to <Item 10>, wherein the voice generation uses different tones for the control agent and the equipment agents.

[0096] <No. 13> In the industrial equipment operation support system of <No. 9>, the level of security of communication between the server and the industrial equipment is higher than that of communication between the server and the input means and output means, and communication between the server and the generating AI engine.

[0097] <Item 16> The industrial equipment operation support system according to <Item 11>, wherein the machine learning engine learns only events related to the industrial equipment.

[0098] <No. 17> An industrial equipment operation support system having a server that can access information related to a plurality of pieces of industrial equipment, the server having a control agent and equipment agents that generate answers to questions received through an input means, the control agent generating answers that identify which of the plurality of pieces of industrial equipment meets the conditions of the questions asked, and the equipment agents generating answers related to the corresponding individual pieces of equipment.

[0099] This embodiment is basically the same as the first embodiment and Fig. 2. The differences will be explained below with reference to Fig. 3.

[0100] In this embodiment, a maintenance terminal 50 is provided. This maintenance terminal performs two-way communication 120 with the server 2 of the driving assistance system 1 via a communication line 202. The maintenance terminal 50 is a terminal different from the input / output means 5, such as a tablet or a PC installed in a factory. The maintenance terminal 50 can also be provided integrally with the industrial equipment 90A, 90B, 90C, etc.

[0101] As a result, if the operations manager becomes aware of an abnormality or detects signs of an abnormality through dialogue between the operations manager and the control agent 12 and the equipment agent 13, he or she can instruct factory workers and maintenance personnel to carry out maintenance and repair work via the operations support system 1.

[0102] In this case, as one embodiment, a response determination engine 20 can be provided in the server 2. This engine is responsible for making the final decision on the response work and issuing detailed instructions in a situation where the driving assistance system 1 is to be entrusted with the decision and instructions.

[0103] Reference numeral 51 denotes a parts supplier. In this example, the driving assistance system 1 sends an order to the parts supplier 51 via the communication line 201 to order and arrange for the parts required for the response. Note that although the figure refers to a parts supplier, this also includes cases where the parts are in stock in a factory or warehouse, or in stock at another factory.

[0104] An example of the flow of driving assistance performed by the driving assistance system of the present invention will be described below in conjunction with the first embodiment using the flowchart of FIG.

[0105] In step S01, the driving assistance system starts a dialogue between the driving manager and the driving assistance system. At this time, as described in the first embodiment, it is desirable to first accept a question input to identify the system with which the dialogue will be held and present the answer to the driving manager.

[0106] In step S02, the driving assistance system accepts an inquiry input about the driving situation as a character input or a voice input from the input / output unit 5 shown in FIG.

[0107] In step S03, the driving assistance system outputs a response regarding the driving situation, either as text information on a display or as voice, or both, via the input / output means 5 shown in Fig. 2 or 3.

[0108] In step S04, the driving assistance system receives an inquiry about the status of a specific machine. Here, the conversation target shifts from the control agent 12 in Fig. 2 or 3 to the device agent 13 for the specific machine that has been inquired about.

[0109] In step S05, the driving assistance system outputs a response about the specific vehicle's status. The output text is created mainly by the device agent 13.

[0110] Here, an example of a display screen using a web browser, which is one form of the input / output means 5 of the present invention, will be described with reference to FIG.

[0111] The display screen 60 is, for example, a part of the area of ​​a web browser. Of course, this does not exclude the use of a dedicated screen.

[0112] When the driver inputs a question in the input area 62 on the dialogue content log screen 61, the reply to the question is displayed in text on the log screen 61 in addition to the past dialogue content.

[0113] When a button 63 for selecting a new conversation is clicked, the log screen 61 is reset and a new conversation such as a question and answer can be started again.

