Server case self-cleaning system and method based on composite cleaning and intelligent decision

Abstract

The invention discloses a server case self-cleaning system and method based on composite cleaning and intelligent decision making, and the system comprises a composite cleaning execution unit, the inner wall of a case and the surface of a heat dissipation assembly are covered with a nano catalytic coating containing a metal organic framework and a compound enzyme, organic pollutants are decomposed, and antibacterial ions are released; micro piezoelectric ceramic pieces are embedded in the heat dissipation fins, the functions of ultrasonic vibration dust removal and thermoelectric power generation are achieved, and the temperature difference between the fins and air is used for supplying energy to the system. The miniature cleaning mechanical arm brush head is of a shape memory alloy and graphene coating combined structure and can be flexibly matched with gaps and self-repair abrasion, the cleaning efficiency and comprehensiveness are remarkably improved, and the service life of the miniature cleaning mechanical arm brush head is prolonged. Full-granularity coverage cleaning from nanoscale (less than or equal to 50nm) to macroscopic particles is realized through multi-dimensional cooperation (decomposition of organic pollutants by a nano catalytic coating, stripping of accumulated dust by piezoelectric vibration, directional removal of charged particles by liquid-electricity driving and the like) of the composite cleaning execution unit.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of server cabinet cleaning, and particularly relates to a server cabinet self-cleaning system and method based on composite cleaning and intelligent decision-making, BACKGROUND

[0002] Server cabinet cleaning mainly relies on mechanical cleaning, airflow blowing and filtering interception. Mechanical cleaning usually adopts fixed brushes or telescopic scrapers to physically wipe the inner wall of the cabinet and the heat dissipation components through a preset trajectory. Airflow blowing uses high-pressure gas or fans to form directional airflow to discharge floating dust from the cabinet. Filtering interception usually sets a dust screen at the air inlet to screen and block external particles through the filter aperture. At the same time, some systems introduce simple sensing control, such as triggering the cleaning action through a dust concentration sensor or starting the maintenance program according to a fixed period. These technologies are widely used in conventional data center environments and can meet the basic cleaning needs, delay the impact of dust accumulation on server heat dissipation performance to a certain extent, and ensure the basic operation stability of the equipment.

[0003] However, the existing technology has obvious limitations in cleaning efficiency, intelligent adaptation and environmental adaptability. Firstly, the cleaning granularity is insufficient, and mechanical cleaning and airflow blowing cannot remove nano-level dust and stubborn dust in the micro-holes of the heat dissipation fins, and the filtering interception has low blocking efficiency for particles less than 50 nm. Secondly, the decision-making mechanism is simple and relies on a single concentration threshold or a fixed time trigger, which is not linked with core operating parameters such as server CPU temperature and load rate, and is easy to interfere with system performance at high load or waste resources at low load. Thirdly, the energy consumption and maintenance cost are high, and the traditional solution relies on external energy and lacks resource recycling design, requiring frequent manual intervention. Fourthly, the extreme environment adaptability is weak, and simple physical cleaning cannot cope with the invasion of pollutants in dusty and humid environments, increasing the risk of equipment failure. These shortcomings make it difficult for the existing technology to meet the development needs of high-density, unattended and green data centers. Therefore, the application provides a server cabinet self-cleaning system and method based on composite cleaning and intelligent decision-making, SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the application provide a server cabinet self-cleaning system and method based on composite cleaning and intelligent decision-making, which solves the problems of incomplete cleaning, lack of intelligent decision-making, high energy consumption, poor adaptability and high maintenance cost in the prior art,

[0005] To achieve the above object, the application provides the following technical scheme:

[0006] The server cabinet self-cleaning system based on composite cleaning and intelligent decision-making comprises:

[0007] The composite cleaning execution unit is covered with a nano-catalytic coating containing metal organic frameworks and composite enzymes on the inner wall of the case and the surface of the heat dissipation assembly to decompose organic pollutants and release antibacterial ions; the heat dissipation fins are embedded with micro piezoelectric ceramic sheets, which have the functions of ultrasonic vibration dust removal and thermoelectric power generation, and the system is powered by the temperature difference between the fins and the air; the micro cleaning mechanical arm brush head adopts a combination structure of shape memory alloy and graphene coating, which can flexibly adapt to gaps and self-repair wear; the surfaces of key components such as the mainboard, memory module and hard disk in the case are coated with an anti-static coating, which is matched with an adjustable static electricity generating device (the output intensity is dynamically adjusted according to the environmental temperature and humidity), to actively adsorb floating dust particles; a micro dust collection device (containing a high-efficiency silent turbine fan) is integrated, which is matched with a hidden three-dimensional air duct (the inner wall is made of polytetrafluoroethylene material), the air duct inlet is distributed in the dust-prone areas such as the CPU radiator and power module, and the outlet is connected with a replaceable filter screen; the micro cleaning mechanical arm, dust collection device and static electricity generating device adopt hot plug modular design, supporting quick replacement without disassembling the server body;

[0008] The intelligent decision control module dynamically optimizes the cleaning strategy by fusing pollution data of multiple cases through a distributed learning network under the premise of protecting data privacy; an event-driven pulse neural network chip is used to simulate the correlation between pollution, heat dissipation and energy consumption in real time combined with the CPU temperature and load rate of the server to predict the dust accumulation risk and autonomously optimize the cleaning scheme; a digital twin engine couples a nanometer physical model with a macro system simulation to dynamically plan the cleaning topology; the intelligent decision algorithm fuses multi-dimensional data such as server load, dust concentration and environmental temperature and humidity to dynamically match the cleaning mode: intermittent weak suction cleaning mode is started when the load is low and the concentration is low (running for 1 minute every 2 hours); switch to strong suction continuous cleaning mode when the load is high and the concentration is high; reduce the strength of the static electricity generating device when the humidity is too high (relative humidity > 60%); add a remote monitoring and early warning sub-module to upload data such as cleaning status, filter screen blockage degree and equipment failure to the cloud management platform through Internet of Things technology, support mobile APP / webpage viewing, and push early warning notifications and solutions (such as "filter screen blockage: suggest replacing within 12 hours") when there is a fault;

