An operation and maintenance management method and system based on a heterogeneous multi-chip cloud architecture

Abstract

The application relates to a kind of operation and maintenance management method and system based on heterogeneous multi-chip cloud architecture, and relates to the field of operation and maintenance management, which comprises: collecting chip multidimensional data;According to the chip multidimensional data to know chip load condition;Based on the resource load condition to generate resource call demand;Collecting chip performance bottleneck;Combining the resource call demand and the chip performance bottleneck to generate dynamic adjustment parameter;Based on the dynamic adjustment parameter to carry out the resource mobilization of multi-chip, simultaneously generate resource adjustment log, and input the resource adjustment log into the preset operation and maintenance vector database.The application has the effect of improving resource utilization.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of operation and maintenance, and in particular to an operation and maintenance method and system based on a heterogeneous multi-chip cloud architecture. BACKGROUND

[0002] The operation and maintenance based on a heterogeneous multi-chip cloud architecture refers to a technical system for unified monitoring, scheduling, configuration and fault handling of computing resources, storage resources, network resources and the like under a cloud infrastructure architecture composed of different types and different functions of chips (such as CPU, GPU, FPGA, ASIC, etc.).

[0003] At present, with the rapid development of domestic chips, cloud computing platforms gradually adopt a heterogeneous multi-chip cloud architecture to balance flexibility and security. However, in the signal creation environment, the wide application of domestic chips (such as Haiguang, Kunpeng, Feiteng, Shenwei) and operating systems (such as Kirin, UOS, and Euler) introduces the challenge of heterogeneity.

[0004] However, the existing operation and maintenance is mostly designed for single-chip homogeneous environment, and the indicators and methods for coping with multi-chip cloud scenarios are not completely applicable, resulting in scattered heterogeneous resources and reduced resource utilization, which needs to be improved. SUMMARY

[0005] In order to improve the resource utilization, the present application provides an operation and maintenance method and system based on a heterogeneous multi-chip cloud architecture.

[0006] In the first aspect, the present application provides an operation and maintenance method based on a heterogeneous multi-chip cloud architecture, which adopts the following technical solution:

[0007] An operation and maintenance method based on a heterogeneous multi-chip cloud architecture, comprising:

[0008] Collecting multi-dimensional chip data;

[0009] According to the chip multi-dimensional data to know the chip load condition;

[0010] Based on the resource load condition to generate resource call demand;

[0011] Collecting chip performance bottlenecks;

[0012] Combining the resource call demand and the chip performance bottlenecks to generate dynamic adjustment parameters;

[0013] Based on the dynamic adjustment parameters to mobilize the resources of multiple chips, generate resource adjustment logs, and input the resource adjustment logs into a preset operation and maintenance vector database.

[0014] By adopting the technical scheme, the system realizes real-time monitoring of the load state through acquisition of multi-dimensional data of the chip, generates accurate resource calling demand in combination with performance bottleneck analysis, and then realizes intelligent scheduling and collaborative work of cross-chip resources through dynamic adjustment of parameters. Meanwhile, the log of each resource adjustment is stored in an operation and maintenance vector database, and vector retrieval technology is used to realize quick matching and reuse of historical scheduling strategies, form knowledge accumulation and evolution ability, so as to realize real-time sensing of chip state difference and dynamic optimization of resource allocation, improve the collaborative efficiency of the heterogeneous chip cluster, and then improve the resource utilization rate.

[0015] Optionally, the method further comprises:

[0016] acquiring a current chip usage signal;

[0017] comparing whether the current chip usage signal is consistent with a preset reference chip usage signal;

[0018] if consistent, continuing to acquire the current chip usage signal;

[0019] if inconsistent, comparing the current chip usage signal with the reference chip usage signal to frame out an abnormal signal frequency band;

[0020] based on the abnormal signal frequency band, matching out an abnormal chip, and processing the abnormal chip by using a preset abnormal chip processing method.

[0021] Optionally, the abnormal chip processing method comprises:

[0022] acquiring solder joint image information of the abnormal chip;

[0023] identifying and positioning a preset solder joint feature from the solder joint image information to obtain a solder joint position;

[0024] controlling a preset power-on test device to perform power-on test on the solder joint position, and acquiring a solder joint temperature value;

[0025] when the solder joint temperature value exceeds a preset standard solder joint temperature value, outputting an abnormal solder joint position, and controlling a preset repair welding device to perform repair welding on the abnormal solder joint position;

[0026] when the solder joint temperature value does not exceed the preset standard solder joint temperature value, reporting a chip abnormality prompt.

[0027] Optionally, the method further comprises:

[0028] acquiring a chip temperature value and an adjusted chip load condition;

[0029] based on the adjusted chip load condition, generating a chip reference temperature interval;

[0030] combine the chip temperature value and a preset thermal conductivity of the thermal grease to generate a reference temperature value, when and only when the chip temperature value is greater than a maximum value of the chip reference temperature interval;

[0031] collect area temperature values of all preset cold head sensing areas;

[0032] arrange the area temperature values of all the cold head sensing areas from large to small, and define the cold head sensing area with an area temperature value lower than the reference temperature value as a thermal grease missing area, and process the thermal grease missing area by using a preset thermal grease missing processing method.

[0033] Optionally, the thermal grease missing processing method comprises:

[0034] collect the number of missing areas of the thermal grease missing area;

[0035] determine whether the number of missing areas is greater than a preset standard number of missing areas;

[0036] when the number of missing areas is not greater than the standard number of missing areas, report a thermal grease shortage prompt and supplement the thermal grease by using a preset thermal grease supplement method;

[0037] when the number of missing areas is greater than the standard number of missing areas, report a thermal grease non-uniformity prompt;

[0038] respond to the thermal grease non-uniformity prompt to re-fix the cold head by using a preset cold head re-fixing method.

