A monitoring method and system for an immersion liquid cooling system

Abstract

The present application relates to a monitoring method and system for an immersion liquid cooling system, and the method comprises the following steps: obtaining basic information of the cooling system, and generating a monitoring layout plan through a preset plan planning model; installing a state parameter acquisition device in the cooling system corresponding to the monitoring layout plan to collect state parameter information of the cooling system, and setting a video acquisition device in the area where the cooling system is located according to the monitoring layout plan to collect image information of the cooling system; obtaining the state parameter information of the cooling system in real time, and performing data cleaning through a preset risk warning model according to the state parameter monitoring threshold information, and extracting high-risk state parameter information to generate state risk warning information; obtaining the image information of the cooling system in real time, and identifying the image information through a preset image recognition model to generate image risk warning information. The present application has the effect of improving the monitoring efficiency and monitoring accuracy of the cooling system.

Description

Technical Field

[0001] The present application relates to the field of liquid cooling systems, and in particular to a monitoring method and system for an immersion liquid cooling system. Background Art

[0002] Liquid cooling system is a cooling technology that uses liquid as a heat transfer medium, immerses the heat-generating device in the liquid, and exchanges heat through direct contact. This technology is mainly used to address energy consumption issues in big data and ultra-density computing, replacing traditional air cooling technology and providing heat dissipation solutions for computer equipment. The immersion liquid cooling system directly immerses the heat-generating components in the coolant, relying on the flow and circulation of the liquid to remove the heat. In order to ensure that the servers and other high-heat-generating devices served by the liquid cooling system can fully dissipate heat, it is often necessary to monitor the liquid cooling system so that faults can be repaired as soon as possible to ensure the stable operation of the liquid cooling system.

[0003] The existing monitoring method of the immersion liquid cooling system often uses the state parameter analysis method for monitoring. This monitoring method often requires a lot of manpower and material resources based on the scale of the cooling system. The monitoring plan is designed according to the actual distribution diagram of the liquid cooling system and the setting method of the state parameter sensor is constantly debugged. Some other liquid cooling systems also use traditional image monitoring and image analysis to monitor the immersion liquid cooling system, which can often only generate alarm information after the accident occurs, which is easy to cause irreparable losses. For example, after a fire occurs in the cabinet, an alarm is issued through image recognition, and the loss is irreparable at this time. For example, the Chinese patent with publication number CN116600085A discloses a server monitoring system, a server monitoring method, a device and a storage medium, which relates to the field of computer technology, including: a multi-node data transmission system composed of multiple camera devices, used to collect server cabinet images of an immersion liquid cooling data center; a Bluetooth routing device, used to receive server cabinet images transmitted by the multi-node data transmission system; a cloud server, used to receive server cabinet images uploaded by the Bluetooth routing device, and process the server cabinet images to obtain immersion liquid cooling server information in the server cabinet images. The application states that the image of the server cabinet of the immersion liquid cooling data center is collected by a camera device, and the image is sent to a Bluetooth routing device to be uploaded to a cloud server for processing to obtain the immersion liquid cooling server information in the server cabinet image, replacing the manual inspection method, making the inspection of the data center more automated and intelligent, and reducing human resource consumption. The application also has the above-mentioned problem of only using video surveillance to monitor the immersion liquid cooling system, which can often only generate alarm information after an accident occurs, which is easy to cause irreparable losses.

[0004] With respect to the above-mentioned related technologies, the inventors believe that the existing immersion liquid cooling system has a single monitoring method, and the planning and debugging steps are cumbersome. It can only generate alarm information after an accident occurs, which can easily cause irreparable losses. Summary of the invention

[0005] In order to solve the problems of low monitoring efficiency of existing immersion liquid cooling systems, inability to achieve risk warning, and easy to cause irreparable losses, the present application provides a monitoring method and system for an immersion liquid cooling system.

[0006] In a first aspect, the present application provides a monitoring method for an immersion liquid cooling system, which adopts the following technical solution:

[0007] A monitoring method for an immersion liquid cooling system comprises the following steps:

[0008] S1. Obtain basic information of the cooling system, and generate a monitoring layout plan through a preset plan planning model; the basic information of the cooling system includes cooling system layout information, cooling system component information, and cooling system operating parameter information; the plan planning model is a machine learning model obtained through historical data training, and the monitoring layout plan includes layout information of at least one state parameter acquisition device, state parameter monitoring threshold information, and layout information of at least one video acquisition device, and the state parameter monitoring threshold information includes a state parameter early warning threshold and a state parameter alarm threshold, and the state parameter early warning threshold is between the state parameter alarm threshold and the normal state parameter recorded in the cooling system operating parameter information;

[0009] S2. Installing state parameter acquisition equipment in the cooling system according to the monitoring layout plan to collect state parameter information of the cooling system, and setting video acquisition equipment in the area where the cooling system is located according to the monitoring layout plan to collect image information of the cooling system;

[0010] S3. Real-time acquisition of the state parameter information of the cooling system, based on the state parameter monitoring threshold information, data cleaning is performed through a preset risk warning model, and high-risk state parameter information is extracted to generate state risk warning information; the risk warning model is a machine learning model obtained by training with historical data;

[0011] S4. Acquire image information of the cooling system in real time, and generate image risk warning information by recognizing the image information through a preset image recognition model. The image recognition model is based on the ChatGPT 4.0 model and is obtained through risk data training.

[0012] Preferably, the step of setting up a video acquisition device in the area where the cooling system is located according to the monitoring layout plan specifically includes the following steps:

[0013] A1. Install the video acquisition device at the designated installation location in the area where the cooling system is located according to the video acquisition device layout information of the monitoring layout plan;

[0014] A2. Build a three-dimensional model of the cooling system and its area based on the basic information of the cooling system, cut and splice the image information of the cooling system collected by each video acquisition device, set the map on the three-dimensional model, and determine whether there is an unmapped area in the three-dimensional model;

[0015] A3. If yes, it is determined that there is a blind spot in monitoring, and adjustment information is generated and sent to the installer. After the installer completes the debugging, jump to step A2;

[0016] A4. If it does not exist, the video capture device is installed.