[0009] The pollution monitoring and tracing network deploys diamond color heart quantum sensors at key positions to realize dust positioning and component identification with a particle size of ≤50nm; a microfluidic chip sensor analyzes the source of dust and links the sodium alginate-based hydrogel protective layer at the air inlet of the case, which can be controlled to secrete to intercept specific pollutants; an intelligent sensor group is added, including: dust concentration sensor (real-time detection of dust concentration in different areas), temperature and humidity sensor (providing basis for static electricity device adjustment), pressure sensor (deployed at key nodes of the air duct to monitor air duct resistance and filter screen blockage status);

[0010] Multi-field synergistic fluid architecture, liquid cooling loop integrated electric field modulation microfluidic channel, stubborn dust accumulation is stripped by electrophoresis of suspended charged nanoparticles; phase change self-adaptive topology component driven by cholesteric liquid crystal reconfigurable microchannel switches negative pressure suction and flow guide cleaning mode with dust heat resistance changes; ion migration auxiliary flow channel cooperates with airflow laminar flow channel to directively remove conductive dust and suppress secondary pollution; airflow laminar flow channel links with hidden air duct and miniature dust collection device to form a closed-loop air circulation path, ensuring that the dust collection device efficiently collects the dust stripped by piezoelectric vibration and suppresses secondary diffusion.

[0011] Preferably, the composite cleaning execution unit comprises a self-repairing cleaning component. When the brush head of the mechanical arm is worn out, the shape memory alloy is restored to its original shape by energizing heating, and the graphene coating fills in the wear gap and maintains the electrostatic adsorption performance. The vibration sensor monitors the wear frequency of the brush head, and automatically completes calibration and predicts the replacement cycle when the server load rate is less than or equal to 20%. The hot plug modular design supports quick replacement without disassembling the server body when the brush head reaches the replacement cycle, and the replacement time is controlled within a few minutes.

[0012] Preferably, the composite cleaning execution unit comprises an energy cycle symbiotic module. The thermoelectric power generation device supplies power to the cleaning system, and the collected carbon-containing dust is cracked into synthesis gas by microwave plasma, which is used to prepare cleaning consumables or nanozyme carriers, forming a pollution-energy-material closed loop. The hot plug modular design supports quick replacement without disassembling the server body when the brush head reaches the replacement cycle, and the replacement time is controlled within a few minutes.

[0013] Preferably, the intelligent decision control module comprises a pollution tracing response mechanism. After locating the dust source through data analysis, the air inlet temporary pre-filter layer is started by sending a filter screen replacement warning through the machine room management system. When PCB board oxidation particles are detected, local dry cleaning and moisture control are triggered synchronously, and the air flow cleaning module is preferentially started when the CPU temperature is greater than or equal to 80°C. Combined with pressure sensor data to judge the degree of filter screen blockage, when the blockage exceeds the threshold, the cloud management platform pushes a warning notification of "filter screen blockage: suggest replacing within 12 hours", and the temporary bypass air duct is started to maintain basic heat dissipation.

[0014] Preferably, the intelligent decision control module comprises an edge inference elastic expansion component. The cleaning resource pooled cabinet deploys a hot plug neuro-morphic accelerator card, dynamically enhances real-time inference capability in high pollution scenarios, and avoids single node algorithm bottleneck. The remote monitoring and early warning sub-module uploads the running state and algorithm load data of the edge inference node to the cloud management platform, supporting remote viewing and scheduling of accelerator card resources by operation and maintenance personnel.

[0015] Preferably, the pollution monitoring and tracing network comprises a quantum sensing-electromagnetic resonance composite cleaning unit. After positioning the nanoscale dust, the near-field electromagnetic resonance antenna array applies resonance energy waves to make the particles detach from the surface, and the magnetic control nanometer collector synchronously captures the charged dust. For the carbon deposition in the micro-holes of the circuit board, a time-domain finite difference optimized microwave emission source outputs 10 4 -10 5 W / m 2 Thermoacoustic pulses are used to remove the carbonized layer. The dust concentration sensor in the intelligent sensor group monitors the macro dust concentration in real time, and the quantum sensor data is fused to realize "nanoscale + macro level" full-scale dust state perception.

[0016] Preferably, the multi-field collaborative fluid architecture comprises a liquid-electricity-gas-ion flow ternary composite flow channel. The main liquid cooling circuit carries away the core heat, the micro-pressure gas flow laminar channel maintains positive pressure filtration, and the ion migration channel drives the salt deposition ions to migrate to the collector under the electric field gradient. The dielectric wetting valve dynamically switches the flow channel, closes the ion channel to save energy at low demand, and activates the collaborative path at high risk. The hidden air duct seamlessly connects with the gas laminar flow channel, and the fan power of the miniature dust collection device is dynamically adjusted according to the gas laminar flow intensity to ensure the balance between dust collection efficiency and gas flow energy consumption.

[0017] Preferably, the server case self-cleaning method based on composite cleaning and intelligent decision-making includes the following steps:

[0018] (1) Real-time pollution perception and composition analysis. The diamond color center quantum sensor and microfluidic chip are used to identify the dust particle size, composition and source. The quantum sensing accuracy is ≤50 nm, and the image recognition is used to classify the dust types such as fiber, metal and salt mist. The intelligent sensor group synchronously collects dust concentration, environmental temperature and humidity, and air duct pressure data, and uploads them to the intelligent decision-making control module.

[0019] (2) Distributed migration learning decision-making. The local data of multiple cases are aggregated and trained to generate a global model after differential privacy protection. A small sample predictor based on graph neural network and hierarchical reinforcement learning generates a hierarchical cleaning path combined with CPU temperature and load rate. The server load, dust concentration, temperature and humidity data are fused to match the corresponding cleaning mode, which is intermittent weak suction, continuous strong suction, and adjustment of electrostatic intensity.

[0020] (3) Composite cleaning strategy execution. According to the cleaning intensity-heat dissipation efficiency-energy consumption surface simulated by digital twin, the optimal parameters are determined by reinforcement learning. The liquid-electricity, piezoelectricity and gas flow modules switch the operation mode according to the dust type. The electrostatic generator adjusts the output intensity according to the temperature and humidity data, and the miniature dust collection device starts with the hidden air duct to realize the closed loop of adsorption, stripping and collection.