[0039] Optionally, the thermal grease supplement method comprises:

[0040] when the number of missing areas is 1, obtain a thermal grease supplement position based on the thermal grease missing area;

[0041] respond to the thermal grease supplement position to match a cold head loosening number, and control a preset cold head fixing device to loosen the bolt corresponding to the cold head loosening number by using a preset reversal distance value;

[0042] after the loosening is completed, control a preset thermal grease supplement device to insert into the thermal grease supplement position to supplement the thermal grease;

[0043] when the number of missing areas is greater than 1, combine the thermal grease missing area to generate a thermal grease supplement route;

[0044] control a preset cold head fixing device to loosen all the bolts by using a preset reversal distance value, and after the loosening is completed, control a preset thermal grease supplement device to supplement the thermal grease along the thermal grease supplement route.

[0045] Optionally, the method further comprises a silicon grease extrusion method:

[0046] collecting silicon grease parameter information and detection area number of the silicon grease missing area;

[0047] generating a silicon grease flow path according to the silicon grease supplement position, the silicon grease parameter information, and a preset silicon grease area injection amount;

[0048] knowing a priority fixing bolt based on the detection area number and a preset cold head bolt position;

[0049] generating a bolt fixing parameter in response to the priority fixing bolt and the silicon grease flow path;

[0050] controlling the cold head fixing device to perform bolt fixing according to the bolt fixing parameter, so as to complete silicon grease addition.

[0051] Optionally, the cold head re-fixing method comprises:

[0052] controlling a preset cold head fixing device to perform tightening fixing with a preset fixing force value, and collecting a tightening circle number and a tightening torque value of each of the tightening circle numbers;

[0053] when the tightening circle number exceeds a preset detection circle number, determining whether the tightening torque value of each of the tightening circle numbers is consistent;

[0054] if consistent, reporting a bolt abnormality prompt;

[0055] if inconsistent, controlling the cold head fixing device to continue tightening fixing, and collecting a current torque value;

[0056] when the current torque value is consistent with a preset complete fixing torque value, completing cold head fixing.

[0057] Optionally, the method further comprises:

[0058] in response to the bolt abnormality prompt, controlling the cold head fixing device to stop tightening operation, simultaneously performing reverse rotation, and after a preset reverse circle number, collecting an abnormal bolt image;

[0059] performing thread wear identification on a thread of the bolt from the abnormal bolt image to obtain a thread wear value;

[0060] determining whether the thread wear value exceeds a preset repair wear value;

[0061] if exceeding, reporting a bolt scrapping prompt;

[0062] if not exceeding, controlling a preset tapping device to tap the bolt again, and reporting a base abnormality prompt.

[0063] Secondly, this application provides an operation and maintenance management system based on a heterogeneous multi-chip cloud architecture, which adopts the following technical solution:

[0064] An operation and maintenance management system based on a heterogeneous multi-chip cloud architecture includes:

[0065] The data acquisition module is used to collect multi-dimensional data from the chip and identify chip performance bottlenecks.

[0066] The memory is used to store programs that implement any of the above-mentioned operation and maintenance management methods based on heterogeneous multi-chip cloud architecture;

[0067] The processor is used to load and execute programs stored in memory.

[0068] In summary, this application includes at least one of the following beneficial technical effects:

[0069] 1. The system monitors the load status in real time by collecting multi-dimensional data from chips, and generates precise resource allocation requirements based on performance bottleneck analysis. It then dynamically adjusts parameters to achieve intelligent scheduling and collaborative work across chip resources. Simultaneously, each resource adjustment log is stored in an operations and maintenance vector database. Vector retrieval technology enables rapid matching and reuse of historical scheduling strategies, forming knowledge accumulation and evolution capabilities. By dynamically optimizing resource allocation based on real-time chip status differences, the system improves the collaborative efficiency of heterogeneous chip clusters, thereby increasing resource utilization.

[0070] 2. The system first uses image recognition technology to determine the number of missing thermal paste areas. When only one missing area exists, the system precisely locates the replenishment position and loosens the corresponding bolts accordingly, avoiding excessive manipulation of the entire cold head assembly. When multiple missing areas exist, the system intelligently plans the replenishment route and simultaneously loosens all bolts to ensure the continuity and coverage of thermal paste replenishment. By using a preset reversal distance value, the system precisely controls the degree of bolt loosening, preventing cold head displacement due to excessive loosening while ensuring the thermal paste replenishment device can be smoothly inserted into the target position. Simultaneously, the dynamic generation algorithm for the thermal paste replenishment route considers the spatial distribution of the missing areas and the device's movement trajectory, making the replenishment process more in line with fluid dynamics principles and effectively improving chip heat dissipation efficiency.

[0071] 3. The system first collects thermal grease parameter information and the detection area numbers of missing areas. Combining the replenishment location with the preset injection volume, it uses a fluid dynamics model to generate the optimal flow path of the thermal grease between the chip and the cold head, ensuring complete coverage. Simultaneously, based on the spatial mapping relationship between the detection area numbers and the cold head bolt positions, it determines the priority bolts for fixing. By calculating the bolt fixing sequence and torque parameters, it ensures that the thermal grease spreads evenly along the preset path during the extrusion process. This achieves digital and precise control of the thermal grease extrusion process, reducing human error and accumulating historical extrusion strategies through an operation and maintenance vector database. This provides a standardized and intelligent maintenance method for large-scale heterogeneous cloud architecture cooling systems. Attached Figure Description

[0072] Figure 1 This is a flowchart of an operation and maintenance management method based on a heterogeneous multi-chip cloud architecture;

[0073] Figure 2 This is a flowchart of the silicone grease extrusion method. Detailed Implementation

[0074] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0075] Reference Figure 1 This application discloses an operation and maintenance management method based on a heterogeneous multi-chip cloud architecture, including the following steps:

[0076] S1: Acquire multidimensional data from the chip.