[0017] Preferably, the step of generating the adjustment information and sending it to the installer specifically comprises the following steps:

[0018] Determine the video acquisition devices corresponding to a plurality of image fragments adjacent to the unmapped area in the three-dimensional model as debuggable video acquisition devices;

[0019] Determine the area ratio of the image fragments captured by each adjustable video capture device and attached to the three-dimensional model compared to the effective image captured by the device, wherein the effective image is the image area where the cooling system exists in the image information captured by the video capture device;

[0020] Sort the adjustable video acquisition devices in ascending order based on the area ratio, and select the adjustable video acquisition device with the smallest area ratio as the recommended adjustment device;

[0021] The recommended adjustment equipment information and the sorting information are packaged to generate adjustment information and sent to the installer.

[0022] Preferably, the real-time acquisition of the state parameter information of the cooling system is based on the state parameter monitoring threshold information, and data cleaning is performed through a preset risk warning model to extract high-risk state parameter information to generate state risk warning information, which specifically includes the following steps:

[0023] Acquire the status parameter information of the cooling system in real time, and use the preset risk warning model to clean the data and remove the status parameter data that does not exceed the status parameter warning threshold to obtain the high-risk status parameter information;

[0024] The high-risk state parameter information is refined through the risk warning model, and the first dangerous state parameter set exceeding the state parameter alarm threshold and the second dangerous state parameter set within the state parameter monitoring threshold and the state parameter alarm threshold are screened out, and the state risk warning information is generated based on the high-risk state parameter information in the first dangerous state parameter set;

[0025] Acquire negative feedback adjustment component information of the cooling system, screen and determine high-risk state parameter information lacking negative feedback means in the second dangerous state parameter set, and provide state risk warning information based on the high-risk state parameter information lacking negative feedback means;

[0026] For the high-risk state parameter information with negative feedback means in the second dangerous state parameter set, the corresponding negative feedback means are activated, and an early warning monitoring curve is generated according to the activation time of the negative feedback means to dynamically generate an alarm coefficient, and various early warning scores of high-risk state parameter information with negative feedback means are calculated through a preset early warning monitoring calculation formula, and when the early warning score value is greater than the preset alarm score threshold, the state risk early warning information corresponding to the high-risk state parameter information is generated;

[0027] The early warning monitoring calculation formula is specifically:

[0028] Y=X*T; where X is the alarm coefficient of the high-risk state parameter information, and T is the activation time of the negative feedback means corresponding to the high-risk state parameter information.

[0029] Preferably, the method of dynamically generating an alarm coefficient by generating an early warning monitoring curve according to the activation time of the negative feedback means specifically includes: after the negative feedback means corresponding to each high-risk state parameter information is activated, generating an early warning monitoring curve based on the activation time of the negative feedback means; calculating the alarm priority of each high-risk state parameter information at each time node after the negative feedback is activated based on the early warning monitoring curve by a preset priority calculation formula, and obtaining the alarm coefficient corresponding to the alarm priority of the high-risk state parameter information at the current time node according to a preset priority coefficient comparison table;

[0030] The priority calculation formula is specifically:

[0031] ;

[0032] in, is the alarm priority of the high-risk state parameter information at the i-th unit time node in the early warning monitoring curve, is the alarm priority of the high-risk state parameter information at the i-1th unit time node in the early warning monitoring curve, where The alarm priority is preset by the management personnel; is the state parameter value of the high-risk state parameter information at the i-th unit time node in the early warning monitoring curve, is the state parameter value of the high-risk state parameter information at the i-1th unit time node in the early warning monitoring curve, and O is the expected unit time change value after the preset negative feedback adjustment of the high-risk state parameter information is started.

[0033] Preferably, the real-time acquisition of image information of the cooling system and the generation of image risk warning information by recognizing the image information through a preset image recognition model specifically include the following steps:

[0034] Acquire the image information of the cooling system in real time, identify the status light information, coolant level information and physical status of the cooling system in the image information through the preset image recognition model, and generate image risk warning information when an abnormality occurs;

[0035] Through the image recognition model based on the image recognition capability of the ChatGPT 4.0 model, the objects entering and leaving the cooling system area and the behaviors of people in the image information are identified. When there are pre-set high-risk objects and high-risk behaviors, image risk warning information is generated. When unauthorized objects and unauthorized persons are found, it is determined whether there are authorized persons within the preset effective range of the unauthorized objects and unauthorized persons.

[0036] If it does not exist, an image risk warning message is generated;

[0037] If it exists, an authorized voice message is generated to prompt the authorized person to authorize the unauthorized items and unauthorized persons. The time is simultaneously counted and an image risk warning message is generated for the unauthorized items and unauthorized persons when the time expires.

[0038] In a second aspect, the present application provides a monitoring system for an immersion liquid cooling system, which adopts the following technical solution:

[0039] A monitoring system for an immersion liquid cooling system, comprising:

[0040] A monitoring layout planning module is used to obtain basic information of the cooling system and generate a monitoring layout plan through a preset plan planning model; the basic information of the cooling system includes cooling system layout information, cooling system component information and cooling system operating parameter information; the plan planning model is a machine learning model obtained through historical data training, the monitoring layout plan includes layout information of at least one state parameter acquisition device, state parameter monitoring threshold information and at least one video acquisition device layout information, the state parameter monitoring threshold information includes a state parameter early warning threshold and a state parameter alarm threshold, and the state parameter early warning threshold is between the state parameter alarm threshold and the normal state parameter recorded in the cooling system operating parameter information;

[0041] The monitoring installation verification module is used to install the status parameter acquisition equipment in the cooling system according to the monitoring layout plan to collect the status parameter information of the cooling system, and to set the video acquisition equipment in the area where the cooling system is located according to the monitoring layout plan to collect the image information of the cooling system;

[0042] The state parameter warning module is used to obtain the state parameter information of the cooling system in real time, and to perform data cleaning based on the state parameter monitoring threshold information through a preset risk warning model, and to extract high-risk state parameter information to generate state risk warning information; the risk warning model is a machine learning model obtained by training historical data;

[0043] The intelligent video warning module is used to obtain image information of the cooling system in real time, and to generate image risk warning information by recognizing the image information through a preset image recognition model. The image recognition model is based on the ChatGPT 4.0 model and is obtained through risk data training.

