Intelligent flow dividing control method for air conditioner flow divider based on flow data dynamic adjustment

Abstract

The invention provides an intelligent flow dividing control method for an air conditioner flow divider based on flow data dynamic adjustment, and relates to the technical field of local area network access configuration and heat management. Calculating a dynamic thermal load coefficient based on the flow data; generating a network flow balance index in combination with the temperature; an air duct opening degree matrix is constructed, and dynamic optimization is conducted through gradient descent; the actuating mechanism is driven to adjust the air duct; and when the index is lower than a threshold value, software defined network rerouting migration overload traffic is activated. Dynamic air supply optimization is achieved by coupling network flow and air conditioner parameters, network rerouting and refrigeration linkage control is established, the problem of local overheating caused by static strategy response lag and cross-system splitting is solved, and data center energy efficiency and equipment safety are improved.

Description

Technical Field

[0001] This invention relates to the field of local area network path configuration and thermal management technology, and in particular to an intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data. Background Art

[0002] As data centers continue to expand, the heat generated by server clusters is increasing dramatically, with cooling systems now accounting for over 40% of total energy consumption. Traditional air conditioning control methods, primarily based on static temperature thresholds or simple timing strategies, cannot detect real-time changes in network traffic, leading to low cooling efficiency. Especially in scenarios with sudden traffic surges, localized hotspots can form due to the dense heat generated by network devices, while the air conditioning system's response is delayed, posing a risk of equipment overheating and downtime. Existing technologies attempt to alleviate the problem by increasing the density of temperature sensors or increasing fan speed, but these methods fail to address the dynamic matching between network traffic and cooling demands, instead exacerbating energy waste.

[0003] Meanwhile, the disconnect between network traffic and thermal environment management is prominent. While network operation and maintenance systems can monitor traffic fluctuations, they cannot coordinate with air conditioning equipment; cooling systems, on the other hand, lack the ability to perceive network characteristics such as TCP throughput and UDP packet loss rate. This disconnect leads to two serious consequences: first, the air conditioning splitter duct opening is fixed, making it impossible to dynamically adjust the air supply ratio according to the service load; second, when local overheating occurs, it can only passively increase the overall cooling intensity, unable to coordinate cooling through traffic migration. Existing solutions, such as SDN-based traffic scheduling, can migrate loads, but they are not deeply coupled with temperature control algorithms, resulting in migration strategies being disconnected from thermodynamic states, easily triggering secondary hotspots.

[0004] Furthermore, existing duct control models have significant shortcomings. Most systems employ open-loop control or simple feedback mechanisms, failing to establish a quantitative relationship between network status and cooling efficiency. Duct opening adjustments rely on manual experience and lack multi-variable collaborative optimization capabilities. Under complex traffic patterns (such as sudden traffic surges caused by BGP routing oscillations), traditional methods struggle to quickly converge to the optimal solution, failing to ensure equipment heat dissipation safety and resulting in significant waste of cooling capacity. Summary of the Invention

[0005] To address the technical problems of existing technologies, such as "static air conditioning control being unable to respond to dynamic changes in network traffic, leading to localized overheating and a surge in energy consumption" and "the disconnect between network traffic scheduling and the refrigeration system causing low thermal management efficiency," this invention provides an intelligent split-flow control method for air conditioning splitters based on dynamic adjustment of traffic data.

[0006] The technical solution provided by this invention is as follows: The present invention provides an intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data, comprising: S1. Real-time monitoring of network traffic data and ambient temperature data in each sub-area of ​​the local area network. The network traffic data includes Transmission Control Protocol throughput, User Datagram Protocol packet loss rate, and Border Gateway Protocol routing fluctuation count. S2. Calculate the dynamic heat load coefficient based on network traffic data. This coefficient characterizes the impact of network traffic changes on the dynamic heat load of the refrigeration system. S3. Generate a network flow balance index based on ambient temperature data and dynamic heat load coefficient. This index comprehensively reflects the degree of matching between the network state and the thermodynamic environment. S4. Construct a multi-dimensional air duct opening matrix for the air conditioning splitter, where the matrix elements represent the opening ratio of each air duct to a sub-region. S5. Using the network flow balance index as the optimization objective, the gradient descent algorithm is used to dynamically adjust the values ​​of all elements of the air duct opening matrix. S6. Generate control commands based on the optimized duct opening matrix to drive the physical actuator of the air conditioning splitter to complete the duct opening adjustment. S7. When the network flow balance index is lower than the alarm threshold, activate the network traffic rerouting mechanism to migrate the traffic data of the overloaded sub-region through the software-defined network controller.