[0077] Multidimensional chip data refers to chip operating status parameters collected through a distributed agent, including real-time data on core temperature, power supply voltage, computing load, memory bandwidth, inter-chip communication latency, and other dimensions. The method of collecting chip operating status parameters through a distributed agent is common knowledge in the field and will not be elaborated upon here.

[0078] S2: Know the chip load based on multi-dimensional chip data.

[0079] Chip load status refers to the real-time operating status index of a chip calculated based on multi-dimensional chip data (such as core temperature, voltage, CPU / GPU utilization, memory bandwidth utilization, instruction execution rate, and other real-time operating parameters).

[0080] By standardizing the multidimensional data of the chip and unifying the dimensions of different indicators, a dynamic scheduling algorithm is used for fusion calculation to obtain the chip load. The specific standardized dimensions are predetermined by those skilled in the art and will not be elaborated here. The method of unifying the dimensions and the dynamic scheduling algorithm are common knowledge in the field and will not be elaborated here.

[0081] S3: Generate resource retrieval requests based on resource load conditions.

[0082] Resource retrieval requirements refer to the specific requirements for cross-chip resource calls. By combining dynamic scheduling algorithms to analyze and process the chip load (resource load), the specific requirements for cross-chip resource calls can be generated, thus obtaining the resource retrieval requirements. Dynamic scheduling algorithms are common knowledge in this field and will not be elaborated upon here.

[0083] S4: Performance bottleneck of the acquisition chip.

[0084] Chip performance bottlenecks refer to critical nodes in heterogeneous multi-chip cloud architectures where the computing, storage, and communication capabilities of a chip cannot be fully utilized due to hardware limitations or abnormal operating conditions. Chip performance bottlenecks are determined through real-time monitoring of multi-dimensional chip data combined with preset performance thresholds. These preset performance thresholds are pre-set by those skilled in the art and will not be elaborated upon here. The method of determining chip performance bottlenecks through real-time monitoring of multi-dimensional chip data combined with preset performance thresholds is common knowledge in the field and will not be elaborated upon here either.

[0085] S5: Combines resource retrieval requirements and chip performance bottlenecks to generate dynamically adjustable parameters.

[0086] Dynamic adjustment parameters refer to specific quantitative indicators used to guide the dynamic allocation of resources among heterogeneous multi-chip systems. Through dynamic scheduling algorithms, specific indicators in resource allocation requirements (such as the need for an additional 2GB of memory) are combined with performance bottlenecks (such as the current memory bandwidth remaining at 30%) to calculate the optimal adjustment parameters, thereby obtaining dynamic adjustment parameters and ensuring that resource allocation both meets the requirements and avoids bottlenecks.

[0087] S6: Based on dynamic parameter adjustment, perform multi-chip resource allocation, generate resource adjustment logs, and input the resource adjustment logs into the preset operation and maintenance vector database.

[0088] Resource adjustment logs are structured data that record the entire process of multi-chip resource allocation, used for tracing adjustment operations, analyzing maintenance effects, and locating faults.

[0089] An operations and maintenance vector database refers to a vector-based database used to store and manage operations and maintenance data for heterogeneous multi-chip systems, supporting rapid retrieval and similarity queries of high-dimensional data. The operations and maintenance vector database is pre-defined by those skilled in the art and will not be elaborated upon here.

[0090] After vectorizing the resource adjustment logs, they are input into the operations and maintenance vector database for subsequent retrieval. The generation method and vectorization process of the resource adjustment logs are common knowledge in this field and will not be elaborated here.

[0091] When resources are dynamically adjusted for multiple chips, resource adjustment logs need to be generated simultaneously, vectorized, and then input into the operation and maintenance vector database.

[0092] It also includes the following steps:

[0093] S7: Collect the current chip usage signals.

[0094] The signals used by the chip today refer to the electrical signals generated during chip operation. These signals are captured in real time by a pre-set signal acquisition module. The signal acquisition module is pre-configured by those skilled in the art and will not be described in detail here.

[0095] S70: Compare whether the current chip signal used is consistent with the preset reference chip signal used.

[0096] The signals used by the reference chip refer to the standard signal template during normal chip operation. The signals used by the reference chip are preset by those skilled in the art and will not be elaborated here.

[0097] By determining whether the signal used by the current chip is consistent with the signal used by the reference chip, we can determine whether the signal used by the current chip is abnormal.

[0098] S71: If they match, continue to collect the current chip's signal.

[0099] If the signal used by the current chip is the same as that used by the reference chip, it means that there is no abnormality in the signal used by the current chip, and you can continue to collect the signal used by the current chip.

[0100] S72: If they are inconsistent, compare the signal used by the current chip with the signal used by the reference chip to select the abnormal signal frequency band.

[0101] An abnormal signal frequency band refers to a frequency range where the signal used by the current chip deviates significantly from the reference signal. This is achieved by performing spectral analysis on the current chip's signal and the reference chip's signal to convert them into frequency domain characteristics. The difference between each frequency band is then calculated, and frequency bands exceeding a preset abnormal threshold are defined as abnormal signal frequency bands.