[0044] Preferably, the state parameter early warning module specifically includes:

[0045] A data cleaning unit is used to obtain the state parameter information of the cooling system in real time, and to perform data cleaning through a preset risk warning model to remove the state parameter data that does not exceed the state parameter warning threshold to obtain high-risk state parameter information;

[0046] A data extraction unit is used to extract high-risk state parameter information through a risk warning model, screen out a first dangerous state parameter set that exceeds a state parameter alarm threshold and a second dangerous state parameter set that is within an interval between a state parameter monitoring threshold and a state parameter alarm threshold, and generate state risk warning information based on the high-risk state parameter information in the first dangerous state parameter set;

[0047] A negative feedback screening unit is used to obtain negative feedback adjustment component information of the cooling system, screen and determine high-risk state parameter information lacking negative feedback means in the second dangerous state parameter set, and provide state risk warning information based on the high-risk state parameter information lacking negative feedback means;

[0048] A dynamic monitoring unit is used to activate the corresponding negative feedback means for the high-risk state parameter information with the negative feedback means in the second dangerous state parameter set, generate an early warning monitoring curve according to the activation time of the negative feedback means, dynamically generate an alarm coefficient, calculate various early warning scores of the high-risk state parameter information with the negative feedback means through a preset early warning monitoring calculation formula, and generate state risk early warning information corresponding to the high-risk state parameter information when the early warning score value is greater than a preset alarm score threshold;

[0049] The early warning monitoring calculation formula is specifically:

[0050] Y=X*T; where X is the alarm coefficient of the high-risk state parameter information, and T is the activation time of the negative feedback means corresponding to the high-risk state parameter information.

[0051] Preferably, the intelligent video warning module includes:

[0052] The system monitoring unit is used to obtain the image information of the cooling system in real time, identify the status light information, coolant level information and physical status of the cooling system in the image information through a preset image recognition model, and generate image risk warning information when an abnormality occurs;

[0053] The risk warning unit is used to identify the objects and personnel behaviors entering and leaving the area where the cooling system is located in the image information through the image recognition capability of the image recognition model based on the ChatGPT 4.0 model, and generate image risk warning information when there are pre-set high-risk objects and high-risk behaviors; and when unauthorized objects and unauthorized persons are found, determine whether there are authorized persons within the preset effective range of the unauthorized objects and unauthorized persons; if not, generate image risk warning information; if so, generate authorization voice information to prompt the authorized persons to authorize the unauthorized objects and unauthorized persons, and simultaneously perform timing to generate image risk warning information for unauthorized objects and unauthorized persons when the time has expired.

[0054] Preferably, the monitoring installation verification module includes a video installation verification unit, which is used to set up a video acquisition device in the area where the cooling system is located according to the monitoring layout plan, and specifically includes the following steps:

[0055] A1. Install the video acquisition device at the designated installation location in the area where the cooling system is located according to the video acquisition device layout information of the monitoring layout plan;

[0056] A2. Build a three-dimensional model of the cooling system and its area based on the basic information of the cooling system, cut and splice the image information of the cooling system collected by each video acquisition device, set the map on the three-dimensional model, and determine whether there is an unmapped area in the three-dimensional model;

[0057] A3. If yes, it is determined that there is a blind spot in monitoring, and adjustment information is generated and sent to the installer. After the installer completes the debugging, jump to step A2;

[0058] A4. If it does not exist, the video capture device is installed.

[0059] In summary, the present application includes at least one of the following beneficial technical effects:

[0060] 1. Through the cooperation of the risk warning model and the image recognition model, based on the powerful image recognition capability of the ChatGPT 4.0 model, there is no traditional machine vision model that can only train samples for recognition. It is more intelligent and accurate, and can identify any objects and personnel behaviors entering and leaving the cooling system area, which helps to avoid interference and damage to the cooling system by foreign objects and personnel, and minimizes the impact of external factors on the cooling system, taking precautions before it happens. At the same time, on the basis of the risk warning model intelligently analyzing and refining the state parameter information of the cooling system to screen high-risk state parameter information for early warning, the image recognition model is used to monitor and identify the state light information, coolant level information and physical state of the cooling system, so that the two can cooperate and verify each other, which can further improve the monitoring and early warning accuracy and efficiency, and ensure the stable and efficient operation of the cooling system.

[0061] 2. Clean, refine and screen the state parameter information of the cooling system through the risk warning model, optimize and reduce the amount of data processing, improve the warning response efficiency, and combine the negative feedback adjustment means of the cooling system. While performing intelligent negative feedback adjustment on the high-risk state parameter information with negative feedback means in the second dangerous state parameter set, calculate its alarm risk, and timely issue risk warnings for high-risk state parameter information with poor or unclear negative feedback adjustment effects, realize intelligent monitoring and early warning, save manpower and material resources, and achieve the effect of accurate and efficient early warning of the cooling system;

[0062] 3. Through the setting of the scheme planning model, the monitoring layout scheme is generated according to the actual layout of the cooling system through intelligent planning, and the status parameter acquisition equipment and video acquisition equipment are reasonably set for the cooling system to ensure that the status parameters of the key parts of the cooling system can be accurately monitored, and the image information of the area where the cooling system is located can be fully collected. This simplifies the construction and debugging process of the cooling system monitoring scheme, realizes accurate and efficient all-round monitoring of the cooling system, and helps to timely and efficiently warn of cooling system failures. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 is a method flow chart of a monitoring method of an immersion liquid cooling system in an embodiment of the present application;

[0064] Figure 2 is a flow chart of a method for setting a video acquisition device in an embodiment of the present application;

[0065] Figure 3 is a flow chart of a method for generating adjustment information in an embodiment of the present application;

[0066] Figure 4 is a flow chart of a method for monitoring state parameter information of a cooling system in an embodiment of the present application;

[0067] Figure 5 This is a flow chart of a method for intelligently monitoring image information of a cooling system in an embodiment of the present application.

[0068] Figure 6 It is a system block diagram of a monitoring system for an immersion liquid cooling system in an embodiment of the present application.

[0069] Explanation of the accompanying drawings: 1. Monitoring layout planning module; 2. Monitoring installation verification module; 21. Video installation verification unit; 3. Status parameter warning module; 31. Data cleaning unit; 32. Data extraction unit; 33. Negative feedback screening unit; 34. Dynamic monitoring unit; 4. Intelligent video warning module; 41. System monitoring unit; 42. Risk warning unit. DETAILED DESCRIPTION

[0070] The following is combined with Figure 1-6 This application is described in further detail.