[0021] (4) Resource closed-loop management, waste heat is used to power the catalytic decomposition module, and electrolytic deposition salt crystallization is converted into a recyclable solution; carbon-based dust is converted into synthesis gas for consumable production, and mycelium-based materials are used to make clean components to reduce carbon emissions; when the filter screen is saturated, a remote early warning prompt is given to replace it, and the modular components are hot-plugged and maintained according to demand;

[0022] (5) System self-evolution fault tolerance, event-driven pulse neural network chip updates the pollution-strategy mapping library according to cleaning feedback, iteratively optimizes parameters through genetic algorithm, and searches for alternative solutions using simulated annealing algorithm when hardware fails and rolls back to the strategy backup tree, and parameter iteration is completed within 24-48 hours after cleaning; the remote monitoring sub-module records cleaning effect and equipment state data to provide basis for strategy iteration.

[0023] Preferably, in the pollution real-time sensing and composition analysis step, after the quantum sensor locates the nanoscale dust, the near-field electromagnetic resonance antenna array applies resonance energy waves to make the particles detach from the surface, and the magnetic control nanometer collector synchronously captures; for carbon deposition in the micro-holes of the circuit board, the output 10 4 -10 5 W / m 2 Microwave thermoacoustic pulse removes carbonized layer; dust concentration sensor monitors macro dust concentration in real time, and triggers the cleaning strategy execution step when the concentration exceeds the threshold.

[0024] Preferably, in the composite cleaning strategy execution step, the liquid-electric module applies a pulse voltage when the cooling liquid flows through the dust accumulation fin, driving the charged nanoparticles to electrophoretically strip the stains; the cholesteric liquid crystal phase transition triggers the expansion of the micro-channel to form a negative pressure cavity to suck the dead angle dust, or shrinks into a flow guide groove to guide the airflow to remove the obstacle, and the server load rate is ≥80% automatically reduces the running noise of the mechanical components; the electrostatic generator reduces the intensity when the humidity is too high (relative humidity > 60%), the miniature dust collector switches the operating power according to the cleaning mode, and the hidden air duct ensures air circulation covering the entire area of the case.

[0025] The technical effects and advantages of the server case self-cleaning system and method based on composite cleaning and intelligent decision-making of the present application are:

[0026] 1、The present application significantly improves cleaning efficiency and comprehensiveness, through multi-dimensional cooperation of the composite cleaning execution unit (nanometer catalytic coating decomposes organic pollutants, piezoelectric vibration peels off dust, liquid-electric drive charged particles are directionally removed, etc.), realizing full-size coverage cleaning from nanometer (≤50nm) to macro-particles, effectively removing dead angles such as gaps between heat dissipation fins and micro-holes of circuit boards that are difficult to reach by traditional cleaning, solving the problem of heat dissipation efficiency decay caused by dust accumulation.

[0027] 2、The invention, intelligent decision-making realizes dynamic adaptation and precise control, the intelligent decision-making control module combines the core operation parameters such as server CPU temperature, load rate, dynamically optimizes cleaning strategy through distributed learning and pulse neural network, automatically reduces mechanical noise in high-load scenarios, preferentially executes deep cleaning and component self-repair in low-load period, avoids interference to server core business, and balances cleaning effect and system performance.

[0028] 3、The invention, energy consumption optimization and resource recycling, the temperature difference power generation and energy cycle symbiosis module uses server waste heat to supply energy for the cleaning system, reduces dependence on external energy;The collected carbon-containing dust is converted into synthesis gas for consumable preparation, forms a“pollution-energy-material”closed loop, reduces carbon emissions and consumable cost, and meets the sustainable development needs of green data center.

[0029] 4、The invention, extreme environment adaptability and system reliability enhancement, pollution monitoring and tracing network accurately identifies the source and composition of pollutants through quantum sensing and microfluidic chip, links the protective adhesive layer and directional removal unit, effectively intercepts external pollutants in extreme environments such as sand and high humidity, reduces the risk of circuit short circuit and other faults;Self-evolution fault-tolerant mechanism improves the stability of long-term operation of the system through strategy iteration and backup rollback.

[0030] 5、The invention, maintenance cost reduction and operation and maintenance efficiency improvement, self-repairing cleaning components prolong the service life of brush heads and other vulnerable parts, edge reasoning elastic expansion supports efficient management of large-scale clusters, reduces the frequency of manual maintenance and downtime, significantly reduces operation and maintenance costs, and improves the continuous operation capability of server clusters by avoiding hardware failures caused by dust accumulation through intelligent early warning. BRIEF DESCRIPTION OF DRAWINGS

[0031] Fig. 1 The system module block diagram of the server case self-cleaning system and method based on composite cleaning and intelligent decision-making is provided by the present application;

[0032] Fig. 2 The method flowchart of the server case self-cleaning system and method based on composite cleaning and intelligent decision-making is provided by the present application. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present application will be described in detail below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor are within the scope of protection of the present application,

[0034] It should be noted that, in this article, the relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations, and the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment, without more limitation, the elements defined by the statement "include" do not exclude the presence of other identical elements in the process, method, article or equipment including the elements,

[0035] Embodiment 1

[0036] Reference Figs. 1-2 The embodiment provides a server case self-cleaning system and method based on composite cleaning and intelligent decision-making, which is used for a standard server room self-cleaning system, and specific implementation contents include:

[0037] Implementation purpose: solve the problem of heat dissipation efficiency reduction caused by dust accumulation in a conventional server room, realize low-energy-consumption and automatic internal case cleaning through the cooperation of a composite cleaning unit (nanometer catalysis + piezoelectric vibration + electrostatic adsorption + micro dust collection) and intelligent decision-making, and adapt to general computing scenarios with large CPU load fluctuations.

[0038] System architecture:

[0039] Composite cleaning execution unit:

[0040] The inner wall of the case is covered with a nanometer catalytic coating (thickness 50-100 μm) containing a metal organic framework and a composite enzyme, and a piezoelectric ceramic sheet (resonance frequency 30-50 kHz) with a diameter of 5 mm is embedded in the heat dissipation fin;

[0041] The surface of the mainboard and the memory module in the case is covered with a 5 μm thick anti-static coating, and a adjustable electrostatic generating device (output voltage 5-15 kV, reduced to 5-8 kV when humidity > 60%, and increased to 12-15 kV when humidity < 30%) is matched;

[0042] Integrated micro dust collection device (high-efficiency silent turbine fan, power 5W), matched with hidden three-dimensional air duct (inner wall made of polytetrafluoroethylene material, air resistance ≤50Pa), air duct inlet distributed on the top of CPU radiator and power module, outlet connected with replaceable filter screen (filtration precision 0.3 μm);

[0043] The micro robotic arm (length 8-12 cm) brush head adopts a 0.1 mm thick shape memory alloy base + 5 μm graphene coating, and the dust collection device and static electricity generation device are designed with hot plug modularization (replaced through the special interface on the side of the case, and the single module replacement time is ≤3 minutes).

