Grey bin model-based facility operation monitoring method and system

Abstract

The invention relates to the technical field of operation monitoring, and particularly discloses a facility operation monitoring method and system based on a grey box model, and the method comprises the steps: analyzing a received operation monitoring demand, so as to obtain a corresponding grey box model and associated monitoring data; acquiring facility simulation data based on the grey box model and the associated monitoring data; based on the facility simulation data, the associated monitoring data and the operation monitoring demand, generating an operation monitoring result through a pre-trained facility agent; wherein the facility agent is obtained based on a grey box model and facility historical data training. Through the technical scheme provided by the invention, the actual operation condition of the technological process executed by the industrial facility can be accurately described by constructing the grey box model, and the grey box model is utilized to perform agent training and cue word optimization in the agent application process, so that the intelligent analysis capability of the agent can be greatly improved.

Description

Technical Field

[0001] This disclosure relates to the field of operation monitoring technology, and in particular to a facility operation monitoring method and system based on a gray box model. Background Technology

[0002] A large-scale intelligent agent (or simply intelligent agent) is an intelligent system built upon a large-scale language model, capable of performing various tasks and interacting with users through natural language. In industrial facility applications, intelligent agents can be used to promptly diagnose and analyze on-site process / equipment faults and alarms, providing corresponding operational suggestions and decision-making support. This assists operators in monitoring and managing the operation of industrial facilities. The construction of an intelligent agent involves using general textual materials such as the process principles, operating procedures, or relevant professional books related to the specific technological processes performed by the industrial facility as training material to train the underlying large-scale language model, thus obtaining the desired intelligent agent.

[0003] However, in real-world applications, the processes executed by industrial facilities are often complex, highly coupled, and exhibit significant differences in how the same process is handled on different equipment and in different environments. Furthermore, the constructed intelligent agents are limited by hardware constraints such as deployment equipment, making it difficult to utilize a large number of model parameters to conduct a comprehensive assessment of the overall operation of the industrial facilities. Consequently, intelligent agents built solely based on general textual data of the process flow have a low level of intelligence, cannot provide users with accurate decision-making information, and are ill-suited for complex process execution scenarios such as chemical processes. Summary of the Invention

[0004] The purpose of this disclosure is to provide a facility operation monitoring method and system based on a gray box model, which can accurately describe the actual operation of the process flow performed by the industrial facility by constructing a gray box model, and use the gray box model to train the agent, which can greatly improve the intelligent analysis capability of the agent.

[0005] The first aspect of this disclosure provides a facility operation monitoring method based on a gray box model. This method may specifically include the following steps: parsing the received operation monitoring requirements to obtain the corresponding gray box model and associated monitoring data; obtaining facility simulation data based on the gray box model and associated monitoring data; and generating operation monitoring results through a pre-trained facility agent based on the facility simulation data, associated monitoring data, and operation monitoring requirements; wherein the facility agent is trained based on the gray box model and historical facility data.

[0006] In one possible implementation of the first aspect above, the facility agent is associated with the technological process performed by the facility. The facility agent is trained based on the following steps: acquiring the associated equipment and associated monitoring items of the technological process; constructing an original gray box model based on preset core monitoring items, associated equipment, and associated monitoring items; performing parameter correction on the original gray box model based on historical monitoring data to obtain a gray box model; generating simulation data corresponding to the associated equipment based on the gray box model; generating a training dataset based on historical monitoring data, simulation data, and the gray box model, and using the training dataset to train a preset base large language model to obtain the facility agent.

[0007] In one possible implementation of the first aspect above, the method further includes: storing the gray box model and its description information, the description information including associated monitoring items related to the gray box model and index tags corresponding to the gray box model, wherein the index tags also correspond to core monitoring items.

[0008] In one possible implementation of the first aspect above, the job monitoring requirements include user-inputted and / or automatically acquired warning information for core monitoring items; the process of parsing the received job monitoring requirements to obtain the corresponding gray box model and associated monitoring data includes the following steps: parsing the warning information to obtain the index label corresponding to the core monitoring item; based on the index label, obtaining the matching gray box model and associated monitoring data items related to the gray box model; and based on the associated monitoring items, obtaining the associated monitoring data.