[0007] Preferably, S2 specifically includes: S201. Collect the peak network traffic and the arithmetic mean traffic within a preset time window; S202. Calculate the theoretical traffic entropy value based on Shannon's channel capacity formula: ; Where B is the physical channel bandwidth, measured in Hz. This represents peak network traffic, measured in bps. Network noise floor power, in watts (W). S203. Calculate the magnitude of the gradient of flow rate change: ; in The sampling time window length, in seconds. This is the start time of the time window; S204, Generating Dynamic Heat Load Coefficient: ; in This is the arithmetic mean of the flow rate, in bps. It is a dimensionless dynamic thermal load coefficient.

[0008] Preferably, the calculation of the gradient magnitude of the flow rate change in step S203 further includes: S2031. Construct a cubic spline interpolation function for discrete flow data within the time window. ; S2032. Solving the time-domain integral of the absolute value of the second derivative using numerical integration: ; in For the first Each sampling time point Given the total number of sampling points, the interpolation function satisfies the natural boundary conditions. .

[0009] Preferably, the calculation of the network flow balancing index in S3 specifically includes: S301. Calculate the temperature variance between sub-regions: ; in The number of sub-regions. For the first Sub-region temperature, in °C The average temperature is expressed in °C. S302, Generation Network Flow Equilibrium Index: ; in For temperature vectors, The preset optimal temperature is expressed in °C. The maximum allowable temperature difference for the system, expressed in °C. It is a vector of all 1s. This is the Euclidean norm, also known as the L2 norm. The temperature variance between sub-regions. For dynamic thermal load coefficient, It is a dimensionless network flow equilibrium index.

[0010] Preferably, the temperature variance calculation in S301 employs a dynamic sliding window mechanism, specifically including: The window size is determined by the following formula: ; in The basic sampling period is measured in seconds (s). This is the lower limit threshold for the thermal load factor; To adapt to the window size, the unit is the number of sampling points.

[0011] Preferably, the initialization of the duct opening matrix in S4 satisfies: ; in Let be the initial opening value of duct i for sub-region j. Let be the physical area of ​​the j-th subregion, in units of . m represents the total number of air ducts, and n represents the total number of sub-regions.

[0012] Preferably, the gradient descent algorithm of S5 employs a proportional-integral-derivative control mechanism, specifically including: Deviation input Defined as: ; in The integration time constant is set to the preset target threshold. Adapts to ambient temperature: ; and This is the calibration coefficient.

[0013] Preferably, the network traffic rerouting mechanism of S7 specifically includes: S701, Identify and satisfy The sub-regions are used as migration sources; S702, Select the option that satisfies The sub-region is used as the migration target; S703, Modify the switch forwarding table entries using the Open Flow Protocol to migrate the traffic as follows: ; in To migrate the real-time traffic load of the source sub-region, in bps. This is the real-time traffic load of the target sub-region to be migrated, in bps. This is the preset traffic balancing threshold, in bps. The actual migration traffic value is expressed in bps. The above formula ensures that the traffic in the source area is not lower than the threshold and the traffic in the target area does not exceed the threshold after migration.

[0014] Preferably, the acquisition of ambient temperature data in step S1 specifically includes: S101. Deploy at least three temperature sensors in each sub-region to form a redundant array; S102. Use weighted median filtering to process the raw temperature data: ; in The measured value of sensor s, Weights for sensor accuracy; S103. Perform time synchronization compensation on the filtered data, compensation amount... , This is the synchronization coefficient.