[0102] The methods for converting signals into frequency domain features and for calculating the differences between frequency bands are common knowledge in the field and will not be elaborated here.

[0103] If the signal used by the current chip is inconsistent with the signal used by the reference chip, it indicates that there is an anomaly in the signal used by the current chip. The frequency band of the abnormal signal needs to be selected first for subsequent steps.

[0104] S73: Match abnormal chips based on abnormal signal frequency bands and process abnormal chips using preset abnormal chip processing methods.

[0105] An abnormal chip refers to a chip whose operating state deviates from the normal standard. A pre-set frequency band chip lookup table can be used to find the abnormal chip corresponding to the abnormal signal frequency band. This table records different abnormal chips corresponding to different abnormal signal frequency bands. The reference content in the frequency band chip lookup table is formed by those skilled in the art by sequentially recording the different abnormal chips corresponding to different abnormal signal frequency bands, and will not be elaborated upon here.

[0106] The abnormal chip handling method refers to the method used to process abnormal chips. The specific abnormal chip handling method will be explained in detail in S730 to S734, and will not be repeated here.

[0107] Once an abnormal chip is identified, it must be processed using the methods for handling abnormal chips.

[0108] The abnormal chip handling method includes the following steps:

[0109] S730: Acquires solder joint image information of abnormal chips.

[0110] Solder joint image information refers to images of the solder joint area of ​​the defective chip. Solder joint image information is obtained by capturing images with a camera.

[0111] S731: Identify and locate the preset solder joint features from the solder joint image information to obtain the solder joint position.

[0112] Solder joint location refers to the specific location of the solder joints in a faulty chip. The solder joint location can be obtained by extracting the three-dimensional coordinates of the solder joint from the solder joint image information using computer vision algorithms. Computer vision algorithms are common knowledge in this field and will not be elaborated upon here.

[0113] S732: Controls the preset power-on test device to perform power-on tests on the solder joint location and collects the solder joint temperature value.

[0114] A power-on testing device is a device used to perform power-on testing on solder joints. The solder joint temperature value refers to the highest surface temperature of the solder joint captured by a preset infrared thermal imager when a rated current is applied to the solder joint for power-on testing.

[0115] The method of power-on testing is common knowledge in this field and will not be elaborated here.

[0116] S733: When the solder joint temperature exceeds the preset standard solder joint temperature, output the abnormal solder joint location and control the preset repair welding device to repair the abnormal solder joint location.

[0117] The standard solder joint temperature value refers to the temperature reference value used to detect whether the solder joint temperature value is abnormal. The standard solder joint temperature value is set in advance by those skilled in the art, and will not be elaborated here.

[0118] Abnormal solder joint locations refer to solder joint locations where the solder joint temperature exceeds the standard solder joint temperature value. Abnormal solder joint locations can be identified by recording these locations.

[0119] A repair welding device is a device used to repair abnormal weld points.

[0120] Since the contact resistance of a cold solder joint increases, and the heat generated when power is applied is positively correlated with the resistance, when the temperature value of the solder joint exceeds the standard solder joint temperature value, it indicates that the solder joint is cold. The abnormal solder joint location needs to be output first, and then the repair welding device should be controlled to repair the abnormal solder joint location.

[0121] S734: When the solder joint temperature does not exceed the preset standard solder joint temperature, a chip abnormality prompt is reported.

[0122] When the solder joint temperature does not exceed the standard solder joint temperature, it indicates that the solder joint is not cold-soldered, and the cause of the chip malfunction is due to the chip itself. A chip malfunction warning should be reported.

[0123] It also includes the following steps:

[0124] S80: Collects chip temperature values ​​and adjusts chip load.

[0125] The chip temperature value refers to the temperature of the chip. The chip temperature value is obtained by a temperature sensor preset on the chip die.

[0126] Post-adjustment chip load refers to the real-time operating status indicator of the chip after dynamic resource adjustments. The post-adjustment chip load is obtained in real time through the performance monitoring unit.

[0127] S81: Generates a chip reference temperature range based on the adjusted chip load conditions.

[0128] The chip reference temperature range refers to the range used to determine whether the chip's operating temperature is normal.

[0129] The reference temperature range of the chip corresponding to the adjusted chip load can be found by using the preset load temperature comparison table. The table records different reference temperature ranges of the chip corresponding to different adjusted chip loads. The comparison content in the load temperature comparison table is formed by those skilled in the art after conducting tests on different reference temperature ranges of the chip corresponding to different adjusted chip loads, and will not be elaborated here.

[0130] S82: Generate a reference temperature value by combining the chip temperature value and the preset thermal conductivity of the thermal paste if and only if the chip temperature value is greater than the maximum value of the chip reference temperature range.

[0131] Thermal conductivity of thermal grease is a physical parameter that measures the thermal conductivity of thermal grease. The thermal conductivity of thermal grease is known to those skilled in the art and will not be elaborated upon here.

[0132] The reference temperature value is a temperature threshold calculated using the chip temperature and the thermal conductivity of the thermal paste. It is used to determine whether there is a lack of thermal paste in the sensing area of ​​the cold head. The specific calculation formula is T. 基准 =T 芯片 ×k, where T 基准 T is the reference temperature value. 芯片 Here, k represents the chip temperature, and k is the thermal conductivity of the thermal grease. k has already been dimensionless. The specific method for dimensionless transformation is common knowledge in this field and will not be elaborated upon here.

[0133] S83: Collect the area temperature values ​​of all preset cold head sensing areas.