[0114] Selection area 64 is a selection area for whether or not to permit operation of the operation assistance system on the maintenance terminal side. As with the maintenance terminal 50 in Figure 3, it is desirable that on-site workers and maintenance personnel, in addition to the operation manager, can also access the operation assistance system to facilitate operation assistance and, particularly, to implement countermeasures for abnormalities and their precursors. However, if unlimited access or unlimited questions are required, there is a concern that inappropriate information may be learned, leading to an increased risk of halucation.

[0115] Therefore, by allowing the operation manager to select whether to avoid operations on the maintenance terminal in the selection area 64, it is possible to permit access only when actual work such as maintenance or repair is required and access to the operation assistance system from the site is necessary. This not only avoids hallucination, but also minimizes security risks and the risk of interference with the operation assistance system that may arise from unauthorized access to the maintenance terminal.

[0116] The selection area 66 switches the audio mode between ON and OFF.

[0117] As mentioned above, there are advantages to using voice when a driver manager responds to a driver assistance system. However, for example, if an emergency situation occurs and the surrounding area is noisy, using voice can lead to entering an incorrect question, the question not being recognized correctly, or the voice response making the noise worse and hindering the situation from being resolved. In such cases, by selectively turning off the voice mode, i.e., by not using voice and responding only with text information, the driver assistance system can be used effectively even in an emergency.

[0118] Selection area 65 is a selection area for turning automatic response ON or OFF. In operation support for industrial equipment, there are usually few situations in which unexpected abnormality signs are identified. Rather, it is thought that the system is more likely to identify and report the arrival of part replacement work due to the end of the lifespan of periodic parts, or adjustments to operating parameters or ambient temperature due to air conditioning operation. For such signs and abnormalities for which a response policy has been firmly established in advance, there are many situations in which the operation support system can automatically determine and respond to the policy, without the operation manager having to make a decision each time.

[0119] Selection area 65 is an area for selecting whether or not to allow the driving assistance system to respond automatically. If ON is selected here, the driving assistance system will automatically respond to abnormal signs that are necessary and can be responded to routinely, without the intervention of the driving manager.

[0120] This automatic response may be set to be always enabled while the driving assistance system is in operation and ON is selected in the selection area 65. In addition, in the flowchart of Fig. 5, the ON / OFF of the automatic countermeasure mode is determined after proceeding to step S05, which is the output of the situation response of the specific machine, but the position of this determination in the flowchart is not limited to this.

[0121] Returning to the flowchart of FIG. 5, the explanation will be continued.

[0122] In step S06, the driving assistance system determines whether the automatic countermeasure mode is ON. If the determination is NO, that is, if the driving manager has not authorized the driving assistance system to take automatic countermeasures, the process proceeds to step S07.

[0123] In step S07, the driving assistance system receives an inquiry about countermeasures from the driving manager.

[0124] In step S08, the driving assistance system outputs a response regarding countermeasures to the driving manager.

[0125] Furthermore, through the continuation of the dialogue, the operation assistance system receives instructions for responding to the actual abnormality signs according to the content of the dialogue, and transmits detailed responses to various parties. At that time, for example, if it is necessary to order parts or replacement parts, the system can place an order with the parts supplier 51 via the communication path 121 in Fig. 3. Furthermore, if on-site work is required, the system can instruct the maintenance terminal 50 on the work content and adjustment execution via the communication path 120 in Fig. 3. Furthermore, if it is necessary to adjust the operating parameters of the industrial equipment 90A, 90B, 90C, etc., the system can receive instructions from the operation manager and directly control the parameters of the industrial equipment via the communication path 111.

[0126] An example in which the automatic countermeasure mode is ON will be described in the branch from step S06.

[0127] In step S11, it is determined whether or not a countermeasure is necessary. If there are no signs of an abnormality, or if signs of an abnormality are detected but are not yet at a level where countermeasures are not necessary, the determination in step S11 is NO, and the flow chart loops and the countermeasure process returns to the previous step.

[0128] In the example of FIG. 3, such a determination is made by a response determination engine 20 deployed on the server 2.

[0129] If it is determined that a response is necessary, a countermeasure is considered in step S12. This is also performed by the response determination engine 20.