[0044] Intelligent decision control module:

[0045] The event-driven pulse neural network chip (computing power 1 TOPS / W) is mounted, communicates with the server BMC through the PCIe interface, and obtains the CPU temperature (sampling frequency 1 Hz) and load rate (sampling frequency 0.1 Hz) in real time;

[0046] The digital twin engine is deployed on the edge server in the computer room, and the simulation accuracy reaches 0.1 mm level.

[0047] A remote monitoring and early warning sub-module is added, which uploads the cleaning state, filter screen blockage degree (based on pressure sensor data), and equipment fault information to the cloud management platform through Internet of Things technology, supports mobile phone APP viewing and early warning push (such as “filter screen blockage: suggest replacing within 24 hours”).

[0048] Pollution monitoring and tracing network:

[0049] One diamond color center quantum sensor is arranged near each of the air inlet of the case and the CPU cooling fin (detection range 0-1000 particles / cm 3 );

[0050] An intelligent sensor group is added: two dust concentration sensors (monitoring the CPU area and the power area, range 0-500 particles / cm 3 ), one temperature and humidity sensor (sampling frequency 1 Hz), and one pressure sensor (outlet of the air duct, monitoring the filter screen resistance, threshold 100 Pa);

[0051] A sodium alginate-based hydrogel protective adhesive layer (area 10×10 cm 2 ) is arranged inside the air inlet and connected to the micro pump-controlled secretion device.

[0052] Multi-field collaborative fluid architecture:

[0053] The liquid cooling circuit integrates an electric field modulation microchannel with a width of 0.5 mm, and built-in TiO2@graphene charged nanoparticles (concentration 0.1wt%);

[0054] The dielectric wetting valve (response time <10 ms) controls the ion migration channel switch, and the laminar flow channel is linked with the hidden air duct and the micro dust collection device to form a closed-loop air circulation.

[0055] Implementation steps:

[0056] (1) Pollution perception and analysis: Quantum sensors detect nanoscale dust (≤50 nm) in real time and microfluidic chips analyze the composition (such as fibers, metal debris); intelligent sensor groups synchronously collect macro dust concentration (such as CPU area > 200 particles / cm 3 ), environmental temperature and humidity (such as 25℃, 50% RH) and air duct pressure data, and upload them to the intelligent decision module after fusion with image recognition results.

[0057] (2) Distributed decision generation: Fusion of local pollution data from 10 servers of the same model in the computer room trains a global model; combined with current CPU temperature (such as 75℃) and load rate (60%), a hierarchical cleaning path is generated through graph neural network, and the piezoelectric vibration + electrostatic adsorption + dust removal collaborative mode is preferentially started.

[0058] (3) Compound cleaning execution: Piezoelectric ceramic sheet vibrates at a frequency of 40 kHz for 10 seconds to remove dust on the surface of the heat dissipation fin; electrostatic generator is started (output 10 kV) to form an electrostatic field on the surface of the mainboard and memory to adsorb floating dust; the micro dust removal device operates at a speed of 3000 rpm for 5 seconds to suck the accumulated dust into the filter screen through the hidden air duct; the dielectric wetting valve closes the ion channel to save energy.

[0059] (4) Resource closed-loop management: The temperature difference power generation module (output voltage 3.3V) supplies power to the sensor and control circuit; the collected dust is temporarily stored in the filter screen, which is replaced after remote early warning prompt, and the filter screen frame can be recycled.

[0060] (5) System self-evolution: Within 24 hours after cleaning, the neural network chip updates the strategy parameters according to the change of heat dissipation efficiency (such as CPU temperature reduced to 68℃), and establishes a linkage mapping relationship of "vibration frequency - electrostatic strength - dust removal power".

[0061] Implementation effect: After continuous operation for 30 days, the nanoscale dust removal rate in the case is > 90%, the macro dust (50nm-10μm) removal rate reaches 95%; the average CPU temperature is reduced by 10℃, the speed of the heat dissipation fan is reduced by 20%, and the total energy consumption is reduced by 8%; remote monitoring makes the filter screen replacement early warning accuracy 100%, modular design shortens the maintenance time of a single server from 30 minutes to 5 minutes, and the operation and maintenance efficiency is improved by 80%.

[0062] Example 2

[0063] This embodiment provides a server case self-cleaning system and method based on composite cleaning and intelligent decision, which is used for desert area data center enhancement system (anti-dust version), and the specific implementation content includes:

[0064] Implementation purpose: For high sand dust environment in desert (PM10 concentration often > 500μg / m 3), strengthen pollution tracing and active interception ability, solve the risk of circuit short circuit caused by silicate particle deposition through enhanced electrostatic adsorption, high negative pressure dust collection and linkage early warning, and adapt to stable operation of servers in extreme environment.

[0065] System architecture:

[0066] Composite cleaning execution unit:

[0067] The area of the air inlet protection adhesive layer is expanded to 20x20cm 2 , and 0.5wt% nano calcium carbonate is added to enhance viscosity;

[0068] The anti-static coating adds 0.3wt% nano calcium carbonate particles to enhance the adsorption capacity of silicate dust; the output voltage of the electrostatic generator is increased to 15-20kV;

[0069] The thickness of the graphene coating of the mechanical arm brush head is increased to 10μm, the micro dust collection device is upgraded to a high negative pressure model (power 10W, wind pressure 200Pa), the hidden air duct inlet is additionally provided with a stainless steel dustproof grid (pore size 1mm), and the filter screen replacement prompt linkage cloud platform is added.

[0070] Pollution monitoring and tracing network:

[0071] Three microfluidic chip sensors are added (deployed at the air inlet of the case, the power module, and the memory slot), which can identify silicate, iron oxide and other dust characteristic components; the detection accuracy of the quantum sensor is improved to ≤30nm;

[0072] The dust concentration sensor in the intelligent sensor group is linked with the microfluidic chip. When the proportion of silicate particles is >30% and the concentration is >500 particles / cm 3 , trigger double early warning (local + cloud).

[0073] Intelligent decision control module:

[0074] Integrate dust pollution special algorithm model. When the proportion of silicate particles is >30%, send filter screen replacement warning through SNMP protocol linkage data center management system.