[0134] The cold head sensing area refers to the multiple temperature measurement zones divided between the contact surface between the heat sink and the chip. The cold head sensing area is predetermined by those skilled in the art and will not be elaborated here.

[0135] The zone temperature value refers to the temperature value of the sensing area of ​​the cold head. The zone temperature value is obtained by measuring the temperature value through the preset temperature sensors on each sensing area of ​​the cold head.

[0136] S84: Arrange the temperature values ​​of all cold head sensing areas from largest to smallest, and define the cold head sensing areas with temperature values ​​lower than the reference temperature as thermal paste missing areas, and process the thermal paste missing areas using the preset thermal paste missing processing method.

[0137] A region lacking thermal paste refers to a section of the cold head's sensing area where the temperature is lower than the reference temperature. When the temperature of this area is lower than the reference temperature, it indicates that the thermal paste in that area is insufficient, leading to increased thermal resistance.

[0138] The method for handling missing silicone grease refers to the method used to treat areas where silicone grease is missing. The specific method for handling missing silicone grease will be explained in detail in S840 to S844, and will not be repeated here.

[0139] After arranging the temperature values ​​of all the cold head sensing areas from largest to smallest, the cold head sensing areas with temperature values ​​lower than the reference temperature value should be defined as areas lacking thermal paste, so that the areas lacking thermal paste can be processed using the thermal paste lack processing method.

[0140] The method for handling missing silicone grease includes the following steps:

[0141] S840: Collects the number of missing areas in the silicone grease-deficient region.

[0142] The number of missing areas refers to the number of areas where thermal paste is missing. This number can be obtained by digitally recording the temperature values ​​of the cold head sensing areas that are lower than the reference temperature.

[0143] S841: Determine whether the number of missing regions is greater than the preset standard number of missing regions.

[0144] The standard missing area number refers to the threshold number used to determine the degree of silicone grease deficiency. The standard missing area number is predetermined by those skilled in the art and will not be elaborated here.

[0145] The specific cause of the missing silicone grease can be determined by judging whether the number of missing areas is greater than the standard number of missing areas.

[0146] S842: When the number of missing areas is not greater than the standard number of missing areas, report a message indicating insufficient thermal paste and replenish the thermal paste using the preset method.

[0147] The method for applying thermal paste refers to the procedure used to apply thermal paste between the cold head and the chip. Specific methods for applying thermal paste will be detailed in subsequent sections S8420 to S8424, and will not be repeated here.

[0148] When the number of missing areas is no greater than the standard number of missing areas, it indicates that the lack of thermal paste in the cold head sensing area is a localized problem and has not yet resulted in large-scale uneven distribution. In this case, a thermal paste shortage warning should be reported, and thermal paste should be replenished using the appropriate method.

[0149] S843: When the number of missing areas is greater than the standard number of missing areas, an uneven silicone grease warning will be reported.

[0150] When the number of missing areas exceeds the standard number of missing areas, it indicates that the distribution of thermal paste on the contact surface between the cold head and the chip has become uneven over a large area, rather than being a localized missing area. This suggests that the cold head fixing bolts were not tightened properly, resulting in insufficient initial compression of the thermal paste and uneven application. It is necessary to report the uneven thermal paste issue for subsequent steps.

[0151] S844: In response to the uneven application of thermal paste, the cold head is re-secured using the preset method for re-secured cold head.

[0152] The method of re-fixing the cold head refers to a complete process that uses automated equipment to retighten the cold head fixing bolts to adjust the pressure distribution on the contact surface between the cold head and the chip, thereby promoting the uniform filling of the thermal interface with thermal paste. The specific method of re-fixing the cold head will be explained in detail in the subsequent sections S8440 to S8444, and will not be repeated here.

[0153] When you receive a message indicating that the thermal paste is not evenly distributed, you need to re-secure the cold head using the same method as before to ensure that the thermal paste is evenly distributed across the heat dissipation interface.

[0154] The method for reapplying silicone grease includes the following steps:

[0155] S8420: When the number of missing regions is 1, the location of the silicone grease replenishment is obtained based on the missing silicone grease regions.

[0156] The thermal paste refill location refers to the specific position where the thermal paste refill device needs to be inserted into the area where thermal paste is lacking. The thermal paste refill location is the center point of the area where thermal paste is lacking; therefore, the geometric center coordinates of the area where thermal paste is lacking can be used to determine the thermal paste refill location.

[0157] In this embodiment, the sensing area of ​​the cold head is pre-defined with grid coordinates. When a certain area is determined to be a missing thermal paste area, the system automatically retrieves the three-dimensional coordinates of the geometric center point of that area to determine the insertion point of the thermal paste replenishment device, ensuring that the thermal paste spreads evenly outward from the center point to cover the entire missing area. The specific grid coordinates are preset by those skilled in the art and will not be elaborated here.

[0158] S8421: In response to the grease replenishment position, the cold head loosening number is matched, and the preset cold head fixing device is controlled to loosen the bolt corresponding to the cold head loosening number by a preset reversal distance value.

[0159] The cold head loosening number refers to the number of the cold head fixing bolt corresponding to the location of the silicone grease replenishment. In this embodiment, there are 4 cold head fixing bolts, which are arranged in a rectangular pattern.

[0160] The cold head loosening number corresponding to the silicone grease replenishment position can be found by using the preset supplementary bolt reference table. The table records the different cold head loosening numbers corresponding to different silicone grease replenishment positions. The reference content in the supplementary bolt reference table is formed by those skilled in the art by recording the different cold head loosening numbers corresponding to different silicone grease replenishment positions in sequence, which will not be elaborated here.