[0009] In one possible implementation of the first aspect above, the gray box model includes at least one or any combination of multiple of the following: an equipment relationship sub-model for describing the relationship between the constituent equipment and / or equipment connection structures in the facility; a data monitoring sub-model for describing the relationship between the data acquisition equipment and / or corresponding regulation and control equipment in the facility; and a key indicator sub-model for describing the relationship between key parameters in the corresponding process flow of the facility.

[0010] In one possible implementation of the first aspect above, the facilities provided in this disclosure are used to implement a chemical process flow; the equipment relationship sub-model is constructed based on the type of constituent equipment and / or equipment connection structure, through the theoretical reaction principle corresponding to the chemical process flow.

[0011] In one possible implementation of the first aspect above, the data monitoring sub-model is constructed based on the distribution of each data acquisition device through linear fitting and / or a transfer function model; and / or the data monitoring sub-model is constructed based on the regulation and control device and the data acquisition device that has a control relationship with the regulation and control device through proportional-integral-derivative algorithm and / or a transfer function model.

[0012] In one possible implementation of the first aspect above, the process of generating operation monitoring results through a pre-trained facility agent includes the following steps: generating agent prompts based on facility simulation data, associated monitoring data, and operation monitoring requirements using a preset prompt template; generating operation monitoring results based on gray-box model prompts, wherein the operation monitoring results include one or more combinations of the following: the cause of the warning information, the actual facility operation status associated with the warning information, and the handling method of the warning information.

[0013] In one possible implementation of the first aspect mentioned above, facility intelligence is applied to facilities related to chemical process flows.

[0014] The second aspect of this disclosure provides a facility operation monitoring system based on a gray box model. This system may specifically include: a requirement parsing module for parsing received operation monitoring requirements to obtain the corresponding gray box model and associated monitoring data; a facility simulation module for obtaining facility simulation data based on the gray box model and associated monitoring data; and a result generation module for generating operation monitoring results using a pre-trained facility agent based on the facility simulation data, associated monitoring data, and operation monitoring requirements. The facility agent is trained based on the gray box model and historical facility data.

[0015] The technical solution disclosed herein can accurately describe the actual operation of the technological process executed by industrial facilities by constructing a gray box model. It utilizes the gray box model and historical data as training material for the intelligent agent. Compared to training the agent using a large amount of disordered professional process data, process descriptions, and operating procedures, the facility intelligent agent obtained by the technical solution disclosed herein better reflects the characteristics and principles of the technological process of specific related equipment, exhibiting higher intelligence and stronger analytical capabilities. It is particularly suitable for processes such as chemical processes with chain structures and known mechanistic characteristics. The technical solution disclosed herein also uses a gray box model in the process of using the intelligent agent for job monitoring and demand response. By using the associated gray box model and its generated simulation data in conjunction, intelligent agent prompts can be generated, achieving more user-relevant and intelligent job monitoring results. Attached Figure Description

[0016] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0017] Figure 1 This is an exemplary flowchart of a facility operation monitoring method based on a gray box model provided in accordance with embodiments of this disclosure; Figure 2 This is an exemplary flowchart of a training facility agent provided according to embodiments of the present disclosure; Figure 3 This is an exemplary flowchart provided in this disclosure for parsing received job monitoring requests to obtain the corresponding gray box model and associated monitoring data; Figure 4 This is an exemplary flowchart of generating operation monitoring results through a pre-trained facility agent, according to an embodiment of the present disclosure; Figure 5 This is a schematic diagram of an exemplary facility structure for a water electrolysis hydrogen production chemical process provided in accordance with embodiments of this disclosure; Figure 6 This is a schematic diagram of a facility operation monitoring system based on a gray box model provided according to an embodiment of the present disclosure; Figure 7 This is an exemplary structural diagram of a server provided according to an embodiment of the present disclosure. Detailed Implementation

[0018] To address the issues raised in the background art, such as the low level of intelligence of intelligent agents constructed solely based on general textual data of process flows, their inability to provide users with accurate decision-making information, and their difficulty in adapting to complex process execution scenarios like chemical processes, some embodiments of this disclosure provide a facility operation monitoring method and system based on a gray box model. This method can accurately describe the actual operation of the process flows executed by industrial facilities by constructing a gray box model, and utilize the gray box model for intelligent agent training and optimization of prompts during intelligent agent application, thereby significantly improving the intelligent analysis capabilities of the intelligent agent.