[0015] Preferably, the cubic spline interpolation function further includes: The cubic spline interpolation function at the nodes The continuity condition is satisfied at: ; in This represents the p-th derivative of the function, and this constraint ensures the smoothness of the gradient calculation for flow rate changes.

[0016] The beneficial effects of the technical solution provided by this invention include at least the following: (1) In this invention, the network traffic characteristics and air conditioning cooling parameters are linked in real time for the first time through the coupling model of dynamic heat load coefficient and network flow balance index, so that the air duct opening can be dynamically optimized according to the network status such as transmission control protocol throughput and user data packet loss rate, thereby fundamentally eliminating local hot spots and significantly reducing cooling energy consumption. (2) In this invention, the gradient descent algorithm is used to collaboratively optimize the multi-dimensional air duct opening matrix, combined with the adaptive mechanism of proportional integral derivative control, to effectively solve the problem of response lag in traditional control strategies. In sudden scenarios such as border gateway protocol routing fluctuations, the precise air supply adjustment can be completed quickly, greatly reducing the risk of equipment overheating. (3) In this invention, the software-defined network rerouting triggered by the network flow balance index is linked with the air conditioning control across systems to build a closed-loop control of network cooling integration: when the air duct adjustment reaches the limit, the flow of the overloaded area is automatically migrated to the low temperature area, breaking through the single cooling bottleneck and improving the overall energy efficiency of the data center. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating the intelligent flow control method for an air conditioner splitter based on dynamic adjustment of flow data provided in an embodiment of the present invention; Figure 2 A schematic diagram illustrating the dynamic heat load coefficient calculation process of the intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data, provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the three-level degradation strategy execution flow of the intelligent diversion control method for air conditioning diverters based on dynamic adjustment of flow data provided in an embodiment of the present invention. Detailed Implementation

[0019] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0020] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0021] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0022] In embodiments of the present invention, sometimes the subscript is as follows: It may be mistakenly written as a non-subscript form such as W1. When the distinction is not emphasized, the meaning they express is the same.

[0023] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0024] Reference manual attached Figure 1 The diagram shows a flowchart of the intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data provided in an embodiment of the present invention.

[0025] This invention provides an intelligent flow control method for air conditioner splitters based on dynamic adjustment of flow data. The processing flow may include the following steps: S1. Real-time monitoring of network traffic data and ambient temperature data in each sub-area of ​​the local area network. The network traffic data includes Transmission Control Protocol throughput, User Datagram Protocol packet loss rate, and Border Gateway Protocol routing fluctuation count.

[0026] Specifically, the temperature sensor uses a PT1000 platinum resistance temperature sensor with an accuracy of ±0.1℃, transmitting data via Modbus RTU protocol at a sampling frequency of 1Hz. The three sensors are arranged in an equilateral triangle within the sub-area, located at the front, middle, and rear doors of the cabinet, respectively, 1.2 meters above the ground. The weights of the weighted median filter are set according to the sensor calibration certificate: 0.4 for sensors with an accuracy of ±0.1℃, and 0.2 for sensors with an accuracy of ±0.2℃. Time synchronization compensation coefficient... Compensation amount The flow rate change rate dF / dt is obtained by calculating the forward difference method. For example, when the flow rate change rate is 10 Gbps / s, the time compensation value is 1 millisecond.

[0027] In one possible implementation, the acquisition of ambient temperature data in S1 specifically includes: S101. Deploy at least three temperature sensors in each sub-region to form a redundant array; S102. Use weighted median filtering to process the raw temperature data: ; in The measured value of sensor s, Weights for sensor accuracy; S103. Perform time synchronization compensation on the filtered data, compensation amount... , This is the synchronization coefficient.

[0028] It should be noted that the high-precision temperature data obtained through the aforementioned environmental temperature data acquisition method will, together with real-time network traffic data, constitute the basic input of the system. This synchronously acquired data will be directly used in the calculation of the dynamic heat load coefficient, a key indicator for measuring the impact of network traffic on the refrigeration system's heat load, providing a quantitative basis for subsequent intelligent control decisions.