[0071] The present application embodiment discloses a monitoring method for an immersion liquid cooling system. Figure 1 , a monitoring method for an immersion liquid cooling system, comprising the following steps:

[0072] S1. Plan and generate a monitoring layout plan: obtain basic information of the cooling system, and plan and generate a monitoring layout plan through a preset plan planning model; the basic information of the cooling system includes cooling system layout information, cooling system component information, and cooling system operating parameter information; the plan planning model is a machine learning model obtained by training with historical data, and its specific training steps are prior art and are not repeated here; the monitoring layout plan includes layout information of at least one state parameter acquisition device, state parameter monitoring threshold information, and layout information of at least one video acquisition device, and the state parameter monitoring threshold information includes a state parameter early warning threshold and a state parameter alarm threshold, and the state parameter early warning threshold is between the state parameter alarm threshold and the normal state parameter recorded in the cooling system operating parameter information;

[0073] It should be noted that the state parameter acquisition equipment includes temperature sensors, liquid level sensors, flow sensors, energy consumption parameter acquisition equipment, coolant liquid circuit sealing parameter acquisition equipment and other equipment for collecting state parameters of the cooling system during operation. The state parameter information is the state parameter information that needs to be monitored in the cooling system operation parameter information, such as temperature parameters, coolant level parameters, coolant flow parameters, energy consumption parameters and sealing parameters, etc.;

[0074] S2. System installation and debugging: based on the monitoring layout plan, status parameter acquisition equipment is installed in the cooling system to collect status parameter information of the cooling system, and video acquisition equipment is set in the area where the cooling system is located according to the monitoring layout plan to collect image information of the cooling system;

[0075] S3, status parameter monitoring and early warning: real-time acquisition of the status parameter information of the cooling system, based on the status parameter monitoring threshold information, data cleaning is performed through a pre-set risk early warning model, and high-risk status parameter information is extracted to generate status risk early warning information; the risk early warning model is a machine learning model obtained through historical data training;

[0076] S4, image information recognition and warning: real-time acquisition of image information of the cooling system, recognition of image information through a pre-set image recognition model to generate image risk warning information, the image recognition model is based on the ChatGPT 4.0 model and is obtained through risk data training. Through the setting of the scheme planning model, the monitoring layout scheme is generated according to the actual layout of the cooling system, and the state parameter acquisition equipment and video acquisition equipment are reasonably set for the cooling system to ensure that the state parameters of the key parts of the cooling system can be accurately monitored, and the image information of the area where the cooling system is located can be fully collected, which simplifies the construction and debugging process of the cooling system monitoring scheme, realizes accurate and efficient full-scale monitoring of the cooling system, and helps to timely and efficiently warn of cooling system failures. Through the cooperation of the risk warning model and the image recognition model, based on the powerful image recognition capability of the ChatGPT 4.0 model, there is no traditional machine vision model that can only train samples for recognition. It is more intelligent and accurate, and can identify any objects and personnel behaviors entering and leaving the cooling system area, which helps to avoid interference and damage to the cooling system by foreign objects and personnel, and minimizes the impact of external factors on the cooling system, so as to prevent problems before they occur. At the same time, on the basis of the risk warning model's intelligent analysis and refinement of the cooling system's status parameter information to screen high-risk status parameter information for early warning, the image recognition model is used to monitor and identify the cooling system's status light information, coolant level information, and physical status, so that the two can cooperate and verify each other, which can further improve the monitoring and early warning accuracy and efficiency, and ensure the stable and efficient operation of the cooling system.

[0077] Reference Figure 2 The above-mentioned video acquisition equipment is set in the area where the cooling system is located according to the monitoring layout plan, which specifically includes the following steps:

[0078] A1. Video acquisition equipment installation: Install the video acquisition equipment at the designated installation location in the area where the cooling system is located according to the video acquisition equipment layout information of the monitoring layout plan;

[0079] A2. Construct a three-dimensional model and determine whether there is an unmapped area in the three-dimensional model: construct a three-dimensional model of the cooling system and its area based on the basic information of the cooling system, cut and splice the image information of the cooling system collected by each video acquisition device, and set the map on the three-dimensional model to determine whether there is an unmapped area in the three-dimensional model;

[0080] A3. If yes, it is determined that there is a blind spot in monitoring, and adjustment information is generated and sent to the installer. After the installer completes the debugging, jump to step A2;

[0081] A4. If it does not exist, the video acquisition device is installed. By constructing a three-dimensional model for video mapping, the cooling system image information collected by each video acquisition device is accurately verified to avoid the existence of monitoring blind spots, which helps to improve the efficiency and accuracy of video acquisition device installation and debugging, and improve the accuracy and efficiency of monitoring and early warning of the cooling system.

[0082] Reference Figure 3 The above-mentioned generation of adjustment information and sending it to the installer specifically includes the following steps:

[0083] B1. Determine the video acquisition devices corresponding to a plurality of image fragments adjacent to the unmapped area in the three-dimensional model as debuggable video acquisition devices;

[0084] B2. Determine the area ratio of the image fragments captured by each adjustable video capture device and attached to the three-dimensional model to the effective image captured by the device, wherein the effective image is the image area where the cooling system exists in the image information captured by the video capture device;

[0085] B3. Sort the adjustable video acquisition devices in ascending order based on the area ratio, and select the adjustable video acquisition device with the smallest area ratio as the recommended adjustment device;

[0086] B4. Pack the recommended adjustment device information and sorting information to generate adjustment information and send it to the installer. After the monitoring blind spot is found, based on the area ratio of the image fragments captured by the corresponding video acquisition device on the 3D model compared to the effective image it captured, the video acquisition device with the least debugging interference is selected as the recommended adjustment device, which helps to improve the debugging efficiency of the installer and save manpower and material resources.