[0075] Implementation steps:

[0076] (1) Pollution perception and analysis: Quantum sensor locates nanoscale silicate particles, microfluidic chip confirms that the dust source is the external environment; intelligent sensor group detects dust concentration >800 particles / cm 3 , humidity 20%, image recognition shows that the dust thickness in the fin gap is 50μm.

[0077] (2) Distributed decision-making generation: call the historical migration model of the desert data center, combined with the current CPU temperature (82°C), decide to start the air inlet adhesive secretion + electrostatic adsorption + high negative pressure dust removal collaborative mode in priority.

[0078] (3) Composite cleaning execution: the adhesive secretion device pumps in 0.5ml hydrogel to form a temporary high-viscosity interception layer; the electrostatic generation device operates at a high strength of 20kV to adsorb small sand dust that penetrates the adhesive layer; the miniature dust removal device operates at a high speed of 5000rpm for 10 seconds to remove gap dust through intensified air duct negative pressure; the airflow laminar channel cooperates with the air duct to form a closed loop circulation to avoid secondary diffusion; for circuit board micropore carbon deposition, output 10 4 W / m 2 Thermoacoustic pulse removal.

[0079] (4) Resource closed-loop management: the collected sand dust is cracked by microwave plasma (power 500W), and the converted synthesis gas is temporarily stored in a high-pressure tank for the preparation of nanozyme carriers.

[0080] (5) System self-evolution: the neural network adjusts the secretion amount according to the sand dust interception efficiency (adhesive capture rate > 80%), and optimizes the corresponding relationship of "sand dust concentration - electrostatic strength - dust removal power".

[0081] Implementation effect: the dust accumulation in the case under the sand dust environment is reduced by 70% compared with the traditional scheme, and the circuit short circuit failure rate is 0 (the traditional scheme is 1-2 times per month); the electrostatic adsorption + dust removal cooperation makes the sand dust removal rate increase by 40%, the remote early warning response time is <5 seconds, which is 24 hours earlier than manual inspection to find high-risk areas; the maintenance cycle is extended to 4 months, and the replacement time of the modular filter screen is shortened to 2 minutes.

[0082] Example 3

[0083] This example provides a server case self-cleaning system and method based on composite cleaning and intelligent decision-making, which is used for financial level high-load server system, and the specific implementation content includes:

[0084] Implementation purpose: adapt to high-load scenes such as financial transactions (CPU load rate often > 90%), through the linkage of silent composite cleaning (low-noise electrostatic + dust removal + liquid-electric) and transaction system, reduce the running noise while ensuring the cleaning effect, avoid interfering with precision electronic components, and ensure business continuity.

[0085] System architecture:

[0086] Composite cleaning execution unit:

[0087] The mechanical arm driving motor is replaced with a brushless silent type (noise ≤ 30dB), and the brush head shape memory alloy is added with 0.1wt% nickel element to enhance flexibility;

[0088] The electrostatic generator is added with a "silent mode": when the load rate is ≥80%, the output voltage is automatically reduced to 5-8kV, reducing electromagnetic noise;

[0089] The mini dust collection device is integrated with sound-absorbing cotton (reducing noise by 15dB), and the air duct adopts an arc-shaped transition design (reducing turbulent noise), with an operating noise ≤30dB.

[0090] Intelligent decision control module:

[0091] New noise sensor (detection range 30-80dB) is added, which is linked with CPU load rate: when the load rate is ≥80%, the total noise should be ≤35dB;

[0092] The remote monitoring and early warning sub-module is linked with the financial transaction system through API interface, and sends a "low-impact mode" warning 10 seconds before cleaning.

[0093] Implementation steps:

[0094] (1) Pollution perception and analysis: the CPU cooling fins are detected to be covered with dust, causing the temperature to rise to 85°C, the load rate to be 95%, and the noise sensor to read 40dB; the intelligent sensor group shows that the dust concentration is >300 particles / cm 3 , and the humidity is 45%.

[0095] (2) Distributed decision generation: the decision model determines the high load state, and preferentially starts liquid-electric cleaning + silent mode electrostatic adsorption + low wind speed dust collection, and disables the high-frequency vibration of the mechanical arm.

[0096] (3) Compound cleaning execution: the liquid cooling circuit applies a 10V pulse voltage to drive charged nanoparticles to quickly peel off the accumulated dust; the electrostatic generator operates in "silent mode" (5kV) to assist in adsorbing dust without generating high-frequency noise; the mini dust collection device operates at low wind speed (1m / s) to stably collect accumulated dust through a hidden air duct; the mechanical arm only rotates at low speed (5rpm) at the gap to clean, with the total noise controlled at 32dB.

[0097] (4) Resource closed-loop management: use the transaction trough period (2-4am, load rate <20%) to start the thermoelectric power generation + dust recovery, to reduce the impact on the main system energy consumption.

[0098] (5) System self-evolution: record the cleaning parameters (voltage, wind speed, electrostatic strength) under high load, and optimize the "load rate-noise-cleaning intensity" three-dimensional mapping table.

[0099] Implementation effect: cleaning noise ≤ 35dB (traditional mechanical cleaning about 55dB) at high load, CPU temperature stable below 75℃; remote linkage mechanism ensures no interference to financial transactions during cleaning process, transaction delay fluctuation <1ms; system performance impact <1%, fully meeting financial-grade real-time requirements.

[0100] Embodiment 4

[0101] This embodiment provides a server case self-cleaning system and method based on composite cleaning and intelligent decision-making, which is used for large cluster edge inference system (elastic expansion version), and the specific implementation content includes:

[0102] Implementation purpose: solve the inference delay problem of cleaning strategy in a million-level server cluster, realize real-time processing and strategy generation of large-scale pollution data through edge inference elastic expansion (hot plug acceleration card) and cluster-level sensor data fusion, and adapt to cloud computing data center scenarios.

[0103] System architecture:

[0104] Intelligent decision-making control module:

[0105] 10 cleaning resource pooling cabinets are deployed in the computer room, each containing 8 hot plugable neuro-morphic acceleration cards (single card computing power 2TOPS); resource dynamic scheduling is realized through Kubernetes, and the acceleration card load rate is linked with the remote monitoring platform (sampling frequency 1Hz);

[0106] Each 100 servers in the cluster share 1 distributed learning node, which aggregates local data using a federated learning framework (privacy protection level: differential privacy ε=10) and fuses macro dust data of the intelligent sensor group.