[0029] S2. Calculate the dynamic heat load coefficient based on network traffic data. This coefficient characterizes the impact of network traffic changes on the dynamic heat load of the refrigeration system.

[0030] In one possible implementation, such as Figure 2 As shown, S2 specifically includes: S201. Collect the peak network traffic and the arithmetic mean traffic within a preset time window; S202. Calculate the theoretical traffic entropy value based on Shannon's channel capacity formula: ; Where B is the physical channel bandwidth, measured in Hz. This represents peak network traffic, measured in bps. Network noise floor power, in watts (W). S203. Calculate the magnitude of the gradient of flow rate change: ; in The sampling time window length, in seconds. This is the start time of the time window; S204, Generating Dynamic Heat Load Coefficient: ; in This is the arithmetic mean of the flow rate, in bps. It is a dimensionless dynamic thermal load coefficient.

[0031] It should be noted that the time window Fixed at 60 seconds. Peak traffic. The 99.9 percentile statistical method is used to effectively eliminate outlier interference such as Distributed Denial of Service (DDoS) attacks. Network noise floor power. During actual testing at off-peak times, the typical value for a 10 Gigabit Ethernet port was -90dBm. Theoretical traffic entropy value. In the calculation, the channel bandwidth B is taken as the physical negotiated rate of the switch port (e.g., 10Gbps / 25Gbps / 100Gbps). The second derivative of the traffic is approximated using the central difference method. ,in The sampling interval is denoted as .

[0032] In one possible implementation, the calculation of the magnitude of the flow rate change gradient in S203 further includes: S2031. Construct a cubic spline interpolation function for discrete flow data within the time window. ; S2032. Solving the time-domain integral of the absolute value of the second derivative using numerical integration: ; in For the first Each sampling time point Given the total number of sampling points, the interpolation function satisfies the natural boundary conditions. .

[0033] In one possible implementation, cubic spline interpolation employs natural boundary conditions and uses the three-moment method to solve for the spline coefficients. The number of nodes, N=60, corresponds to 60 sampling points per minute. Numerical integration uses the composite Simpson formula: for each subinterval... , , where h is the length of the sub-interval. In the FPGA hardware implementation, the Q2.30 fixed-point format is used to ensure that the cumulative calculation error is <0.1%.

[0034] In one possible implementation, the cubic spline interpolation function further includes: The cubic spline interpolation function at the nodes The continuity condition is satisfied at: ; in This represents the p-th derivative of the function, and this constraint ensures the smoothness of the gradient calculation for flow rate changes.

[0035] Specifically, node continuity is guaranteed by the following mathematical constraints: Function value continuity ; First derivative continuity ; Continuity of the second derivative ; Solving the three bending moment equations results in a 52×52 banded matrix (when N=60). The LU decomposition algorithm is used, with an iteration tolerance set to... Natural boundary conditions are enforced through constraints. and accomplish.

[0036] S3. Generate a network flow balance index based on ambient temperature data and dynamic heat load coefficient. This index comprehensively reflects the matching degree between the network state and the thermodynamic environment.

[0037] In one possible implementation, the calculation of the network flow balancing index in S3 specifically includes: S301. Calculate the temperature variance between sub-regions: ; in The number of sub-regions. For the first Sub-region temperature, in °C The average temperature is expressed in °C. S302, Generation Network Flow Equilibrium Index: ; in For temperature vectors, The preset optimal temperature is expressed in °C. The maximum allowable temperature difference for the system, expressed in °C. It is a vector of all 1s. This is the Euclidean norm, also known as the L2 norm. The temperature variance between sub-regions. For dynamic thermal load coefficient, It is a dimensionless network flow equilibrium index.