[0130] Next, it is determined in step S13 whether additional parts are required and whether parts need to be procured. Even if a part needs to be replaced, if the part is in stock, it is determined that there is no need to procur an additional part.

[0131] If it is determined that additional parts are not required, a response instruction is issued in step S14. As explained with reference to FIG. 3, the operation assistance system 1 displays the response, work, or replacement details on the maintenance terminal 50, and instructs the site on specific response. Note that if the target industrial equipment is compatible with replacement or adjustment work by a robot or automatic work machine, instructions may be issued to the robot or automatic work machine. In this case, the work details may be directly communicated without using text or voice.

[0132] If it is determined in step S13 that additional parts are required, an order for the parts is placed in step S15. Explained with reference to FIG. 3, this involves issuing an ordering instruction to the parts supplier 51 via the communication path 121. This may also include ordering work. Furthermore, if parts are in stock at another factory of the company, an instruction to source the parts from there may also be included.

[0133] 3 shows the generation AI engine 3 and the speech synthesis engine 4 configured outside the server 2, i.e., on a different server or a different cloud. However, the location of the generation AI engine and the speech synthesis engine is not essential to the present invention. Therefore, the present invention also includes the implementation of the generation AI engine 3 and the speech synthesis engine on the same server 2 or the same cloud, on a different CPU 30 or CPU 10, as shown in FIG. 4.

[0134] In Example 1, an example of the technical idea of ​​the present invention was explained from <No. 1> to <No. 14>. However, if an additional technical idea is explained based on the configuration disclosed in this example, it will be as follows.

[0135] <No. 14> In the industrial equipment operation support system described in <No. 4>, when an automatic response mode is selected, the server determines whether any of the plurality of industrial equipment requires response, and if response is required, determines whether additional parts need to be arranged, and if additional parts are not required, gives instructions to a worker to respond or adjusts the operating parameters of the industrial equipment, and if additional parts are required, arranges for the parts.

[0136] <No. 15> In the industrial equipment operation support system described in <No. 8>, the server is capable of switching whether to give instructions to the control agent from a maintenance worker terminal or each piece of industrial equipment.

[0137] The present embodiment is basically the same in concept as the first and second embodiments. Here, the first and second embodiments are explained with a focus on the hardware configuration or the function engine realized on the hardware. In contrast, the present embodiment explains an embodiment of the technical concept of the present invention with a focus on the flow of data processing.

[0138] This will be explained using FIG.

[0139] The user terminal 500 includes, for example, a microphone 510 as an input device, a keyboard 511 as an input device, a display 512 as a character / text output device, and a speaker 513 as an audio output device.

[0140] The user interface 514 accepts text input 515 from the keyboard 511. This may be a question from the driver's manager, or may also be voice input from the microphone 510 that has been converted into text.

[0141] The industrial equipment operation support system 600 is configured as hardware on a server or on the cloud.

[0142] The industrial equipment operation support system 600 includes a master system 601. This corresponds to a control agent or a part of the control agent.

[0143] The text input 515 is input to the input reception 602 as user utterance text 800 and passed to the generation AI call unit 603. In response to the input content, the user utterance text 801 is transmitted to the generation AI prompt generation process 612, and a response is made by the control agent or the device agent.

[0144] It may also be linked to the generation AI service 700. Because many commercial generation AI services are already available, when creating natural dialogue sentences, it is more efficient to utilize an external generation AI service to realize advance dialogue.

[0145] Prompt 830 is a prompt input to the generation AI engine 701 from the generation AI prompt generation process.

[0146] The generated AI utterance text 831 is the generated AI utterance text sent from the generated AI engine 701 to the generated AI prompt generation process 612.

[0147] From the generated AI prompt generation process 612, the generated AI speech text 802 is input to the generated AI call unit 603 as an answer.

[0148] When using voice, the generated AI speech text is transmitted from the generation AI call unit 603 to the voice synthesis call unit 604. When not using voice, the voice synthesis call unit 604 is skipped and the text is transmitted directly to the output management unit 605.