[0107] Pollution monitoring and tracing network:

[0108] A centralized temperature and humidity sensor network is deployed at the cluster level (1 node per 50 servers), providing a unified environmental parameter benchmark for the electrostatic devices of each server.

[0109] Implementation steps:

[0110] (1) Pollution perception and analysis: 1000 servers in the cluster synchronously upload dust data (sampling frequency 0.5Hz), and the total data volume of quantum sensors reaches 10GB / h; the intelligent sensor group aggregates the average dust concentration and filter resistance data of the cluster.

[0111] (2) Distributed decision-making generation: detect sudden increase in dust data (e.g. air conditioner failure in the computer room), automatically call 3 idle acceleration cards for intensive inference; the federated learning node aggregates the global model within 10 seconds, combines with the filter resistance data, and preferentially allocates computing power to high-risk nodes.

[0112] (3) Composite cleaning execution: cleaning according to server importance: core database server priority start liquid-electricity + airflow + static electricity cooperative mode, non-core storage server delay to load low valley execution.

[0113] (4) Resource closed-loop management: the energy consumption of the accelerator card is supplied by the heat recovery system in the machine room, accounting for 30% of the total energy consumption.

[0114] (5) System self-evolution: when the load rate of the accelerator card is > 80%, automatically trigger new card hot plug expansion, reasoning delay control < 50ms.

[0115] Implementation effect: The cluster cleaning strategy generation time is shortened from 50 seconds to 8 seconds, and the reasoning delay is reduced by 84%; 10,000 servers support concurrent cleaning decision-making, resource pooling improves accelerator utilization to 70% (traditional fixed allocation mode is only 30%); The remote platform realizes the visualization of the cluster state, and the operation and maintenance efficiency is improved by 60%.

[0116] Embodiment 5

[0117] This embodiment provides a server case self-cleaning system and method based on composite cleaning and intelligent decision-making, which is used for a green data center closed-loop system (environment-friendly version), and the specific implementation content includes:

[0118] Implementation purpose: realize the full closed loop of "pollution-energy-material", realize dust resourceization (carbon-based dust → synthesis gas) and clean energy utilization (thermoelectric power generation) through degradable components (mycelium-based brush head + filter screen), reduce the carbon footprint of data centers, and adapt to carbon neutralization targets.

[0119] System architecture:

[0120] Composite cleaning execution unit:

[0121] The mechanical arm brush head adopts a mycelium-based biological composite material (instead of a traditional plastic base), and is covered with a degradable anti-static coating (chitosan-graphene composite), with a degradable rate of 90%;

[0122] The filter screen frame of the micro dust collection device adopts mycelium material, which is degraded synchronously with the brush head.

[0123] Energy cycle symbiosis module:

[0124] The thermoelectric power generation device has a power of 5W, and is matched with a microwave plasma cracking furnace (processing capacity 100g / day) and a mycelium incubator (temperature 25℃, humidity 60%);

[0125] The non-carbon-based dust collected by the dust collection device is marked "recyclable" through the remote platform, and is linked to the resource recovery system.

[0126] Implementation steps:

[0127] (1) Pollution perception and analysis: Focus on monitoring carbon-based dust (such as PCB board organic volatile condensation particles), quantum sensor identifies its proportion >40%; intelligent sensor group synchronously monitors temperature and humidity, providing parameters for mycelium culture.

[0128] (2) Distributed decision-making: Decision-making model prioritizes resource recycling process, linking cleaning and material regeneration.

[0129] (3) Composite cleaning execution: Mechanical arm collects carbon-based dust into a special storage box, piezoelectric vibration removes dust while providing energy for thermoelectric power generation device (power generation efficiency 20%); carbon-based dust absorbed by electrostatic device is collected by dust collection device and directly introduced into microwave cracking furnace.

[0130] (4) Resource closed-loop management: Carbon-based dust is cracked into synthesis gas (CO:H2=1:2) by microwave, which is used as a nanozyme carrier for 3D printed mycelium brush head; thermoelectric power generation meets 70% of the energy consumption needs of the sensor.

[0131] (5) System self-evolution: Genetic algorithm optimizes cracking parameters (power, time), increasing synthesis gas conversion rate from 60% to 85%; based on sensor data, optimize mycelium culture parameters, shorten brush head production cycle by 10%.

[0132] Implementation effect: Data center carbon emissions reduced by 35%, cleaning consumables procurement cost reduced by 60%; mycelium brush head service life reaches 80% of traditional brush head, naturally degrades within 3 months after disposal; resource recycling marking function of remote platform increases dust resource utilization rate to 90%.

[0133] Comparative Example 1

[0134] This comparative example provides a traditional passive cleaning solution.

[0135] System architecture:

[0136] Single layer of dust screen (pore size 50μm) + monthly manual compressed air blowing (pressure 0.6MPa), equipped with only one infrared dust sensor (detection lower limit 500nm), no intelligent decision-making module.

[0137] Implementation steps:

[0138] Dust sensor detects concentration >1000 particles / cm 3 Alarm, manual arrangement of shutdown cleaning;

[0139] During cleaning, only compressed air is used for omnidirectional blowing, without directional removal or component differentiation;

[0140] No resource recycling mechanism, dust is directly discharged into the machine room environment.

[0141] Implementation effect:

[0142] There are many cleaning blind spots (such as carbon accumulation in the micro-holes of the circuit board cannot be removed), and the CPU temperature rises by 15℃ after 30 days, and the heat dissipation efficiency decreases by 30%;

[0143] It needs to be shut down for 2 hours every month, the labor cost is high, and the compressed air energy consumption is 5 times that of the present application;

[0144] It cannot be cleaned under high load (to avoid the impact on business due to shutdown), and the accumulated dust causes hardware failure twice a year on average.

[0145] The comparison between the above-mentioned examples 1-5 and the comparative example 1 is summarized as follows:

[0146] In terms of cleaning efficiency, examples 1-5 achieve full-size dust removal through composite cleaning units (nanometer catalytic coating, piezoelectric vibration, liquid-electricity cooperation, etc.), with a nanometer dust (≤50nm) removal rate of more than 90% and a micro-hole carbon accumulation removal rate of more than 85%; the comparative example 1 can only remove particles larger than 500nm, and there are many cleaning blind spots, with a 30% decrease in heat dissipation efficiency after 30 days.