[0161] A cold head fixing device is a device used to fix the bolts of a cold head.

[0162] The reversal distance value refers to the number of rotations required to loosen the bolt, preventing excessive loosening that could cause the cold head to shift. This reversal distance value is preset by those skilled in the art and will not be elaborated upon here.

[0163] Once the cold head loosening number is matched, the cold head fixing device needs to be controlled to loosen the bolts corresponding to the cold head loosening number by reversing the distance value, so as to facilitate subsequent steps.

[0164] S8422: After the adjustment is completed, the preset silicone grease replenishment device is inserted into the silicone grease replenishment position to replenish the silicone grease.

[0165] A silicone grease refill device is a device used to refill silicone grease.

[0166] After the thermal paste loosening is completed, the thermal paste refill device needs to be inserted into the thermal paste refill position, and thermal paste should be refilled according to the preset thermal paste area injection amount. The thermal paste area injection amount refers to the amount of thermal paste used when refilling a thermal paste-deficient area. The thermal paste area injection amount is preset by those skilled in the art and will not be elaborated here.

[0167] S8423: When the number of missing regions is greater than 1, combine the missing regions of the thermal paste to generate a thermal paste replenishment route.

[0168] The thermal paste refill route refers to the path taken by the thermal paste refill device when refilling multiple areas lacking thermal paste. By identifying all the areas lacking thermal paste, the refill locations can be determined, and then a path optimization algorithm can be used to determine the thermal paste refill route. The path optimization algorithm is common knowledge in this field and will not be elaborated upon here.

[0169] When the number of missing regions is greater than 1, it indicates that there are multiple missing thermal paste regions at the same time. It is necessary to first generate a thermal paste replenishment route through a path optimization algorithm for subsequent steps.

[0170] S8424: Controls the preset cold head fixing device to loosen all bolts at a preset reverse distance value, and after the loosening is completed, controls the preset silicone grease replenishment device to replenish silicone grease along the silicone grease replenishment route.

[0171] The control device for fixing the cold head loosens all bolts by reversing the distance value. After loosening, the control device for replenishing the grease moves along the grease replenishment route and replenishes the grease according to the amount injected into the grease area.

[0172] Reference Figure 2 The silicone grease extrusion method includes the following steps:

[0173] S8425: Collects silicone grease parameter information and the detection area number of the silicone grease missing area.

[0174] Silicone grease parameter information refers to data on various characteristics and properties related to silicone grease. Examples include the thermal conductivity, viscosity, density, volatility, component ratio, and operating temperature range of the silicone grease. This parameter information is pre-entered by those skilled in the art and will not be elaborated upon here.

[0175] The detection area number refers to the number of each sensing area on the cold head. It is used to associate temperature data with physical location in order to pinpoint the specific location of areas lacking thermal paste. The detection area number is obtained through preset cold head sensing area division rules.

[0176] The sensing area of ​​the cold head is divided according to grid coordinates (such as 5×5 grid) and assigned a unique number (such as Row1_Col1, Row2_Col3) during system initialization. When the temperature of a certain area is determined to be abnormal, the system automatically retrieves the preset number of that area to obtain the detection area number.

[0177] The rules for dividing the sensing area of ​​the cold head are set in advance by those skilled in the art and will not be elaborated here.

[0178] S8426: Generates a silicone grease flow path based on the silicone grease replenishment location, silicone grease parameter information, and preset silicone grease injection amount for the designated area.

[0179] The thermal paste flow path refers to the expected trajectory of the thermal paste from the injection point to cover the entire missing area. A pre-set flow path lookup table can be used to find the thermal paste replenishment location, thermal paste parameters, and the corresponding thermal paste injection amount for each area. This table records different thermal paste flow paths corresponding to different thermal paste replenishment locations, thermal paste parameters, and thermal paste injection amounts. The reference information in the flow path lookup table was obtained by those skilled in the art using fluid dynamics simulation models to simulate the diffusion trajectory of thermal paste at the contact surface between the cold head and the chip; details will not be elaborated here.

[0180] S8427: Know the priority fixing bolt based on the detection area number and the preset cold head bolt position.

[0181] The position of the cold head bolts refers to the location of the fixing bolts of the cold head on the chip surface. The position of the cold head bolts is predetermined by those skilled in the art and will not be elaborated here.

[0182] The priority bolts are those that need to be tightened first. Following the principle of prioritizing the bolts furthest from the silicone grease injection point, the distance from each cold head bolt position to the center point of the corresponding silicone grease-deficient area is calculated, and the bolt furthest away is selected as the priority bolt. The distance calculation formula is common knowledge in this field and will not be elaborated here.

[0183] S8428: Responds to prioritizing bolt and grease flow paths to generate bolt fixing parameters.

[0184] Bolt fixing parameters refer to a specific combination of quantitative indicators used to guide the bolt tightening operation of the cold head fixing device. A preset bolt fixing reference table can be used to look up the bolt fixing parameters corresponding to the preferred fixing bolts and silicone grease flow paths. This table records different bolt fixing parameters corresponding to different preferred fixing bolts and silicone grease flow paths. The reference content in the bolt fixing reference table was compiled by those skilled in the art after conducting sequential experiments on the different bolt fixing parameters corresponding to different preferred fixing bolts and silicone grease flow paths; details will not be elaborated here.

[0185] S8429: Controls the cold head fixing device to fix the bolts with bolt fixing parameters, thereby completing the addition of thermal paste.

[0186] The control device for fixing the cold head uses bolt fixing parameters to fix the bolts, thereby completing the application of thermal grease.