[0038] Understandably, data validity verification is necessary before calculating temperature variance: when If the time is considered a sensor fault, the average value of adjacent sensors is used as a substitute. L2 norm The calculation is performed using the `cblas_dnrm2` function from the OpenBLAS library. (Exponential term, denominator) Set a lower limit of 0.01 to prevent division by zero errors. Preset the optimal temperature during actual deployment. The system allows a maximum temperature difference The calculation results are normalized to the [0,1] interval using the sigmoid function.

[0039] In one possible implementation, the temperature variance calculation of S301 employs a dynamic sliding window mechanism, specifically including: The window size is determined by the following formula: ; in The basic sampling period is measured in seconds (s). This is the lower limit threshold for the thermal load factor; To adapt to the window size, the unit is the number of sampling points.

[0040] Specifically, the basic sampling period Seconds, lower limit of thermal load factor The window size W is calculated and rounded up, with a maximum value limited to 300 (corresponding to 25 minutes of data). A circular buffer with a depth of 300 is used to store temperature data. Variance calculation is triggered every W sampling periods. Pause window updates until The recovery threshold is above 1.

[0041] Understandably, the network flow balance index As a core parameter reflecting the matching degree between network state and thermodynamic environment, it provides a decision-making basis for the precise control of air conditioning system. Based on the control requirements generated by this index, the system will initiate the process of constructing the duct opening matrix, and transform the control requirements into executable control parameters through mathematical modeling, laying the foundation for subsequent gradient optimization algorithms.

[0042] S4. Construct a multi-dimensional air duct opening matrix for the air conditioning splitter, where the matrix elements represent the opening ratio of each air duct to a sub-region.

[0043] In one possible implementation, the initialization of the duct opening matrix in S4 satisfies: ; in Let be the initial opening value of duct i for sub-region j. Let be the physical area of ​​the j-th subregion, in units of . m represents the total number of air ducts, and n represents the total number of sub-regions.

[0044] In one possible implementation, the total number of air ducts m=8 corresponds to the number of air conditioning terminals, and the number of sub-zones n=48 corresponds to the number of standard server racks. Physical area Calculated based on the floor area of ​​the server rack ( Initial opening value The normalization constraints are written to the PLC's holding register and calculated in real time. It also checks for deviations, and when the deviation is greater than 1%, it triggers an audible and visual alarm and automatically renormalizes.

[0045] Specifically, the renormalization operation is achieved through the following steps: Calculate the current opening degree of each duct i: ; Perform scaling on each matrix element: ; The normalized matrix is ​​written to the PLC holding register via Modbus TCP protocol; Verify the scaling error: If This will trigger a level 2 alarm; The above process is completed within 200ms, ensuring the real-time requirements of the control system.

[0046] Specifically, after initializing and verifying the air duct opening matrix, the system will enter the dynamic optimization phase. This phase uses the network flow balance index as the core optimization objective, and intelligent algorithms adjust the air duct opening parameters in real time to achieve the optimal match between the cooling capacity of the air conditioning system and the heat dissipation characteristics of the network equipment, thus realizing precise control of the data center's thermal environment.

[0047] S5. Using the network flow balance index as the optimization objective, the gradient descent algorithm is used to dynamically adjust the values ​​of all elements of the air duct opening matrix.

[0048] In one possible implementation, the gradient descent algorithm of S5 employs a proportional-integral-derivative control mechanism, specifically including: Deviation input Defined as: ; in The integration time constant is set to the preset target threshold. Adapts to ambient temperature: ; and This is the calibration coefficient.

[0049] It should be noted that the target threshold Set to 0.85. Calibration coefficient. Integral time constant Follow Dynamic adjustment: When the temperature difference is >3℃ (Fast response mode), when the temperature difference is <1℃ (Anti-oscillation mode). Gradient descent step size. The control cycle is synchronized with the data acquisition cycle, which is 5 seconds.

[0050] It should be noted that the specific implementation process of the gradient descent algorithm is as follows: Initialize the Jacobian matrix It is a zero matrix; For each matrix element Perform disturbance analysis: a. Positive disturbance: ; b. Negative disturbance: ; c. Calculate partial derivatives: ; Update the opening matrix: ; Constraint handling: If Then set it to 0, if Then set it to 1.