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the various embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details are presented in the various embodiments of this disclosure to facilitate a better understanding of the disclosure. However, the technical solutions claimed in this disclosure can be implemented even without these technical details and various variations and modifications based on the following embodiments. The division of the various embodiments below is for ease of description and should not constitute any limitation on the specific implementation of this disclosure. The various embodiments can be combined with and referenced by each other without contradiction.

[0020] In some embodiments of this disclosure, Figure 1 An exemplary flowchart of a facility operation monitoring method based on a gray box model is shown, such as... Figure 1 As shown, process 100 may specifically include: Step 110: Parse the received job monitoring request to obtain the corresponding gray box model and associated monitoring data. In some embodiments, the job monitoring request can be actively input by the user or automatically generated by the facility's operation monitoring system based on fluctuations in monitoring data; this is not limited here. In some embodiments, the gray box model corresponding to the job monitoring request refers to the gray box model associated with the monitoring object corresponding to the job monitoring request; correspondingly, the associated monitoring data refers to other monitoring data associated with the monitoring object that can be directly obtained; this is not limited here. The specific parsing process for the job monitoring request will be described in detail later and will not be repeated here.

[0021] Step 120: Obtain facility simulation data based on the gray-box model and associated monitoring data. The gray-box model is a modeling model between the black-box and white-box models, combining data-driven and theory-driven elements. It requires initial modeling using some known information about the facility and its corresponding process flow, while also relying on historical data to fine-tune the initial modeling results. This allows for a better construction of the mass-energy relationships and potential mathematical relationships between the various components of the facility. Using the gray-box model and associated monitoring data, the current operating status of the facility can be simulated to obtain some operational data that is difficult to obtain directly, while simultaneously achieving a comprehensive understanding and acquisition of the facility's current operating status. The specific construction and implementation of the gray-box model will be explained in detail later and will not be elaborated upon here.

[0022] Step 130: Based on facility simulation data, associated monitoring data, and operational monitoring requirements, generate operational monitoring results through a pre-trained facility agent. In some embodiments, the facility agent can be trained based on a gray-box model and historical facility data. The training and practical application of the facility agent will be described in detail later and will not be repeated here. It is understood that, based on the relevant steps of process 100 above, the gray-box model can be used for agent training and optimization of prompts during agent application. The generated operational monitoring results are more consistent with the actual application of the facility and can provide users with accurate intelligent analysis suggestions. The specific implementation of process 100 above will be further explained below with reference to specific embodiments.

[0023] In some embodiments, the facility agent is associated with the process flow executed by the facility; that is, the facility agent deployed locally on the facility can be adaptively trained and configured according to the specific process flow executed by the facility. Considering that the same process flow can be implemented by different devices, the facility agent needs to fully consider the constituent devices of the current facility and the connections between these devices during training. When the constituent devices of the current facility or the connections between them change, the facility agent needs to be adjusted promptly. In some embodiments, Figure 2 An exemplary flowchart of a training facility agent is shown, such as Figure 2 As shown, process 200 may specifically include...

[0024] Step 210: Obtain the associated equipment and associated monitoring items of the process flow. It is understood that a facility can consist of multiple component devices, and some of these component devices can implement an independent process flow. Therefore, the associated equipment of the process flow can include all component devices participating in that process flow, and the associated monitoring items can include the various monitoring data items corresponding to the aforementioned associated equipment; no limitations are imposed here.

[0025] Step 220: Construct an initial gray-box model based on preset core monitoring items, associated devices, and associated monitoring items. In some embodiments, the core monitoring items may be one or more specified associated monitoring items, serving as components of subsequent job monitoring requirements. In some embodiments, the constructed initial gray-box model is associated with the core monitoring items, so that when subsequent user job monitoring requirements include core monitoring items, the associated gray-box model can be invoked to assist the agent in generating job monitoring results.

[0026] Step 230: Based on historical monitoring data, perform parameter correction on the original gray box model to obtain the gray box model. It can be understood that for the completed original gray box model, the coefficients in the original gray box model can be correlated and corrected based on the facility's historical monitoring data, and the model after parameter correction can be used as the gray box model that truly reflects the facility's operational status.