[0149] The speech synthesis invocation unit 604 passes the generated AI spoken text 806 to the speech synthesis interface 615 .

[0150] The speech synthesis interface 615 interacts with the speech synthesis engine 702 to generate synthetic speech.

[0151] The generated synthetic speech 805 is passed through the speech playback process 519 as a speech log 616 in the form of synthetic speech 807 and is output as speech by the speaker 513 .

[0152] The output management unit 605 mentioned above transmits the generated AI speech text 808 as display data 517, which is displayed, for example, in the relevant area of ​​the web browser using the display 512.

[0153] Furthermore, the output management unit stores a conversation log as a chat log 617, which is referenced and displays the past log when an instruction to call up the log is received from the user interface 514.

[0154] In particular, the generated AI prompt generation process 612 involves the equipment management agent inferring the occurrence of an abnormality or signs of an abnormality based on the input operation data 610 of each industrial equipment in a specific AI inference unit 611, and transmitting the sign diagnosis judgment result 803 to the generated AI prompt generation process 612.

[0155] In particular, the crisis management agent evaluates the correlation with the current event, the degree of matching, etc. in the vector database 613 based on the instruction manual text and past cases 614, and passes appropriate content as countermeasures to the AI ​​prompt generation process 612.

[0156] Note that 804 is RAG (Retrieval-Augmented Generation). This process retrieves facts from a database of information, knowledge, case studies, etc., and has the generation AI create an answer based on accurate information. The device agent cooperates with the generation AI prompt generation process to create an answer while referencing the vector database 613, instruction manual text 614, etc. At this time, cooperation with the generation AI engine 701 is also performed in order to create natural-sounding answer sentences.

[0157] In the above, the present embodiment has been described as an example of the technical concept of the present invention, focusing on the flow of data processing. However, as long as the technical concept disclosed throughout this specification is used, even if the details of the description in this embodiment differ from those in part of the configuration or the order of the configuration, it is still within the scope of the disclosure of this embodiment.

[0158] Furthermore, as long as the technical concepts detailed in the above-described embodiments are applied, modifications and slight differences in construction are also included within the scope of the present invention.

[0159] Furthermore, although this specification was written primarily with industrial equipment operation assistance systems in mind, the methodology for avoiding hallucination in fields requiring high reliability of AI may also be applicable to other fields. Therefore, as long as at least a portion of the technical ideas disclosed in this specification is used, and if it is an independently patentable technical idea, even if it is applied to something other than industrial equipment, it is within the scope of the present invention and is claimed as a right.

[0160] Possible examples include operation management of railways, public transport, commercial vehicles, aircraft, etc.; maintenance management of power transmission and distribution systems; operational support for power generation systems that parallel multiple power generation sources such as solar and wind power; and production management support for production systems such as factories.

[0161] In addition, the term "driving manager" in this specification is used in a broad sense to refer to any person whose job requires him or her to ask and answer questions to the driving assistance system.

[0162] This embodiment is basically the same as the first embodiment and FIG.

[0163] This embodiment is characterized in that a function for calling a generated AI is provided to the industrial equipment described in the first and second embodiments. An example will be described below with reference to FIG.

[0164] 8, the industrial equipment 301 corresponds to 90A, 90B, 90C, etc. in FIGS. 1 and 2. All of these industrial equipment may have the configuration described below. Also, cases where some of 90A, 90B, 90C, etc. have the configuration described below are also included in the scope of this embodiment.

[0165] The industrial equipment 301 is characterized by having a function for calling a generated AI. Specifically, the industrial equipment 301 is characterized by having a generated AI call button 302. The generated AI call button 302 may be a physical button or a virtual button.

[0166] For example, the operation manager can speak a specific word that serves as a wake word, such as "tell me about the compressor," and the control unit 303 can recognize this voice to replace the generated AI call button. Also, codes containing connection destination information, such as two-dimensional or three-dimensional codes affixed to the industrial equipment 301, can be read by the operation manager's smartphone, tablet device, or the like to replace the generated AI call button.