[0087] Reference Figure 4 The above-mentioned real-time acquisition of the state parameter information of the cooling system is based on the state parameter monitoring threshold information, and data cleaning is performed through a preset risk warning model to extract high-risk state parameter information to generate state risk warning information. Specifically, the following steps are included:

[0088] C1. Data cleaning: Real-time acquisition of the state parameter information of the cooling system, data cleaning through the preset risk warning model, eliminating the state parameter data that does not exceed the state parameter warning threshold to obtain high-risk state parameter information;

[0089] C2. Data extraction: extract high-risk state parameter information through the risk warning model, screen out the first dangerous state parameter set that exceeds the state parameter alarm threshold and the second dangerous state parameter set that is within the range of the state parameter monitoring threshold and the state parameter alarm threshold, and generate state risk warning information based on the high-risk state parameter information in the first dangerous state parameter set;

[0090] C3. Data screening: obtaining the negative feedback adjustment component information of the cooling system, screening and determining the high-risk state parameter information lacking negative feedback means in the second dangerous state parameter set, and providing state risk warning information based on the high-risk state parameter information lacking negative feedback means;

[0091] C4. Dynamic data monitoring: For high-risk state parameter information with negative feedback means in the second dangerous state parameter set, the corresponding negative feedback means are activated, and an early warning monitoring curve is generated according to the activation time of the negative feedback means to dynamically generate an alarm coefficient. The early warning score values ​​of various high-risk state parameter information with negative feedback means are calculated through a preset early warning monitoring calculation formula, and when the early warning score value is greater than the preset alarm score threshold, the state risk early warning information corresponding to the high-risk state parameter information is generated;

[0092] The early warning monitoring calculation formula is specifically:

[0093] Y=X*T; where X is the alarm coefficient of high-risk state parameter information, and T is the activation time of the negative feedback means corresponding to the high-risk state parameter information. The state parameter information of the cooling system is cleaned, refined and screened through the risk warning model, the data processing volume is optimized and reduced, and the warning response efficiency is improved. At the same time, combined with the negative feedback adjustment means of the cooling system, the high-risk state parameter information with negative feedback means in the second dangerous state parameter set is intelligently negatively adjusted, and its alarm risk is calculated. Risk warning is timely issued for high-risk state parameter information with poor or unclear negative feedback adjustment effects, realizing intelligent monitoring and early warning, saving manpower and material resources, and achieving the effect of accurate and efficient early warning of the cooling system.

[0094] In addition, the above-mentioned dynamic generation of the alarm coefficient by generating the early warning monitoring curve according to the activation time of the negative feedback means specifically includes: after the negative feedback means corresponding to each high-risk state parameter information is activated, generating the early warning monitoring curve based on the activation time of the negative feedback means; calculating the alarm priority of each high-risk state parameter information at each time node after the negative feedback is activated based on the early warning monitoring curve by a preset priority calculation formula, and obtaining the alarm coefficient corresponding to the alarm priority of the high-risk state parameter information at the current time node according to a preset priority coefficient comparison table;

[0095] The priority calculation formula is specifically:

[0096] ;

[0097] in, is the alarm priority of the high-risk state parameter information at the i-th unit time node in the early warning monitoring curve, is the alarm priority of the high-risk state parameter information at the i-1th unit time node in the early warning monitoring curve, where The alarm priority is preset by the management personnel; is the state parameter value of the high-risk state parameter information at the i-th unit time node in the early warning monitoring curve, is the state parameter value of the high-risk state parameter information at the i-1th unit time node in the early warning monitoring curve, and O is the expected unit time change value after the preset high-risk state parameter information negative feedback adjustment is activated. The above-mentioned early warning monitoring curve is generated after the negative feedback means of the high-risk state parameter information is activated, and the alarm priority of each high-risk state parameter information is dynamically corrected according to the negative feedback performance, thereby achieving the effect of intelligent risk early warning.

[0098] Reference Figure 5 The real-time acquisition of the image information of the cooling system and the generation of the image risk warning information by recognizing the image information through a preset image recognition model specifically include the following steps:

[0099] D1. Equipment monitoring: Real-time acquisition of image information of the cooling system, identification of the status light information, coolant level information and physical status of the cooling system in the image information through a pre-set image recognition model, and generation of image risk warning information when an abnormality occurs;

[0100] D2. Intelligent image recognition to determine whether there are authorized persons within the preset effective range of unauthorized items and unauthorized persons: Through the image recognition model based on the image recognition capability of the ChatGPT 4.0 model, the objects entering and leaving the area where the cooling system is located in the image information and the behaviors of people are identified. When there are pre-set high-risk items and high-risk behaviors, image risk warning information is generated, and when unauthorized items and unauthorized persons are found, it is determined whether there are authorized persons within the preset effective range of the unauthorized items and unauthorized persons;

[0101] D3. If it does not exist, generate image risk warning information;

[0102] D4. If it exists, an authorized voice message is generated to prompt the authorized person to authorize the unauthorized items and unauthorized persons, and the timing is performed simultaneously to generate image risk warning information for the unauthorized items and unauthorized persons that have exceeded the time limit. Based on the powerful image recognition capability of the ChatGPT 4.0 model, there is no traditional machine vision model that can only train samples for recognition, which is more intelligent and accurate. It can identify any objects and human behaviors entering and leaving the cooling system area, which helps to avoid interference and damage to the cooling system caused by foreign objects and personnel, and minimize the impact of external factors on the cooling system, so as to effectively improve the monitoring accuracy and efficiency of the cooling system.