[0027] In some embodiments, the gray box model may specifically include at least one or any combination of several of the following: an equipment relationship sub-model for describing the relationships between the constituent devices and / or device connection structures in the facility; a data monitoring sub-model for describing the relationships between the data acquisition devices and / or corresponding regulation and control devices in the facility; and a key indicator sub-model for describing the relationships between key parameters in the corresponding process flow of the facility. In some embodiments, taking a chemical process flow as an example, in a scenario where the facility is used to implement a chemical process flow, the equipment relationship sub-model can be constructed based on the type of constituent devices and / or device connection structures, using the theoretical reaction principles corresponding to the chemical process flow. For example, in a chemical process flow, the constituent devices of the facility can be interconnected through pipelines. In this case, the equipment relationship sub-model is constructed based on the type of each constituent device (e.g., separation equipment, heat exchanger, reactor, etc.) and the principles of mass balance and energy balance, without limitation. In some embodiments, the data monitoring sub-model can be constructed based on the distribution of each data acquisition device, using linear fitting and / or transfer function models. For example, each measuring instrument in a chemical process flow can use linear fitting and transfer functions (first-order, second-order, or higher) to construct the data monitoring sub-model based on its measuring point location. In some embodiments, the data monitoring sub-model can also be constructed based on the regulating and control equipment and the data acquisition equipment that has a control relationship with the regulating and control equipment, using proportional-integral-derivative (PID) algorithms and / or transfer function models. For example, for instruments with control relationships in chemical process flows (such as flow measuring instruments and regulating valves in flow control loops), the data monitoring sub-model can be constructed using PID algorithms and transfer function models, without limitation. In some embodiments, the quantitative key indicator sub-model can be multi-input. In the process of constructing the key indicator sub-model, all input data are first obtained, then the correlation between the input data and each key control indicator is analyzed, and finally the input conditions are selected. A linear relationship is constructed based on the input data and the least squares algorithm to form the key indicator sub-model, without limitation.

[0028] Step 240: Based on the gray box model, generate simulation data corresponding to the associated devices.

[0029] Step 250: Generate a training dataset based on historical monitoring data, simulation data, and the gray box model. Use the training dataset to train a pre-defined base large language model to obtain a facility agent. It can be understood that in steps 240 to 250 above, based on the gray box model, simulation data corresponding to each associated device in the current process flow can be generated. Multiple training samples are generated based on a comprehensive dataset including simulation data, historical monitoring data, and associated devices. Each training sample corresponds to an actual operating condition of the process flow, reflecting the operational relationships between the various associated devices. These training samples are unified into a training dataset, which is then used to train the pre-defined base large language model. This allows the trained facility agent to independently summarize the actual operating conditions of the process flow under different operating conditions from the training dataset, providing users with more targeted intelligent analysis services and decision-making suggestions. It is understandable that the facility intelligence agent provided in this disclosure uses a gray box model and simulation data and historical data optimized using the gray box model as training data for the agent. Compared with using a large amount of disordered process flow professional data, process flow descriptions and operating procedures, the above simulation data can better describe the relationship between various equipment, instruments and key monitoring indicators in the process flow, reduce the interference of invalid information on the training of the agent, and the trained facility intelligence agent can better reflect the process flow characteristics and principles of specific related equipment, and has higher intelligence.

[0030] In some embodiments, further, such as Figure 2 As shown, process 200 may further include the following step 260: storing the gray box model and its descriptive information, the descriptive information including associated monitoring items related to the gray box model and the index tags corresponding to the gray box model, wherein the index tags also correspond to the core monitoring items. It is understood that, while training the facility agent, in order to continue applying the constructed gray box model in subsequent use of the facility agent, it is necessary to store the gray box model; simultaneously, in order to select and call the corresponding gray box model according to the operation monitoring requirements, a corresponding index tag can be generated for each gray box model. This index tag also corresponds to the core monitoring items preset during the gray box model construction process, which is not limited here.

[0031] In some embodiments, the operation monitoring requirement may include user-inputted and / or automatically acquired warning information of core monitoring items. Specifically, in some practical application scenarios, when the monitoring data of a core monitoring item triggers a preset warning, the user can manually input the warning information to inquire about solutions from the facility's intelligent agent. Alternatively, the facility's automatic monitoring system can automatically generate an operation monitoring requirement to inquire about solutions from the facility's intelligent agent when it detects that the monitoring data of a core monitoring item has triggered a preset warning. In some embodiments, the operation monitoring requirement can be input in the form of prompt words based on a preset prompt template, or the user can freely input using natural language; no limitation is imposed here.