[0187] The method for re-fixing the cold block includes the following steps:

[0188] S8440: Controls the preset cold head fixing device to tighten and fix it with a preset fixing force value, and collects the number of tightening turns and the tightening torque value of each tightening turn.

[0189] The fixing force value refers to the force required to fix the cold head. The fixing force value is set in advance by those skilled in the art and will not be elaborated here.

[0190] The number of tightening turns refers to the number of rotations of the bolts when the cold head fixing device is used to secure them. This number is obtained through an encoder built into the cold head fixing device. The encoder records the bolt rotation angle in real time and converts it into a number of turns.

[0191] The tightening torque value refers to the real-time torque corresponding to each number of tightening turns. The tightening torque value is acquired by a torque sensor integrated into the cold head fixing device.

[0192] S8441: When the number of tightening turns exceeds the preset number of detection turns, determine whether the tightening torque value of each tightening turn is consistent.

[0193] The number of test revolutions refers to the threshold number of revolutions used to detect whether the torque is abnormal. The number of test revolutions is preset by those skilled in the art and will not be elaborated here.

[0194] When the number of tightening turns exceeds the test number, it indicates that the bolt's fixation needs to be tested for torque to determine if there is any abnormality. It is necessary to first verify that the tightening torque value is consistent for each tightening turn in order to proceed with subsequent steps.

[0195] S8442: If consistent, report a bolt abnormality warning.

[0196] If they match, it means the bolts cannot be secured, and a bolt abnormality alert needs to be reported for subsequent steps.

[0197] S8443: If inconsistent, control the cold head fixing device to continue tightening and fixation, and collect the current torque value.

[0198] The current torque value refers to the bolt torque collected in real time. This current torque value is obtained through a torque sensor integrated into the cold head fixing device.

[0199] If there is a discrepancy, it means the bolts can be secured. The cold head fixing device needs to be tightened further, and the current torque value should be collected for subsequent steps.

[0200] S8444: When the current torque value matches the preset completion torque value, the cold head is fixed.

[0201] The target torque value refers to the torque value required to complete the cold head fixing operation. This target torque value is preset by those skilled in the art and will not be elaborated upon here.

[0202] When the current torque value matches the final fixed torque value, it indicates that the cold head fixing operation has been completed.

[0203] It also includes the following steps:

[0204] S8445: In response to a bolt abnormality alert, control the cold head fixing device to stop tightening, while simultaneously rotating in the opposite direction, and after a preset number of reverse rotations, acquire an image of the abnormal bolt.

[0205] The number of reverse rotations refers to the number of rotations the cold head fixing device makes to remove the bolts. The number of reverse rotations is preset by those skilled in the art and will not be elaborated here.

[0206] An abnormal bolt image refers to an image of the bolt after it has been removed. Abnormal bolt images are obtained by capturing images using a camera.

[0207] Upon receiving a bolt abnormality alert, the cold head fixing device must stop tightening and simultaneously rotate in the opposite direction. After rotating in the opposite direction a certain number of times, an image of the abnormal bolt should be captured for subsequent steps.

[0208] S8446: Identify thread wear from abnormal bolt images to obtain thread wear values.

[0209] Thread wear value refers to the degree of thread wear obtained by quantitatively analyzing the thread profile in an abnormal bolt image using image recognition technology. Image recognition technology is common knowledge in this field and will not be elaborated upon here.

[0210] S8447: Determine whether the thread wear value exceeds the preset repair wear value.

[0211] Repairable wear value refers to the maximum repairable wear threshold of the thread. The repairable wear value is preset by those skilled in the art and will not be elaborated here.

[0212] By determining whether the thread wear value exceeds the repair wear value, it can be determined whether the bolt can be repaired and used.

[0213] S8448: If the limit is exceeded, a bolt scrapping notice will be reported.

[0214] If the value exceeds this limit, it indicates that the thread strength is insufficient and the tightening ability cannot be restored through repair. A bolt scrapping notice must be reported.

[0215] S8449: If not exceeded, control the preset tapping device to re-tap the bolt and report a base abnormality.

[0216] A tapping device is a device used to repair worn threads.

[0217] If the torque requirement for fixing the cold head is not exceeded, it indicates that the thread can still meet the torque requirement after tapping repair. The tapping device should be controlled to re-tap the bolt, and the base condition should be checked after the bolt is repaired to avoid the thread from wearing again due to base problems. In other words, the base abnormality should be reported.

[0218] Based on the same inventive concept, embodiments of the present invention provide an operation and maintenance management system based on a heterogeneous multi-chip cloud architecture, comprising:

[0219] The acquisition module is used to acquire multi-dimensional chip data, chip performance bottlenecks, current chip usage signals, solder joint image information, solder joint temperature values, chip temperature values, adjusted chip load conditions, regional temperature values, number of missing regions, thermal paste parameter information, detection area number, number of tightening turns, tightening torque value, current torque value, and abnormal bolt images.

[0220] The memory is used to store the program that implements an operation and maintenance management method based on a heterogeneous multi-chip cloud architecture;

[0221] The processor is used to load and execute programs stored in memory.