[0103] The present application also discloses a monitoring system for an immersion liquid cooling system. Figure 6, a monitoring system for an immersion liquid cooling system, comprising a monitoring layout planning module 1, a monitoring layout planning module 2, a state parameter early warning module 3 and an intelligent video early warning module 4, wherein the intelligent video early warning module 4 is connected to a ChatGPT 4.0 model. The monitoring layout planning module 1 is used to obtain basic information of the cooling system, and generate a monitoring layout plan through a preset plan planning model. The basic information of the cooling system includes cooling system layout information, cooling system component information and cooling system operating parameter information; the plan planning model is a machine learning model obtained by training historical data, and the monitoring layout plan includes layout information of at least one state parameter acquisition device, state parameter monitoring threshold information and at least one video acquisition device layout information. The state parameter monitoring threshold information includes a state parameter early warning threshold and a state parameter alarm threshold, and the state parameter early warning threshold is between the state parameter alarm threshold and the normal state parameter recorded in the cooling system operating parameter information. The monitoring installation verification module 2 is used to install a state parameter acquisition device in the cooling system corresponding to the monitoring layout plan to collect the state parameter information of the cooling system, and to set a video acquisition device in the area where the cooling system is located according to the monitoring layout plan to collect image information of the cooling system. The state parameter warning module 3 is used to obtain the state parameter information of the cooling system in real time. According to the state parameter monitoring threshold information, data cleaning is performed through a preset risk warning model, and high-risk state parameter information is extracted to generate state risk warning information. The risk warning model is a machine learning model obtained through historical data training. The intelligent video warning module 4 is used to obtain the image information of the cooling system in real time, and the image risk warning information is generated by recognizing the image information through a preset image recognition model. The image recognition model is based on the ChatGPT 4.0 model and is obtained through risk data training. Through the setting of the scheme planning model, the monitoring layout scheme is generated according to the actual layout of the cooling system, and the state parameter acquisition equipment and video acquisition equipment are reasonably set for the cooling system to ensure that the state parameters of the key parts of the cooling system can be accurately monitored, and the image information of the area where the cooling system is located can be collected in all directions, which simplifies the construction and debugging process of the cooling system monitoring scheme, realizes accurate and efficient all-round monitoring of the cooling system, and helps to timely and efficiently warn of cooling system failures.Through the cooperation of the risk warning model and the image recognition model, based on the powerful image recognition capability of the ChatGPT 4.0 model, there is no traditional machine vision model that can only train samples for recognition. It is more intelligent and accurate, and can identify any objects and personnel behaviors entering and leaving the cooling system area, which helps to avoid interference and damage to the cooling system by foreign objects and personnel, and minimizes the impact of external factors on the cooling system, so as to prevent problems before they occur. At the same time, on the basis of the risk warning model's intelligent analysis and refinement of the cooling system's status parameter information to screen high-risk status parameter information for early warning, the image recognition model is used to monitor and identify the cooling system's status light information, coolant level information, and physical status, so that the two can cooperate and verify each other, which can further improve the monitoring and early warning accuracy and efficiency, and ensure the stable and efficient operation of the cooling system.

[0104] Reference Figure 6 , the state parameter early warning module 3 specifically includes:

[0105] The data cleaning unit 31 is used to obtain the state parameter information of the cooling system in real time, and to perform data cleaning through a preset risk warning model to remove the state parameter data that does not exceed the state parameter warning threshold to obtain high-risk state parameter information;

[0106] The data extraction unit 32 is used to extract the high-risk state parameter information through the risk warning model, screen out the first dangerous state parameter set that exceeds the state parameter alarm threshold and the second dangerous state parameter set that is within the interval between the state parameter monitoring threshold and the state parameter alarm threshold, and generate state risk warning information based on the high-risk state parameter information in the first dangerous state parameter set;

[0107] A negative feedback screening unit 33 is used to obtain negative feedback adjustment component information of the cooling system, screen and determine high-risk state parameter information lacking negative feedback means in the second dangerous state parameter set, and provide state risk warning information based on the high-risk state parameter information lacking negative feedback means;

[0108] The dynamic monitoring unit 34 is used to activate the corresponding negative feedback means for the high-risk state parameter information with the negative feedback means in the second dangerous state parameter set, generate an early warning monitoring curve according to the activation time of the negative feedback means, dynamically generate an alarm coefficient, calculate the early warning score values ​​of various high-risk state parameter information with negative feedback means through a preset early warning monitoring calculation formula, and generate state risk early warning information corresponding to the high-risk state parameter information when the early warning score value is greater than a preset alarm score threshold;

[0109] The early warning monitoring calculation formula is specifically:

[0110] Y=X*T; where X is the alarm coefficient of high-risk state parameter information, and T is the activation time of the negative feedback means corresponding to the high-risk state parameter information. The state parameter information of the cooling system is cleaned, refined and screened through the risk warning model, the data processing volume is optimized and reduced, and the warning response efficiency is improved. At the same time, combined with the negative feedback adjustment means of the cooling system, the high-risk state parameter information with negative feedback means in the second dangerous state parameter set is intelligently negatively adjusted, and its alarm risk is calculated. Risk warning is timely issued for high-risk state parameter information with poor or unclear negative feedback adjustment effects, realizing intelligent monitoring and early warning, saving manpower and material resources, and achieving the effect of accurate and efficient early warning of the cooling system.

[0111] Reference Figure 6 , the intelligent video warning module 4 includes a system monitoring unit 41 and a risk warning unit 42. The system monitoring unit 41 is used to obtain the image information of the cooling system in real time, identify the status light information, coolant level information and physical state of the cooling system in the image information through a preset image recognition model, and generate image risk warning information when an abnormality occurs. The risk warning unit 42 is used to identify the objects and personnel behaviors in and out of the area where the cooling system is located in the image information through the image recognition model based on the image recognition capability of the ChatGPT 4.0 model, and generate image risk warning information when there are pre-set high-risk objects and high-risk behaviors. At the same time, when unauthorized objects and unauthorized persons are found, it is determined whether there are authorized persons within the preset effective range of the unauthorized objects and unauthorized persons. If not, an image risk warning information is generated. If so, an authorized voice message is generated to prompt the authorized person to authorize the unauthorized objects and unauthorized persons, and timing is performed simultaneously to generate image risk warning information for unauthorized objects and unauthorized persons that exceed the preset authorization time threshold. Based on the powerful image recognition capability of the ChatGPT 4.0 model, it does not have the defect of traditional machine vision models that can only identify training samples. It is more intelligent and accurate, and can identify any objects and human behaviors entering and leaving the cooling system area, which helps to avoid interference and damage to the cooling system caused by foreign objects and personnel, and minimizes the possible impact of external factors on the cooling system, thereby effectively improving the monitoring accuracy and efficiency of the cooling system.

[0112] Reference Figure 6 The monitoring installation verification module 2 includes a video installation verification unit 21, which is used to set up a video acquisition device in the area where the cooling system is located according to the monitoring layout plan, and specifically includes the following steps:

[0113] A1. Install the video acquisition device at the designated installation location in the area where the cooling system is located according to the video acquisition device layout information of the monitoring layout plan;

[0114] A2. Build a three-dimensional model of the cooling system and its area based on the basic information of the cooling system, cut and splice the image information of the cooling system collected by each video acquisition device, set the map on the three-dimensional model, and determine whether there is an unmapped area in the three-dimensional model;

[0115] A3. If yes, it is determined that there is a blind spot in monitoring, and adjustment information is generated and sent to the installer. After the installer completes the debugging, jump to step A2;

[0116] A4. If it does not exist, the video acquisition device is installed. Through the setting of the video installation verification unit 21, by constructing a three-dimensional model for video mapping, the cooling system image information collected by each video acquisition device is accurately verified to avoid the existence of monitoring blind spots, which helps to improve the efficiency and accuracy of video acquisition device installation and debugging, and improve the monitoring and early warning accuracy and efficiency of the cooling system.