[0147] In terms of intelligent decision-making, examples 1-5 dynamically adjust strategies in combination with CPU temperature, load rate and other multi-dimensional data, such as example 3 automatically reducing noise to ≤35dB when the load rate is ≥80%, and example 4 shortening the cluster decision-making delay from 50 seconds to 8 seconds through edge elastic expansion; the comparative example 1 relies on a single dust concentration threshold without load linkage, which cannot be cleaned under high load, resulting in frequent hardware failure.

[0148] In terms of energy consumption and environmental protection, examples 1-5 achieve a 60% reduction in energy consumption and a 30%-35% reduction in carbon footprint through temperature difference power generation and resource closed loop (such as example 5 carbon-based dust conversion to synthesis gas); the energy consumption of the comparative example 1 is 5 times that of the examples, and there is no recycling mechanism, with dust directly discharged to pollute the environment.

[0149] In terms of adaptability and reliability, examples 1-5 can adapt to desert high dust (example 2 reduces dust accumulation by 70%) and financial high load scenarios, with a failure rate of almost 0; the comparative example 1 has an average monthly failure of 1-2 times under extreme environment, with a maintenance period of only 1-3 months.

[0150] In terms of maintenance cost, examples 1-5 reduce the cost by 50%-60% through self-repairing components (brush head life extended by 3 times) and intelligent early warning (maintenance period extended to 4-6 months); the comparative example 1 needs manual intervention every month, with 2 hours of shutdown per month, and the comprehensive cost is more than 3 times that of the examples.

[0151] The above-mentioned examples can be realized by software, hardware, firmware or any combination thereof, and when realized by software, the above-mentioned examples can be realized in the form of a computer program product.

[0152] Those skilled in the art can understand that the modules and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0153] In addition, each functional module in each embodiment of the present application can be integrated in one processing module, or each module can exist physically independently, or two or more modules can be integrated in one module.

[0154] The above is merely specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any modification or replacement within the technical scope disclosed by the present application can be easily thought by those skilled in the art, and should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0155] Finally: the above is only the preferred embodiment of the present application, and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A server chassis self-cleaning system based on composite cleaning and intelligent decision-making, characterized in that, Comprising: The composite cleaning execution unit, the inner wall of the case and the surface of the heat dissipation component are covered with a nano-catalytic coating containing metal organic framework and composite enzyme, which decomposes organic pollutants and releases antibacterial ions; the micro piezoelectric ceramic sheet is embedded in the heat dissipation fin, which has the functions of ultrasonic vibration dust removal and thermoelectric generation, and the system is powered by the temperature difference between the fin and the air; the micro cleaning mechanical arm brush head adopts a combination structure of shape memory alloy and graphene coating, which can flexibly adapt to the gap and self-repair wear; The surfaces of key components such as mainboards, memory modules and hard drives in the case are coated with an anti-static coating, which cooperates with an adjustable static electricity generating device to actively attract floating dust particles; a micro dust collection device is integrated, which is matched with a hidden three-dimensional air duct, the air duct inlet is distributed in the areas prone to dust accumulation such as CPU radiator and power module, and the outlet is connected with replaceable filter screen; the micro cleaning mechanical arm, dust collection device and static electricity generating device adopt hot plug modular design, supporting quick replacement without disassembling the server main body; The intelligent decision control module dynamically optimizes the cleaning strategy by fusing pollution data from multiple cases through a distributed learning network while protecting data privacy; an event-driven pulse neural network chip is used to simulate the correlation between pollution, heat dissipation and energy consumption in real time based on CPU temperature and load rate, to predict dust accumulation risk and optimize cleaning solutions autonomously; the digital twin engine couples nanometer physical models with macro system simulation to dynamically plan cleaning topology; The intelligent decision algorithm integrates multi-dimensional data such as server load, dust concentration, environmental temperature and humidity to dynamically match the cleaning mode: intermittent weak suction cleaning mode is started when the load is low and the concentration is low; switch to strong suction continuous cleaning mode when the load is high and the concentration is high; reduce the strength of the static electricity generating device when the humidity is too high; add a remote monitoring and early warning sub-module, which uploads data such as cleaning status, filter screen blockage level and equipment failure to the cloud management platform through Internet of Things technology, supports mobile APP / webpage viewing, and pushes warning notifications and solutions when faults occur; The pollution monitoring and tracing network deploys diamond color heart quantum sensors at key positions to realize dust positioning and composition identification with a particle size of ≤50nm; microfluidic chip sensors analyze the source of dust, and the sodium alginate-based hydrogel protective layer at the air inlet of the case can be controlled to secrete to intercept specific pollutants; Add intelligent sensor groups, including: dust concentration sensor, temperature and humidity sensor, pressure sensor; Multi-field collaborative fluid architecture, liquid cooling circuit integrates electric field modulation microfluidic channel, removes stubborn dust through electrophoresis of suspended charged nanoparticles; the phase change adaptive topology component is driven by cholesteric liquid crystal to reconfigure the micro channel, which switches between negative pressure suction and flow guiding obstacle removal modes according to the change of dust thermal resistance; ion migration auxiliary flow channel cooperates with airflow laminar channel to directionally remove conductive dust and suppress secondary pollution; the airflow laminar channel is linked with the hidden air duct and the micro dust collection device to form a closed-loop air circulation path, ensuring that the dust collection device can efficiently collect the dust removed by piezoelectric vibration and suppress secondary diffusion.

2. The server chassis self-cleaning system based on composite cleaning and intelligent decision of claim 1, wherein, The composite cleaning execution unit includes a self-repairing cleaning component. When the brush head of the mechanical arm is worn out, the shape memory alloy is restored to its original shape by being powered and heated, the graphene coating fills in the wear gap and maintains the electrostatic adsorption performance; the vibration sensor monitors the wear frequency of the brush head, and automatically completes calibration and predicts the replacement cycle when the server load rate is less than or equal to 20%; the hot plug modular design supports quick replacement without disassembling the server body when the brush head reaches the replacement cycle, and the replacement time is controlled within a few minutes.

3. The server chassis self-cleaning system based on composite cleaning and intelligent decision of claim 1, wherein, The composite cleaning execution unit includes an energy cycle symbiosis module. The thermoelectric power generation device supplies power to the cleaning system. The collected carbon-containing dust is cracked into synthesis gas by microwave plasma, which is used to prepare cleaning consumables or nanoenzyme carriers, forming a pollution-energy-material closed loop. The hot plug modular design supports quick replacement without disassembling the server body when the brush head reaches the replacement cycle, and the replacement time is controlled within a few minutes.