[0222] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0223] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An operation and maintenance management method based on a heterogeneous multi-chip cloud architecture, characterized in that, include: Collect multidimensional data from the chip; The chip load can be determined based on the chip's multidimensional data. Based on the chip load conditions, resource retrieval requirements are generated. Collect chip performance bottlenecks; combine the resource retrieval requirements and chip performance bottlenecks to generate dynamic adjustment parameters; perform multi-chip resource allocation based on the dynamic adjustment parameters, generate resource adjustment logs, and input the resource adjustment logs into a preset operation and maintenance vector database; This also includes: The process involves collecting chip temperature values ​​and adjusted chip load conditions, where adjusted chip load conditions refer to the real-time operating status indicators of the chip after dynamic resource adjustments; generating a chip reference temperature range based on the adjusted chip load conditions; generating a baseline temperature value by combining the chip temperature value and a preset thermal conductivity of thermal grease only when the chip temperature value is greater than the maximum value of the chip reference temperature range; collecting the regional temperature values ​​of all preset cold head sensing areas; arranging the regional temperature values ​​of all cold head sensing areas from largest to smallest, and defining the cold head sensing areas with regional temperature values ​​lower than the baseline temperature value as thermal grease-deficient areas, while processing the thermal grease-deficient areas using a preset thermal grease-deficient processing method; The method for treating missing silicone grease includes: Collect the number of missing areas in the thermal paste-deficient region; determine whether the number of missing areas is greater than a preset standard number of missing areas; when the number of missing areas is not greater than the standard number of missing areas, report a thermal paste shortage warning and replenish thermal paste using a preset thermal paste replenishment method; when the number of missing areas is greater than the standard number of missing areas, report an uneven thermal paste distribution warning; in response to the uneven thermal paste distribution warning, re-fix the cold head using a preset cold head re-fixation method. The method for replenishing silicone grease includes: When the number of missing areas is 1, a silicone grease replenishment location is obtained based on the missing silicone grease area; in response to the silicone grease replenishment location, a cold head loosening number is matched, and a preset cold head fixing device is controlled to loosen the bolt corresponding to the cold head loosening number at a preset reversal distance value; after the loosening is completed, a preset silicone grease replenishment device is controlled to insert into the silicone grease replenishment location to replenish silicone grease; when the number of missing areas is greater than 1, a silicone grease replenishment route is generated based on the missing silicone grease areas; the preset cold head fixing device is controlled to loosen all bolts at a preset reversal distance value, and after the loosening is completed, the preset silicone grease replenishment device is controlled to replenish silicone grease along the silicone grease replenishment route.

2. The operation and maintenance management method based on a heterogeneous multi-chip cloud architecture according to claim 1, characterized in that, Also includes: Collect the current chip usage signals; Compare whether the current chip's signal is consistent with the preset reference chip's signal; If they match, continue collecting the current chip's usage signal; If they are inconsistent, the signal used by the current chip is compared with the signal used by the reference chip to select the abnormal signal frequency band. The abnormal chip is matched based on the abnormal signal frequency band, and the abnormal chip is processed using a preset abnormal chip processing method.

3. The operation and maintenance management method based on a heterogeneous multi-chip cloud architecture according to claim 2, characterized in that, The abnormal chip processing method includes: Collect solder joint image information of the abnormal chip; The solder joint position is obtained by identifying and locating preset solder joint features from the solder joint image information; The preset power-on test device is controlled to perform power-on test on the solder joint location and collect the solder joint temperature value; When the temperature value of the solder joint exceeds the preset standard solder joint temperature value, the abnormal solder joint location is output, and the preset repair welding device is controlled to repair the abnormal solder joint location. When the temperature of the solder joint does not exceed the preset standard solder joint temperature value, a chip abnormality prompt is reported.

4. The operation and maintenance management method based on a heterogeneous multi-chip cloud architecture according to claim 1, characterized in that, This also includes silicone grease extrusion methods: Collect silicone grease parameter information and the detection area number of the silicone grease-deficient area; A silicone grease flow path is generated based on the silicone grease replenishment location, the silicone grease parameter information, and the preset silicone grease injection amount for the region. Based on the detection area number and the preset cold head bolt position, the priority fixing bolt is determined; In response to the preferred fixing bolt and the silicone grease flow path, bolt fixing parameters are generated; The cold head fixing device is controlled to fix the bolts according to the bolt fixing parameters, thereby completing the addition of silicone grease.

5. The operation and maintenance management method based on a heterogeneous multi-chip cloud architecture according to claim 1, characterized in that, The method for re-fixing the cold head includes: The preset cold head fixing device is controlled to tighten and fix with a preset fixing force value, and the number of tightening turns and the tightening torque value of each tightening turn are collected; When the number of tightening turns exceeds the preset number of detection turns, it is determined whether the tightening torque value of each tightening turn is consistent; If they match, report the bolt abnormality. If there is a discrepancy, the cold head fixing device is controlled to continue tightening and fixing, and the current torque value is collected; When the current torque value matches the preset torque value for completion, the cold head is fixed.

6. The operation and maintenance management method based on a heterogeneous multi-chip cloud architecture according to claim 5, characterized in that, Also includes: In response to the bolt abnormality alert, the cold head fixing device is controlled to stop tightening and rotate in the opposite direction. After a preset number of reverse rotations, an image of the abnormal bolt is captured. The wear value of the thread is obtained by identifying the thread wear of the bolt from the abnormal bolt image; Determine whether the thread wear value exceeds the preset repair wear value; If the quantity exceeds the limit, a bolt scrapping notice should be reported. If the error does not exceed the limit, the preset tapping device will be controlled to re-tap the bolts and an abnormality warning for the base will be reported.

7. An operation and maintenance management system based on a heterogeneous multi-chip cloud architecture, characterized in that, include: The data acquisition module is used to collect multi-dimensional data from the chip and identify chip performance bottlenecks. A memory for storing a program that implements an operation and maintenance management method based on a heterogeneous multi-chip cloud architecture as described in any one of claims 1 to 6; The processor is used to load and execute programs stored in memory.

Images