[0032] In some embodiments, Figure 3 An exemplary flowchart is shown, illustrating how to parse received job monitoring requests to obtain the corresponding gray-box model and associated monitoring data. Figure 3 As shown, process 300 may specifically include the following steps.

[0033] Step 310: Parse the warning information to obtain the index tags corresponding to the core monitoring items. In some embodiments, if the job monitoring requirement explicitly includes core monitoring items, the index tags corresponding to the core monitoring items can be directly obtained and subsequent gray-box model verification and matching can be performed. In some embodiments, the job monitoring requirement can also be a natural language request input by the user. It can be parsed by semantic analysis and understanding of the natural language request to obtain the core monitoring items contained in the natural language request, or the index tags that can be summarized from the natural language request can be directly obtained, without limitation. In some embodiments, information such as equipment, instruments, control relationships, and associated monitoring items contained in the job monitoring requirement can also be obtained by performing semantic analysis on the job monitoring requirement, and the index tags can be obtained by combining the above information, without limitation.

[0034] Step 320: Based on the index tags, obtain the matching gray box model and the associated monitoring data items related to the gray box model. In some embodiments, the stored gray box model description information can be retrieved and matched based on the index tags. Specifically, the index tag information corresponding to the gray box model can be stored based on an independent tag database, and the tag database can be referenced for verification and matching when a job monitoring request is received. Those skilled in the art can choose an appropriate verification and matching method according to actual needs, which is not limited here.

[0035] Step 330: Obtain related monitoring data based on related monitoring items.

[0036] In some embodiments, Figure 4 An exemplary flowchart is shown, illustrating the generation of operational monitoring results using a pre-trained facility agent. Figure 4As shown, process 400 may specifically include the following steps.

[0037] Step 410: Based on facility simulation data, associated monitoring data, and operational monitoring requirements, generate intelligent agent prompts using a preset prompt template. In some embodiments, the intelligent agent prompts may specifically include descriptions of core monitoring items corresponding to operational monitoring requirements, names of associated devices corresponding to core monitoring items, operational data of associated devices, simulation data of associated devices, directly associated input devices, indirectly associated input devices, directly associated output devices, indirectly associated data devices, key monitoring outputs, and the output format of operational monitoring results, which are not limited here. For specific details regarding the presentation of intelligent agent prompts, please refer to the relevant explanations in the later examples of the water electrolysis hydrogen production chemical process.

[0038] Step 420: Based on the gray box model prompts, generate operation monitoring results, which include one or more combinations of the following: the cause of the warning information, the actual facility operation status associated with the warning information, and the handling method of the warning information.

[0039] In some embodiments, facility intelligence agents can be applied to facilities related to chemical processes. It is understood that chemical processes are complex, highly coupled, and diverse generative processes. The facility operation monitoring method based on the gray box model provided in this disclosure is particularly suitable for processes like chemical processes, which have chain structures and known mechanistic characteristics. The technical solution provided in this disclosure will be further explained below using the chemical process of hydrogen production through water electrolysis as an example.

[0040] In some embodiments, Figure 5 A schematic diagram of an exemplary facility structure for a water electrolysis hydrogen production chemical process is shown, such as... Figure 5 As shown, in the electrolysis of water to produce hydrogen chemical process, a certain concentration of alkaline solution is metered by flow meter 501 and enters electrolytic cell 502, where an electrochemical reaction occurs. Hydrogen is produced from the hydrogen side and mixed with some alkaline solution, entering gas-liquid separator 504 through first pipe 503. Hydrogen is produced from the top of gas-liquid separator 504 through second pipe 505, and the outlet is metered by hydrogen flow meter 506. Alkaline solution is drawn from the bottom of gas-liquid separator 504 through third pipe 507 by alkaline solution circulation pump 508. The outlet of alkaline solution circulation pump 508 is equipped with throttle valve 509, which can be used to adjust the flow rate of alkaline solution circulation to control the liquid level in gas-liquid separator 504. The outlet of throttle valve 509 is equipped with alkaline solution flow meter 510 for monitoring alkaline solution circulation flow rate. Gas-liquid separator 504 also has a liquid level detector 511 for detecting the liquid level in gas-liquid separator 504.