[0051] Learning rate The perturbation analysis is performed once every 5 iterations to reduce the computational load.

[0052] S6. Generate control commands based on the optimized duct opening matrix to drive the physical actuator of the air conditioning splitter to complete the duct opening adjustment.

[0053] In one possible implementation, the optimized duct opening matrix is ​​distributed via the OPC UA protocol. The actuator uses an electric damper controlled by a 0-10V analog signal, with a linear voltage-opening correspondence (10V = 100% opening), an opening resolution of 0.1%, and a mechanical response time of <2 seconds. A full calibration process is performed every 15 minutes: first resetting to the initial opening. Then, adjust the value step by step in increments of 0.5% to reach the current optimal solution, and finally record the position sensor feedback value to compensate for mechanical hysteresis.

[0054] It should be noted that while the air conditioning system is adjusting the air ducts, the system continuously monitors the changing trend of the network flow balance index. When this index remains below the safety threshold, it indicates that single air conditioning adjustment can no longer meet the thermal management requirements. At this point, the network-level collaborative control mechanism will be activated, and the spatial transfer of heat load will be achieved through flow rerouting, forming a closed-loop linkage control between the air conditioning system and the network system.

[0055] S7. When the network flow balance index is lower than the alarm threshold, activate the network traffic rerouting mechanism to migrate the traffic data of the overloaded sub-region through the software-defined network controller.

[0056] In one possible implementation, the network traffic rerouting mechanism of S7 specifically includes: S701, Identify and satisfy The sub-regions are used as migration sources; S702, Select the option that satisfies The sub-region is used as the migration target; S703, Modify the switch forwarding table entries using the Open Flow Protocol to migrate the traffic as follows: ; in To migrate the real-time traffic load of the source sub-region, in bps. This is the real-time traffic load of the target sub-region to be migrated, in bps. This is the preset traffic balancing threshold, in bps. The actual migration traffic value is expressed in bps. The above formula ensures that the traffic in the source area is not lower than the threshold and the traffic in the target area does not exceed the threshold after migration.

[0057] Specifically, alarm threshold Temperature deviation threshold The migration source is selected using a min-heap data structure and processed first. The lowest sub-region. Traffic balancing threshold. Use 70% of the switch port speed (7Gbps for a 10 Gigabit port). The migration will be performed in three steps: 1) Modify the forwarding table using OpenFlow's Flow-Mod message; 2) Phased migration (single migration ≤ 350Mbps); 3) Post-migration verification If the increase value does not meet expectations, VMware vMotion virtual machine migration will be triggered.

[0058] Understandably, Flow-Mod messages contain the following key fields: Matching fields: Source IP address range ( / 24 subnet mask), destination port number; Action field: The output port (output_port) is set to the target area switch port; The queue identifier (queue_id) corresponds to the QoS priority; Timeout field: hard_timeout=120s (temporary migration) Batch migration uses the token bucket algorithm to control the migration rate. ; in The allowed migration rate at time t is in Mbps, where t is the time after migration starts in seconds, ensuring an initial low-speed migration (50 Mbps). With a midpoint time for rate adjustment (default set to 10 seconds), the algorithm ensures an initial migration rate of approximately 50 Mbps (at t=0), which approaches the upper limit of 350 Mbps over time, thus avoiding instantaneous congestion.

[0059] It should be noted that when software-defined networking migration fails, a three-level degradation mechanism is initiated: Level 1 reduces the CPU frequency of the source region server to 80% of its base frequency; Level 2 migrates non-critical virtual machines via vMotion; Level 3 shuts down backup nodes. Event logs contain 12 fields, including timestamps, migration traffic, and temperature fluctuation rate, and are stored in Apache Parquet format with a retention period of 30 days. Log analysis utilizes Spark Streaming for real-time anomaly detection.