[0167] The two-dimensional or three-dimensional code may be attached nearby rather than directly to the industrial equipment 301. The industrial equipment 301 may be a crane or hoist for transportation, and in such cases may be installed at a high place. In this situation, by attaching the two-dimensional or three-dimensional code nearby rather than directly to the industrial equipment 301, the operation manager can communicate with the operation assistance system 1 without having to go to the trouble of climbing up to a high place. This has the effect of ensuring the safety of the operation manager and allowing him or her to safely communicate with the operation assistance system 1.

[0168] Furthermore, when an abnormality or a sign of an abnormality occurs in the industrial equipment 301, the control unit 303 can display on the display unit 305 a suggestion to the operation manager to call the operation assistance system 1. In this case, it is desirable that the industrial equipment 301 has a display as the display unit 305, but this does not exclude the use of other means such as a lamp or audio communication means.

[0169] In this way, the operation manager can issue instructions at the necessary timing to the operation assistance system 1. This has the effect of reducing downtime of the industrial equipment 301 by performing appropriate maintenance of the industrial equipment 301.

[0170] Furthermore, when the industrial equipment 301 is equipped with an image recognition means such as a camera, instructions can be given by detecting the motion of the operation manager's hand signals or movements.

[0171] When the control unit 303 detects that the generation AI call button 302 has been pressed, it displays a two-dimensional or three-dimensional code on the display unit 305. The control unit 303 acquires the two-dimensional or three-dimensional code data from the server 2 via the communication unit 304. The driving manager can interact with the driving assistance system 1 by reading the two-dimensional or three-dimensional code with a smartphone, tablet device, or the like.

[0172] That is, by reading the two-dimensional or three-dimensional code, it is possible to access the web page on which the driving assistance system 1 operates.

[0173] The industrial equipment 301 may be a small air compressor, a crane or a hoist for transportation, and in such cases may be installed at a foot or in a high place. In such cases, by using a smartphone or tablet device of the operation manager at hand to interact with the operation assistance system 1, it is possible to have the effect of interacting with the operation assistance system 1 in a natural posture even if the industrial equipment 301 is installed in a place that is difficult to access.

[0174] It goes without saying that the industrial equipment 301 can be provided with a keyboard so that the operation manager can use the keyboard as an input means and the display unit 305 as an output means, which together form the input / output means 5. In this way, it is not necessary for the operation manager to own a smartphone, and there is an effect that the operation assistance system 1 can be easily used.

[0175] Furthermore, it goes without saying that a microphone and speaker provided on the industrial equipment 301 can be used as the input / output means 5. In this case, the voice picked up by the microphone is converted into text by voice recognition and input to the driving assistance system 1. By performing voice dialogue in this way, it is not necessary for the driving manager to prepare a smartphone, and it is also not necessary to install a keyboard on the industrial equipment 301. In this case, it is possible to easily use the driving assistance system 1, and there is an effect that the industrial equipment 301 can be made smaller.

[0176] 1 and 2 is preferably controlled so that the operations manager can interact with the equipment agent 13 of the industrial equipment 301. In this way, it is possible to provide a dialogue that meets the operations manager's expectations. In addition, by configuring the dialogue so that it can be intuitively understood by the operations manager, it is possible to reduce the learning cost required for the operations manager to be able to use the operations assistance system 1.

[0177] When the driving manager issues an instruction to switch to the control agent, the driving manager becomes able to converse with the control agent 12. One example of a method of issuing an instruction is for the driving manager to input a sentence such as "I want to talk to the control agent" into a smartphone or tablet device, which is the input / output means 5 in this embodiment. Note that a similar switching instruction may also be issued by voice input or GUI.

[0178] It is desirable that the control agent 12 has shallow information about a plurality of industrial devices 301. On the other hand, it is desirable that the device agents 13 have deep information about individual industrial devices 301.