[0041] In some embodiments, such as Figure 5The illustrated electrolytic water hydrogen production process includes an electrolyzer 502, multiple pipelines, a gas-liquid separator 504, and an alkali circulation pump 508. The pipelines directly related to the high-high level alarm of the gas-liquid separator 504 are the first pipeline 503, the second pipeline 505, and the third pipeline 507. The gas-liquid flow rate in the first pipeline 503 is related to the flow rate of the electrolyzer 502 and the alkali solution entering the electrolyzer 502 (the reading of the alkali solution via flow meter 501). The flow rate in the third pipeline 507 is related to the alkali circulation pump 508 and the alkali solution flow rate at the outlet of the alkali circulation pump 508 (the reading of the alkali solution flow meter 510). The liquid level height of the gas-liquid separator 504 is controlled by the throttle valve 509 at the outlet of the alkali circulation pump 508.

[0042] Based on the above analysis, for the core monitoring item of liquid level in gas-liquid separator 504 in the water electrolysis hydrogen production process, the following gray box model can be constructed: The first gray box model can be a model describing the relationship between the first pipeline 503, the electrolyzer 502, and the alkaline flow rate (the reading of the alkaline solution through the flow meter 501) in the electrolyzer 502. Since the material in the first pipeline 503 is a gas-liquid mixture, the constructed first gray box model can include gas phase flow constraint relationship and liquid phase flow constraint relationship. The gas phase flow constraint relationship is determined based on the current density of the electrolyzer 502 and the residence time of the alkaline solution in the electrolyzer 502. The liquid phase flow constraint relationship is determined by the current density of the electrolyzer 502, the residence time of the alkaline solution in the electrolyzer 502, and the reading of the alkaline solution through the flow meter 501. In some embodiments, the constructed second gray box model can be a model describing the relationship between the alkali flow rate at the outlet of the alkali circulation pump 508 and the flow rate in the third pipe 507. Specifically, it can include the relationship between the flow rate in the third pipe 507 and the alkali flow rate at the outlet of the alkali circulation pump 508, and the relationship coefficient can be obtained by fine-tuning based on historical monitoring data. In some embodiments, the key indicator sub-model of the constructed third gray box model can be a model describing the control relationship between the throttle valve 509 at the outlet of the alkali circulation pump 508 and the liquid level in the gas-liquid separator 504. In some embodiments, the constructed fourth gray box model can be a model describing the relationship between the liquid level in the gas-liquid separator 504 and the first pipe 503, the second pipe 505, and the third pipe 507. Specifically, it can include the constraint relationship between the liquid level in the gas-liquid separator 504 and the gas-liquid phase flow rates in the first pipe 503, the second pipe 505, and the third pipe 507, which is not limited here.

[0043] In some implementations, after the gray box model mentioned above has been constructed, historical data of the water electrolysis hydrogen production process can be obtained, and the coefficients in the gray box model can be correlated and corrected based on the historical data. Simulation data is generated based on the corrected model. Furthermore, a training dataset is generated from the historical data, simulation data, and each gray box model, and the training dataset is used to train a preset base large language model to obtain the facility intelligent agent of the water electrolysis hydrogen production process, and the corresponding index labels are generated and saved to the database.

[0044] In some embodiments, when an early warning message regarding the liquid level height of the gas-liquid separator appears, the warning message can be parsed, and the corresponding gray box model can be obtained based on the index label corresponding to the liquid level height of the gas-liquid separator to acquire associated monitoring data and simulation data, thereby realizing the construction of intelligent agent prompt words. In some embodiments, the generated intelligent agent prompt words can be as follows: Alarm message: <Gas-liquid separator high liquid level alarm>; Constraints: Location of occurrence: Equipment name: <Gas-liquid separator>; Operating data: <Gas-liquid separator liquid level reading>; Simulation data: <Gas-liquid separator liquid level simulation height>; Additional notes: <None>; Input association: Direct equipment: <First pipeline>; Indirect equipment: <Electrolytic cell>; Output association: Direct devices: <Third pipe>, <Second pipe>; Related data: <Electrolyzer parameters>; <Third pipe flow rate>; <Second pipe flow rate>; <First pipe flow rate>; Output format: Generate an analysis and diagnostic report in JSON format. The JSON data format is as follows: { "why": Analyze the reasons for the alarm. "what": Analyze the nature of the alarm. "how": Provides standard advice on how to handle the alarm. } Task Description: Diagnose alarm information and constraints based on alarm information, analyze alarm causes, provide standard recommendations for alarm handling, and generate output according to the specified format.