[0060] In one possible implementation, such as Figure 3 The strategy execution logic shown has the following trigger conditions for each level of degradation strategy: Level 1 trigger condition: When And the number of SDN migration failures The CPU frequency is set via the IPMI raw 0x080x05 instruction: ; in This refers to the adjusted CPU frequency, in GHz. This refers to the CPU's base frequency (e.g., 2.4GHz). This refers to the CPU's maximum turbo frequency (e.g., 3.8GHz). The current network flow balancing index k is the frequency reduction factor (default value 0.2).

[0061] Level 2 trigger condition: 10 minutes after Level 1 execution Still <0.65: Virtual machine migration selection criteria: If memory usage < 40% and CPU utilization < 50%, and the service type is not necessarily a database... Then migration priority = high Migrate the selected virtual machine via the vMotion API, with a migration bandwidth limit of 1Gbps.

[0062] Level 3 trigger condition: After Level 2 is executed Still <0.6: Shutdown order: Backup nodes → Test environment → Non-core business containers (Kubernetes Pods) Commands to stop: `systemctl stop service_name` or `kubectl scale deployment --replicas=0` Understandably, the aforementioned three-tiered degradation strategy constitutes the system's security assurance framework. When the main control strategy fails, this framework ensures the system's security boundary by gradually reducing service load, while simultaneously buying valuable time for fault diagnosis and system recovery. This layered and progressive fault-tolerant mechanism, together with the core control algorithm, forms a complete technical solution, ensuring the reliability and robustness of data center thermal environment management.

[0063] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: (1) In this invention, the network traffic characteristics and air conditioning cooling parameters are linked in real time for the first time through the coupling model of dynamic heat load coefficient and network flow balance index, so that the air duct opening can be dynamically optimized according to the network status such as transmission control protocol throughput and user data packet loss rate, thereby fundamentally eliminating local hot spots and significantly reducing cooling energy consumption. (2) In this invention, the gradient descent algorithm is used to collaboratively optimize the multi-dimensional air duct opening matrix, combined with the adaptive mechanism of proportional integral derivative control, to effectively solve the problem of response lag in traditional control strategies. In sudden scenarios such as border gateway protocol routing fluctuations, the precise air supply adjustment can be completed quickly, greatly reducing the risk of equipment overheating. (3) In this invention, the software-defined network rerouting triggered by the network flow balance index is linked with the air conditioning control across systems to build a closed-loop control of network cooling integration: when the air duct adjustment reaches the limit, the flow of the overloaded area is automatically migrated to the low temperature area, breaking through the single cooling bottleneck and improving the overall energy efficiency of the data center.

[0064] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0065] Where there is no conflict, the embodiments of the present invention and the features thereof can be combined with each other to obtain new embodiments.

[0066] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for intelligent flow control of an air conditioning splitter based on dynamic adjustment of flow data, characterized in that, include: S1. Real-time monitoring of network traffic data and ambient temperature data in each sub-area of ​​the local area network. The network traffic data includes Transmission Control Protocol throughput, User Datagram Protocol packet loss rate, and Border Gateway Protocol routing fluctuation count. S2. Calculate the dynamic heat load coefficient based on network traffic data. This coefficient characterizes the impact of network traffic changes on the dynamic heat load of the refrigeration system. S3. Generate a network flow balance index based on ambient temperature data and dynamic heat load coefficient. This index comprehensively reflects the degree of matching between the network state and the thermodynamic environment. S4. Construct a multi-dimensional air duct opening matrix for the air conditioning splitter, where the matrix elements represent the opening ratio of each air duct to a sub-region. S5. Using the network flow balance index as the optimization objective, the gradient descent algorithm is used to dynamically adjust the values ​​of all elements of the air duct opening matrix. S6. Generate control commands based on the optimized duct opening matrix to drive the physical actuator of the air conditioning splitter to complete the duct opening adjustment. S7. When the network flow balance index is lower than the alarm threshold, activate the network traffic rerouting mechanism to migrate the traffic data of the overloaded sub-region through the software-defined network controller.