[0179] The control agent 12 limits information to shallow information, allowing the operation manager to know the operating status of each piece of equipment in short sentences, while the equipment agent 13 has deep information, allowing the operation manager to obtain a wide range of information about the equipment.

[0180] Specifically, the knowledge that the control agent 12 should have is preferably product instruction manuals, information on alarms and malfunctions (abnormality information) of the multiple industrial devices 301, and information on the presence or absence of signs of abnormality.

[0181] On the other hand, it is desirable for the equipment agent 13 to have the product's instruction manual, operation data for the industrial equipment 301 that the equipment agent 13 is responsible for, and information on whether or not there are any signs of anomalies. The operation data includes the presence or absence of anomalies and their types, as well as regularly transmitted data. The regularly transmitted data includes the discharge temperature, discharge pressure, and ambient temperature for a compressor, the remaining ink level and ambient temperature for an industrial inkjet printer, and the current value and ambient temperature for a water supply pump. It is desirable for this data to be acquired periodically, such as every 30 minutes, and transmitted to the server 2.

[0182] Here, it is desirable that the scheduled transmission data undergoes statistical processing in the generation AI prompt generation process 612 in Figure 7 before being input to the generation AI engine 701. Statistical processing involves averaging and determining maximum, minimum, and median values. It is also desirable to aggregate the compressor load factor only when the compressor is in a loaded state (when compressed air is being sent out).

[0183] By performing statistical processing, it is possible to reduce the number of characters input to the generation AI engine 701, thereby obtaining a more accurate answer. The generation AI is limited in the number of characters that can be input, and can only handle input of, for example, about 30,000 characters. This is because it is known that the accuracy of the answer decreases as the number of input characters increases.

[0184] The operations manager can grasp the operating status of the industrial equipment 301 by interacting with the equipment agent 13. Furthermore, by switching to the control agent 12 and interacting therewith, the operations manager can grasp the operating status of the multiple pieces of industrial equipment 301 managed by the operations manager in a short time. Furthermore, even if not all of the industrial equipment 301 managed by the operations manager is compatible with the operation assistance system 1, as long as at least one piece of industrial equipment is compatible, the operations manager can grasp the operating status of the other pieces of industrial equipment 301 through interaction with the control agent 12.

[0185] As described above, by providing the generated AI call button 302 on the industrial equipment 301, the operation manager can intuitively and quickly call up the operation assistance system 1 when he or she wants to interact with the operation assistance system 1, which has the effect of making it easier to receive support from the operation assistance system 1.

[0186] It is more desirable to be able to interact with other industrial equipment 301 in addition to the equipment agent 13 in charge of the industrial equipment 301 having the generation AI call button 302 .

[0187] Here, an example will be described in which the serial number of the industrial equipment 301 having the generation AI call button 302 is A, and the serial number of the other industrial equipment 301 is B. For example, an operations manager can enter a sentence such as "I would like to talk to the equipment agent with serial number B" into a smartphone or tablet device, thereby interacting with the other industrial equipment 301. In this way, the operations manager can obtain detailed information about the industrial equipment 301 with serial number B without having to go from in front of the industrial equipment 301 with serial number A to in front of the industrial equipment 301 with serial number B.

[0188] To reiterate, as long as the technical concepts detailed in the above-described embodiments are applied, modifications and slight differences in structure are also included within the scope of the present invention.

[0189] DESCRIPTION OF SYMBOLS 1: Driving assistance system 2: Server 3: Generating AI engine 4: Speech synthesis engine 5: Input / output means 10: CPU 11: Character-to-speech conversion engine 12: Control agent 13: Equipment agent 14: Equipment diagnosis engine 15: Machine learning engine 16: Memory 17: Database 18: Manual database 19: Learning case database 20: Response judgment engine 50: Maintenance terminal 51: Parts supplier 60: Display screen 61: Log screen of dialogue content 62: Input area 63: New dialogue selection button 64: Maintenance terminal operation availability selection area 65: Automatic response availability selection area 66: Voice mode availability selection area 90A, 90B, 90C, 301: Industrial equipment 201, 202, 203: Communication line 302: Generating AI call button 303: Control unit 304: Communication unit 305: Display unit