[0045] In some embodiments, the facility agent can generate corresponding operation monitoring results based on the agent prompts shown above. Specifically, these results may include the cause of excessively high liquid level in the gas-liquid separator, the current actual operating conditions of the facility, and possible ways to handle excessively high liquid level in the gas-liquid separator.

[0046] Some embodiments of this disclosure also relate to a facility operation monitoring system based on a gray box model, specifically, Figure 6 A schematic diagram of a facility operation monitoring system based on a gray box model is shown, as follows: Figure 6 As shown, it may specifically include a requirements analysis module 610, a facility simulation module 620, and a result generation module 630.

[0047] In some embodiments, the requirement parsing module 610 can be used to parse the received operation monitoring requirements to obtain the corresponding gray box model and associated monitoring data. In some embodiments, the facility simulation module 620 can be used to obtain facility simulation data based on the gray box model and associated monitoring data. In some embodiments, the result generation module 630 can be used to generate operation monitoring results through a pre-trained facility agent based on the facility simulation data, associated monitoring data, and operation monitoring requirements, wherein the facility agent can be trained based on the gray box model and historical facility data. In some embodiments, the specific functional implementation of the requirement parsing module 610 to the result generation module 630 can be implemented with reference to the various steps in the facility operation monitoring method provided in the foregoing embodiments, and will not be elaborated here.

[0048] Some embodiments of this disclosure also relate to a facility operation monitoring device based on a gray box model, specifically, Figure 7 An exemplary structural diagram of a facility operation monitoring device based on a gray box model is shown, such as... Figure 7 As shown, the facility operation monitoring device based on the gray box model includes at least one processor 710 and a memory 720 communicatively connected to the at least one processor. The memory 720 stores instructions that can be executed by the at least one processor 710. The instructions are executed by the at least one processor 710 to enable the at least one processor 710 to execute the facility operation monitoring method based on the gray box model provided in the foregoing embodiments.

[0049] The memory 720 and processor 710 are connected via a bus, which may include any number of interconnecting buses and bridges, connecting various circuits of one or more processors 710 and memory 720 together. The bus may also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. A bus interface provides an interface between the bus and the transceiver. The transceiver may be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by processor 710 is transmitted over a wireless medium via an antenna, which further receives data and transmits it to the processor.

[0050] In some embodiments, the processor 710 may be responsible for managing the bus and general processing, and may also provide various functions, including timing, peripheral interfaces, voltage regulation, power management and other control functions, while the memory 720 may be used to store data used by the processor when performing operations, without limitation.

[0051] Some embodiments of this disclosure also relate to a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the steps of the facility operation monitoring method based on the gray box model provided in the foregoing embodiments. In some embodiments, the computer-readable storage medium may include flash memory, hard disk, multimedia card, card-type memory (e.g., SD or D6 memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the computer-readable storage medium may be an internal storage unit of a computer device, such as the hard disk or memory of the computer device. In other embodiments, the computer-readable storage medium may also be an external storage device of a computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. Of course, the computer-readable storage medium may also include both internal storage units and external storage devices of a computer device. In this embodiment, the computer-readable storage medium is typically used to store the operating system and various application software installed on the computer device, such as the program code of the facility operation monitoring method in this embodiment. Furthermore, the computer-readable storage medium can also be used to temporarily store various types of data that have been output or will be output.

[0052] Some embodiments of this disclosure also relate to a computer program product, including a computer program that, when executed by a processor, implements the steps of the facility operation monitoring method based on the gray box model provided in the foregoing embodiments.

[0053] In some embodiments, the computer program product may involve only a computer program, which may be carried on a storage medium or processing device. In other embodiments, the computer program product may also be a storage medium or processing device containing the aforementioned computer program. The processing device may include one or more processors, and the storage medium. Those skilled in the art will understand that all or part of the steps in the facility operation monitoring method provided in the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this disclosure.

[0054] The basic concepts have been described above. It is obvious that the detailed disclosure above is merely illustrative and does not constitute a limitation of this specification. Although not explicitly stated herein, various modifications, improvements, and corrections may be made to this specification by those skilled in the art. Such modifications, improvements, and corrections are taught in this specification and therefore remain within the spirit and scope of the exemplary embodiments described herein.