[0117] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the scope of protection of the invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on these embodiments, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in this field can still combine, add, delete or make other adjustments to the features in the various embodiments of the present invention according to the circumstances without conflict, without making creative work, so as to obtain different other technical solutions that do not deviate from the concept of the present invention in essence, and these technical solutions also belong to the scope of protection of the present invention.

Claims

1. A monitoring method for an immersion liquid cooling system, characterized in that: The following steps are involved: S1. Obtain basic information of the cooling system, and generate a monitoring layout plan through a preset plan planning model; the basic information of the cooling system includes cooling system layout information, cooling system component information, and cooling system operating parameter information; the plan planning model is a machine learning model obtained through historical data training, and the monitoring layout plan includes layout information of at least one state parameter acquisition device, state parameter monitoring threshold information, and layout information of at least one video acquisition device, and the state parameter monitoring threshold information includes a state parameter early warning threshold and a state parameter alarm threshold, and the state parameter early warning threshold is between the state parameter alarm threshold and the normal state parameter recorded in the cooling system operating parameter information; S2. Installing state parameter acquisition equipment in the cooling system according to the monitoring layout plan to collect state parameter information of the cooling system, and setting video acquisition equipment in the area where the cooling system is located according to the monitoring layout plan to collect image information of the cooling system; S3. Real-time acquisition of the state parameter information of the cooling system, based on the state parameter monitoring threshold information, data cleaning is performed through a pre-set risk warning model, and high-risk state parameter information is extracted to generate state risk warning information; The risk warning model is a machine learning model obtained by training historical data; S4. Acquire image information of the cooling system in real time, and generate image risk warning information by recognizing the image information through a preset image recognition model, wherein the image recognition model is based on the ChatGPT 4.0 model and is obtained through risk data training; The real-time acquisition of the state parameter information of the cooling system is based on the state parameter monitoring threshold information, and data cleaning is performed through a preset risk warning model to extract high-risk state parameter information to generate state risk warning information, which specifically includes the following steps: Acquire the status parameter information of the cooling system in real time, and use the preset risk warning model to clean the data and remove the status parameter data that does not exceed the status parameter warning threshold to obtain the high-risk status parameter information; The high-risk state parameter information is refined through the risk warning model, and the first dangerous state parameter set exceeding the state parameter alarm threshold and the second dangerous state parameter set within the state parameter warning threshold and the state parameter alarm threshold are screened out, and the state risk warning information is generated based on the high-risk state parameter information in the first dangerous state parameter set; Acquire negative feedback adjustment component information of the cooling system, screen and determine high-risk state parameter information lacking negative feedback means in the second dangerous state parameter set, and generate state risk warning information based on the high-risk state parameter information lacking negative feedback means; For the high-risk state parameter information with negative feedback means in the second dangerous state parameter set, the corresponding negative feedback means are activated, and an early warning monitoring curve is generated according to the activation time of the negative feedback means to dynamically generate an alarm coefficient, and various early warning scores of high-risk state parameter information with negative feedback means are calculated through a preset early warning monitoring calculation formula, and when the early warning score value is greater than the preset alarm score threshold, the state risk early warning information corresponding to the high-risk state parameter information is generated; The early warning monitoring calculation formula is specifically: Y=X*T; wherein X is the alarm coefficient of the high-risk state parameter information, and T is the activation time of the negative feedback means corresponding to the high-risk state parameter information.

2. The monitoring method of an immersion liquid cooling system according to claim 1, characterized in that: The step of setting up a video acquisition device in the area where the cooling system is located according to the monitoring layout plan specifically includes the following steps: A1. Install the video acquisition device at the designated installation location in the area where the cooling system is located according to the video acquisition device layout information of the monitoring layout plan; A2. Build a three-dimensional model of the cooling system and its area based on the basic information of the cooling system, cut and splice the image information of the cooling system collected by each video acquisition device, set the map on the three-dimensional model, and determine whether there is an unmapped area in the three-dimensional model; A3. If yes, it is determined that there is a blind spot in monitoring, and adjustment information is generated and sent to the installer. After the installer completes the debugging, jump to step A2; A4. If it does not exist, the video capture device is installed.

3. The monitoring method of an immersion liquid cooling system according to claim 2, characterized in that: The generating and sending the adjustment information to the installer specifically comprises the following steps: Determine the video acquisition devices corresponding to a plurality of image fragments adjacent to the unmapped area in the three-dimensional model as debuggable video acquisition devices; Determine the area ratio of the image fragments captured by each adjustable video capture device and attached to the three-dimensional model compared to the effective image captured by the device, wherein the effective image is the image area where the cooling system exists in the image information captured by the video capture device; Sort the adjustable video acquisition devices in ascending order based on the area ratio, and select the adjustable video acquisition device with the smallest area ratio as the recommended adjustment device; The recommended adjustment equipment information and the sorting information are packaged to generate adjustment information and sent to the installer.

4. The monitoring method of an immersion liquid cooling system according to claim 1, characterized in that: The method of dynamically generating an alarm coefficient by generating an early warning monitoring curve according to the activation time of the negative feedback means specifically includes: after the negative feedback means corresponding to each high-risk state parameter information is activated, generating an early warning monitoring curve based on the activation time of the negative feedback means; calculating the alarm priority of each high-risk state parameter information at each time node after the negative feedback is activated based on the early warning monitoring curve by a preset priority calculation formula, and obtaining the alarm coefficient corresponding to the alarm priority of the high-risk state parameter information at the current time node according to a preset priority coefficient comparison table; The priority calculation formula is specifically: Among them, M i is the alarm priority of the high-risk state parameter information at the i-th unit time node in the early warning monitoring curve, M i-1 is the alarm priority of the high-risk state parameter information at the i-1th unit time node in the early warning monitoring curve, where the alarm priority of M0 is preset by the management personnel; N i is the state parameter value of the high-risk state parameter information at the i-th unit time node in the early warning monitoring curve, N i-1 is the state parameter value of the high-risk state parameter information at the i-1th unit time node in the early warning monitoring curve, and O is the expected unit time change value after the preset negative feedback adjustment of the high-risk state parameter information is started.