4. The server chassis self-cleaning system based on composite cleaning and intelligent decision of claim 1, wherein, The intelligent decision control module includes a pollution tracing response mechanism. After locating the dust source through data analysis, the filter screen replacement warning is sent and the temporary pre-filter layer of the air inlet is started by linking the machine room management system. When the PCB board oxidation particles are detected, the local dry cleaning and moisture-proof control are triggered synchronously, and the airflow cleaning module is started preferentially when the CPU temperature is greater than or equal to 80°C. Combined with the pressure sensor data, the degree of filter screen blockage is judged. When the blockage exceeds the threshold, the "filter screen blockage: suggest replacing within 12 hours" warning notice is pushed through the cloud management platform, and the temporary bypass air duct is opened to maintain basic heat dissipation.

5. The server chassis self-cleaning system based on composite cleaning and intelligent decision of claim 1, wherein, The intelligent decision control module includes an edge inference elastic expansion component. The cleaning resource pooled cabinet is deployed with hot-pluggable neuromorphic accelerator cards. In high-pollution scenarios, the real-time inference capability is dynamically enhanced to avoid single-node computing power bottlenecks. The remote monitoring and early warning sub-module uploads the running state and computing power load data of the edge inference node to the cloud management platform, supporting remote viewing and scheduling of accelerator card resources by operation and maintenance personnel.

6. The server chassis self-cleaning system based on composite cleaning and intelligent decision of claim 1, wherein, The pollution monitoring and tracing network comprises a quantum sensing-electromagnetic resonance composite removal unit, after positioning the nano-sized dust, the near-field electromagnetic resonance antenna array applies resonance energy wave to make the particles separate from the surface, and the magnetic control nano collector synchronously captures the charged dust; for the carbon deposition in the micro-holes of the circuit board, a microwave emission source optimized by finite difference in time domain is adopted to output 10 4 -10 5 W / m 2 Thermoacoustic pulses are used to remove the carbonized layer; the dust concentration sensor in the intelligent sensor group monitors the change of the macro dust concentration in real time, and the data of the quantum sensor are fused to realize the full-scale dust state perception of the nano level and the macro level.

7. The server chassis self-cleaning system based on composite cleaning and intelligent decision of claim 1, wherein, The multi-field collaborative fluid architecture includes a ternary composite flow channel of liquid-electricity-airflow-ion flow. The main liquid cooling circuit carries away the core heat, the micro-pressure airflow laminar flow channel maintains positive pressure filtration, and the ion migration channel drives salt accumulation ions to migrate to the collector under the electric field gradient. The dielectric wetting valve dynamically switches the flow channel, closes the ion channel for energy saving in low demand, and activates the collaborative path in high-risk situations. The hidden air duct seamlessly connects with the airflow laminar flow channel, and the fan power of the miniature dust collection device is dynamically adjusted according to the airflow laminar flow intensity, ensuring the balance between dust collection efficiency and airflow energy consumption.

8. A method for cleaning a server cabinet based on composite cleaning and intelligent decision-making, characterized in that, The method comprises the following steps: (1) Real-time pollution sensing and component analysis. The diamond color center quantum sensor and microfluidic chip are used to identify dust particle size, composition and source. The quantum sensing accuracy is ≤50 nm. Image recognition is used to classify dust types such as fiber, metal and salt mist. Intelligent sensor groups synchronously collect dust concentration, environmental temperature and humidity, and air duct pressure data, which are uploaded to the intelligent decision control module; (2) Distributed transfer learning decision. The local data of multiple cabinets are aggregated and trained to form a global model after differential privacy protection. The model is adapted to new pollution scenarios. Based on graph neural network and hierarchical reinforcement learning, small sample predictor generates hierarchical cleaning path combining CPU temperature and load rate; Fusion server load, dust concentration, temperature and humidity data, match the corresponding cleaning mode, intermittent weak suction, strong suction continuous, adjust the strength of static electricity; (3) Composite cleaning strategy execution, according to the cleaning strength-heat dissipation efficiency-energy consumption surface simulated by digital twin, the optimal parameters are determined by reinforcement learning; Liquid electricity, piezoelectric and airflow module switch operation mode according to dust type; The output strength of electrostatic generator is adjusted according to temperature and humidity data, and the hidden air duct is started with micro dust collector to realize the closed loop of adsorption, stripping and collection. (4) Resource closed loop management, waste heat is used to power the catalytic decomposition module by thermoelectric generator, and electrolytic deposition salt crystallization is converted into recyclable solution; Carbon-based dust is converted into synthesis gas for consumable production, and mycelium-based material is used to make cleaning components to reduce carbon emissions; When the filter screen is saturated, replace it through remote early warning; Modular components are hot-plugged and maintained according to demand; (5) System self-evolution fault tolerance, event-driven pulse neural network chip updates pollution-strategy mapping library according to cleaning feedback, iteratively optimizes parameters through genetic algorithm, and searches for alternative solutions and reverts to strategy backup tree when hardware fails, and completes parameter iteration within 24-48 hours after cleaning. Remote monitoring sub-module records cleaning effect and equipment state data to provide basis for strategy iteration.

9. The method of claim 8, wherein the server cabinet cleaning method based on composite cleaning and intelligent decision-making is characterized by, The pollution real-time sensing and component analysis step, after the quantum sensor locates the nano dust, the near-field electromagnetic resonance antenna array applies resonance energy wave to make the particles separate from the surface, and the magnetic control nano collector synchronously captures; for the carbon deposition in the micro holes of the circuit board, the output 10 4 -10 5 W / m 2 Microwave thermoacoustic pulse removes the carbonized layer; the dust concentration sensor monitors the macro dust concentration in real time, and triggers the cleaning strategy execution step when the concentration exceeds the threshold value.

10. The method of claim 8, wherein the server cabinet cleaning method based on composite cleaning and intelligent decision-making is characterized by, In the composite cleaning strategy execution step, the liquid electricity module applies pulse voltage when the cooling liquid flows through the dust accumulation fin to drive the charged nanoparticles to strip the dirt; Cholesteric liquid crystal phase transition triggers microchannel expansion to form negative pressure cavity to suck dead angle dust, or shrink into guide groove to guide airflow to remove obstacles, and automatically reduce mechanical component running noise when server load rate ≥80%; The electrostatic generator reduces the strength when the humidity is too high, the micro dust collector switches the running power according to the cleaning mode, and the hidden air duct ensures air circulation covering the whole area of the case.

Images