Claims

1. An industrial equipment operation support system comprising: a server that can access information relating to a plurality of pieces of industrial equipment; said server having a control agent and equipment agents that generate answers to questions received through an input means; said control agent receiving only a portion of the information relating to said plurality of pieces of industrial equipment for each of said plurality of pieces of industrial equipment; and said equipment agents receiving only information relating to some of the pieces of industrial equipment for each of said plurality of pieces of industrial equipment.

2. An operation support system for industrial equipment according to claim 1, wherein information is input to one or both of said control agent and said equipment agents by prompts.

3. An industrial equipment operation support system according to claim 1, wherein the server, upon receiving a request for dialogue with a specific equipment agent from the input means, is capable of starting a dialogue between the equipment agent and an operation manager.

4. An industrial equipment operation support system as described in claim 3, wherein, when the server receives an inquiry from an operation manager to check the status of the specific equipment agent, the server causes the equipment agent to create a response regarding the operating status of the corresponding industrial equipment and responds to the operation manager.

5. An operation support system for industrial equipment according to claim 4, wherein the operating conditions include signs of failure or warning, or maintenance information.

6. An industrial equipment operation support system according to claim 1, wherein the server receives questions from the operation manager regarding countermeasures, causes the equipment agent to prepare a response on the estimated cause or a response on the countermeasures, and responds to the operation manager.

7. An operation support system for industrial equipment according to claim 1, wherein the server has an equipment diagnostic engine, and the equipment diagnostic engine uses operation information from the industrial equipment to output diagnostic information on failures, alarms and their precursors, and inputs the diagnostic information to the control agent or the equipment agent.

8. An operation support system for industrial equipment as described in claim 7, wherein the server has an instruction manual database, and the equipment agent creates a response to the operation manager regarding the estimated cause or countermeasures based on the contents of the instruction manual database and the judgment of the equipment diagnosis engine regarding failures, alarms, and their precursors.

9. An industrial equipment operation support system according to any one of claims 1 to 8, wherein the server works in conjunction with an internal or external generative AI engine to interpret questions between the control agent or the equipment agent and the operation manager, and to create replies to the operation manager.

10. An industrial equipment operation support system according to claim 4, wherein the server works in conjunction with an internal or external speech synthesis engine to convert into voice responses from the control agent or the equipment agent to the operation manager.

11. An operation support system for industrial equipment according to claim 8, wherein the server has a machine learning engine and accumulates corresponding cases in a learning case database.

12. The industrial equipment operation support system according to claim 10, wherein the voice generation uses different tones for the control agent and the equipment agents.

13. An operation support system for industrial equipment as described in claim 9, wherein the level of security for communication between the server and the industrial equipment is higher than the communication between the server and the input means and output means, and the communication between the server and the generating AI engine.

14. An industrial equipment operation support system as described in claim 4, wherein when the automatic response mode is selected, the server determines whether any of the plurality of industrial equipment requires response, and if response is required, determines whether additional parts need to be procured, and if additional parts are not required, gives instructions to a worker to respond or adjusts the operating parameters of the industrial equipment, and if additional parts are required, procures the parts.

15. An operation support system for industrial equipment according to claim 8, wherein the server is capable of switching between allowing and not allowing instructions to be given to the control agent from a maintenance worker terminal or each piece of industrial equipment.

16. An operation support system for industrial equipment according to claim 11, wherein the machine learning engine learns only events related to the industrial equipment.

17. An industrial equipment operation support system comprising: a server that can access information relating to a plurality of pieces of industrial equipment; said server having a control agent and equipment agents that generate answers to questions received through an input means; said control agent generating answers that identify which of said plurality of pieces of industrial equipment meets the conditions of the questions asked; and said equipment agents generating answers relating to the corresponding individual pieces of equipment.

Images