Claims

1. A facility operation monitoring method based on a gray box model, characterized in that, The method includes the following steps: Parse the received job monitoring requests to obtain the corresponding gray box model and associated monitoring data; Based on the gray box model and the associated monitoring data, facility simulation data is obtained; Based on the facility simulation data, the associated monitoring data, and the operation monitoring requirements, operation monitoring results are generated through a pre-trained facility agent. The facility agent is trained based on the gray box model and historical facility data.

2. The facility operation monitoring method based on the gray box model according to claim 1, characterized in that, The facility agent is associated with the process flow executed by the facility, and the facility agent is trained based on the following steps; Obtain the associated equipment and associated monitoring items of the process flow; Based on the preset core monitoring items, associated devices and associated monitoring items, construct the original gray box model; Based on historical monitoring data, the parameters of the original gray box model are corrected to obtain the gray box model; Based on the gray box model, simulation data corresponding to the associated equipment is generated; A training dataset is generated based on the historical monitoring data, the simulation data, and the gray box model. The training dataset is then used to train a preset base large language model to obtain the facility intelligent agent.

3. The facility operation monitoring method based on the gray box model according to claim 2, characterized in that, The method further includes: The gray box model and its description information are stored. The description information includes associated monitoring items related to the gray box model and index tags corresponding to the gray box model. The index tags also correspond to the core monitoring items.

4. The facility operation monitoring method based on the gray box model according to claim 3, characterized in that, The job monitoring requirements include early warning information for the core monitoring items, which are input by the user and / or automatically acquired. The process of parsing the received job monitoring requirements to obtain the corresponding gray box model and associated monitoring data includes the following steps: The warning information is parsed to obtain the index tag corresponding to the core monitoring item; Based on the index tags, obtain the matching gray box model and the associated monitoring data items related to the gray box model; Based on the associated monitoring items, obtain the associated monitoring data.

5. The facility operation monitoring method based on a gray box model according to any one of claims 1 to 4, characterized in that, The gray box model includes at least one or any combination of multiple sub-models, such as an equipment relationship sub-model for describing the relationship between the constituent equipment and / or equipment connection structures in the facility, a data monitoring sub-model for describing the relationship between the data acquisition equipment and / or corresponding regulation and control equipment in the facility, and a key indicator sub-model for describing the relationship between key parameters in the corresponding process flow of the facility.

6. The facility operation monitoring method based on the gray box model according to claim 5, characterized in that, The facility is used to implement chemical process flows; The equipment relationship sub-model is constructed based on the type of constituent equipment and / or equipment connection structure, and is obtained through the theoretical reaction principle corresponding to the chemical process flow.

7. The facility operation monitoring method based on a gray box model according to claim 5 or 6, characterized in that, The data monitoring sub-model is constructed based on the distribution of each data acquisition device, through linear fitting and / or a transfer function model; and / or The data monitoring sub-model is constructed based on the regulation and control device and the data acquisition device that has a control relationship with the regulation and control device, and is obtained through proportional-integral-differential algorithm and / or transfer function model.

8. The facility operation monitoring method based on the gray box model according to claim 1, characterized in that, The process of generating operation monitoring results through pre-trained facility agents includes the following steps: Based on the facility simulation data, the associated monitoring data, and the operation monitoring requirements, intelligent agent prompts are generated using preset prompt templates. Based on the intelligent agent prompts, the operation monitoring results are generated. The operation monitoring results include one or more combinations of the cause of the early warning information, the actual facility operation status associated with the early warning information, and the handling method of the early warning information.

9. The facility operation monitoring method based on a gray box model according to any one of claims 1 to 8, characterized in that, The facility intelligence agent is applied to facilities related to chemical process flows.

10. A facility operation monitoring system based on a gray box model, characterized in that, The system includes: The requirement parsing module is used to parse the received job monitoring requirements in order to obtain the corresponding gray box model and associated monitoring data; The facility simulation module is used to obtain facility simulation data based on the gray box model and the associated monitoring data; The result generation module is used to generate operation monitoring results based on the facility simulation data, the associated monitoring data, and the operation monitoring requirements, through a pre-trained facility agent. The facility agent is trained based on the gray box model and historical facility data.

Images