5. The monitoring method of an immersion liquid cooling system according to claim 1, characterized in that: The real-time acquisition of image information of the cooling system and the generation of image risk warning information by recognizing the image information through a preset image recognition model specifically include the following steps: Acquire the image information of the cooling system in real time, identify the status light information, coolant level information and physical status of the cooling system in the image information through the preset image recognition model, and generate image risk warning information when an abnormality occurs; Through the image recognition model based on the image recognition capability of the ChatGPT 4.0 model, the objects entering and leaving the cooling system area and the behaviors of people in the image information are identified. When there are pre-set high-risk objects and high-risk behaviors, image risk warning information is generated. When unauthorized objects and unauthorized persons are found, it is determined whether there are authorized persons within the preset effective range of the unauthorized objects and unauthorized persons. If it does not exist, an image risk warning message is generated; If it exists, an authorized voice message is generated to prompt the authorized person to authorize the unauthorized items and unauthorized persons. The time is simultaneously counted and an image risk warning message is generated for the unauthorized items and unauthorized persons when the time expires.

6. A monitoring system for an immersion liquid cooling system, applied to a monitoring method for an immersion liquid cooling system according to any one of claims 1 to 5, characterized in that: include: A monitoring layout planning module (1) is used to obtain basic information of a cooling system and generate a monitoring layout plan through a preset plan planning model; the basic information of the cooling system includes cooling system layout information, cooling system component information and cooling system operating parameter information; the plan planning model is a machine learning model obtained through historical data training, the monitoring layout plan includes at least one layout information of a state parameter acquisition device, state parameter monitoring threshold information and at least one layout information of a video acquisition device, the state parameter monitoring threshold information includes a state parameter early warning threshold and a state parameter alarm threshold, and the state parameter early warning threshold is between the state parameter alarm threshold and the normal state parameter recorded in the cooling system operating parameter information; A monitoring installation verification module (2) is used to install a state parameter acquisition device in the cooling system according to the monitoring layout plan to acquire state parameter information of the cooling system, and to set a video acquisition device in the area where the cooling system is located according to the monitoring layout plan to acquire image information of the cooling system; A state parameter warning module (3) is used to obtain state parameter information of the cooling system in real time, perform data cleaning based on state parameter monitoring threshold information, and extract high-risk state parameter information to generate state risk warning information through a preset risk warning model; The risk warning model is a machine learning model obtained by training historical data; An intelligent video warning module (4) is used to obtain image information of the cooling system in real time, and to generate image risk warning information by recognizing the image information through a preset image recognition model, wherein the image recognition model is based on a ChatGPT 4.0 model and is obtained through risk data training; The state parameter early warning module (3) specifically comprises: A data cleaning unit (31) is used to obtain state parameter information of the cooling system in real time, and to perform data cleaning through a preset risk warning model to remove state parameter data that does not exceed a state parameter warning threshold value to obtain high-risk state parameter information; A data extraction unit (32) is used to extract high-risk state parameter information through a risk warning model, screen out a first dangerous state parameter set that exceeds a state parameter alarm threshold and a second dangerous state parameter set that is within an interval between a state parameter alarm threshold and a state parameter alarm threshold, and generate state risk warning information based on the high-risk state parameter information in the first dangerous state parameter set; A negative feedback screening unit (33) is used to obtain negative feedback adjustment component information of the cooling system, screen and determine high-risk state parameter information lacking negative feedback means in the second dangerous state parameter set, and generate state risk warning information based on the high-risk state parameter information lacking negative feedback means; A dynamic monitoring unit (34) is used to activate the corresponding negative feedback means for the high-risk state parameter information with the negative feedback means in the second dangerous state parameter set, generate an early warning monitoring curve according to the activation time of the negative feedback means, dynamically generate an alarm coefficient, calculate the early warning score values ​​of various high-risk state parameter information with negative feedback means through a preset early warning monitoring calculation formula, and generate state risk early warning information corresponding to the high-risk state parameter information when the early warning score value is greater than a preset alarm score threshold; The early warning monitoring calculation formula is specifically: Y=X*T; wherein X is the alarm coefficient of the high-risk state parameter information, and T is the activation time of the negative feedback means corresponding to the high-risk state parameter information.

7. The monitoring system of an immersion liquid cooling system according to claim 6, characterized in that: The intelligent video warning module (4) comprises: A system monitoring unit (41) is used to obtain image information of the cooling system in real time, identify status light information, coolant level information and physical status of the cooling system in the image information through a preset image recognition model, and generate image risk warning information when an abnormality occurs; The risk warning unit (42) is used to identify the objects and personnel behaviors entering and leaving the cooling system area in the image information through the image recognition model based on the image recognition capability of the ChatGPT 4.0 model, and generate image risk warning information when there are pre-set high-risk objects and high-risk behaviors; and when unauthorized objects and unauthorized persons are found, determine whether there are authorized persons within the preset effective range of the unauthorized objects and unauthorized persons; if not, generate image risk warning information; if so, generate authorization voice information to prompt the authorized person to authorize the unauthorized objects and unauthorized persons, and simultaneously perform timing to generate image risk warning information for the unauthorized objects and unauthorized persons when the time limit is exceeded.

8. The monitoring system of an immersion liquid cooling system according to claim 6, characterized in that: The monitoring installation verification module (2) comprises a video installation verification unit (21), which is used to set up a video acquisition device in the area where the cooling system is located according to the monitoring layout plan, and specifically comprises the following steps: A1. Install the video acquisition device at the designated installation location in the area where the cooling system is located according to the video acquisition device layout information of the monitoring layout plan; A2. Build a three-dimensional model of the cooling system and its area based on the basic information of the cooling system, cut and splice the image information of the cooling system collected by each video acquisition device, set the map on the three-dimensional model, and determine whether there is an unmapped area in the three-dimensional model; A3. If yes, it is determined that there is a blind spot in monitoring, and adjustment information is generated and sent to the installer. After the installer completes the debugging, jump to step A2; A4. If it does not exist, the video capture device is installed.

Images