2. The intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data according to claim 1, characterized in that, S2 specifically includes: S201. Collect the peak network traffic and the arithmetic mean traffic within a preset time window; S202. Calculate the theoretical traffic entropy value based on Shannon's channel capacity formula: ； Where B is the physical channel bandwidth, measured in Hz. This represents peak network traffic, measured in bps. Network noise floor power, in watts (W). S203. Calculate the magnitude of the gradient of flow rate change: ； in The sampling time window length, in seconds. This is the start time of the time window; S204, Generating Dynamic Heat Load Coefficient: ； in This is the arithmetic mean of the flow rate, in bps. It is a dimensionless dynamic thermal load coefficient.

3. The intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data according to claim 2, characterized in that, The calculation of the gradient magnitude of the flow rate change in S203 further includes: S2031. Construct a cubic spline interpolation function for discrete flow data within the time window. ; S2032. Solving the time-domain integral of the absolute value of the second derivative using numerical integration: ； in For the first Each sampling time point Given the total number of sampling points, the interpolation function satisfies the natural boundary conditions. .

4. The intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data according to claim 1, characterized in that, The calculation of the network flow balancing index in S3 specifically includes: S301. Calculate the temperature variance between sub-regions: ； in The number of sub-regions. For the Sub-region temperature, in °C The average temperature is expressed in °C. S302, Generation Network Flow Equilibrium Index: ； in For temperature vectors, The preset optimal temperature is expressed in °C. This represents the maximum allowable temperature difference of the system, expressed in degrees Celsius (°C). It is a vector of all 1s. This is the Euclidean norm, also known as the L2 norm. The temperature variance between sub-regions. For dynamic thermal load coefficient, It is a dimensionless network flow equilibrium index.

5. The intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data according to claim 4, characterized in that, The temperature variance calculation of S301 adopts a dynamic sliding window mechanism, specifically including: The window size is determined by the following formula: ； in The basic sampling period is measured in seconds (s). This is the lower limit threshold for the thermal load factor; To adapt to the window size, the unit is the number of sampling points.

6. The intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data according to claim 1, characterized in that, The initialization of the air duct opening matrix in S4 satisfies: ； in Let be the initial opening value of duct i for sub-region j. Let be the physical area of ​​the j-th subregion, in units of . m represents the total number of air ducts, and n represents the total number of sub-regions.

7. The intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data according to claim 1, characterized in that, The gradient descent algorithm of S5 employs a proportional-integral-derivative control mechanism, specifically including: Deviation input Defined as: ； in The integration time constant is set to the preset target threshold. Adapts to ambient temperature: ； and This is the calibration coefficient.

8. The intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data according to claim 1, characterized in that, The network traffic rerouting mechanism of S7 specifically includes: S701, Identify and satisfy The sub-regions are used as migration sources; S702, Select the option that satisfies The sub-region is used as the migration target; S703, Modify the switch forwarding table entries using the Open Flow Protocol to migrate the traffic as follows: ； in To migrate the real-time traffic load of the source sub-region, in bps. This is the real-time traffic load of the target sub-region to be migrated, in bps. This is the preset traffic balancing threshold, in bps. The actual migration traffic value is expressed in bps. The above formula ensures that the traffic in the source area is not lower than the threshold and the traffic in the target area does not exceed the threshold after migration.

9. The intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data according to claim 1, characterized in that, The collection of ambient temperature data in S1 specifically includes: S101. Deploy at least three temperature sensors in each sub-region to form a redundant array; S102. Use weighted median filtering to process the raw temperature data: ； in The measured value of sensor s, Weights for sensor accuracy; S103. Perform time synchronization compensation on the filtered data, compensation amount... , This is the synchronization coefficient.

10. The intelligent flow control method for air conditioning splitters based on dynamic adjustment of flow data according to claim 3, characterized in that, The cubic spline interpolation function further includes: The cubic spline interpolation function at the nodes The continuity condition is satisfied at: ； in This represents the p-th derivative of the function, and this constraint ensures the smoothness of the gradient calculation for flow rate changes.

Images