Cloud computing service platform based on artificial intelligence

Abstract

The embodiment of the utility model discloses a cloud computing service platform based on artificial intelligence, which comprises a physical server cluster, network connection equipment, an artificial intelligence module and a distributed storage unit, the physical server cluster comprises a plurality of servers arranged in a distributed manner, and the cloud computing service platform further comprises a distributed temperature sensing module, a cloud computing module and a cloud computing module, the plurality of servers are arranged at key heating parts of the servers in a distributed manner; the cooling execution unit is arranged close to the key heating part and comprises a plurality of groups of micro PWM (Pulse Width Modulation) speed regulation fans and / or micro liquid cooling devices; the PWM dynamic closed-loop speed regulation circuit comprises a main control module, a PWM speed regulation module and a state monitoring module connected with the cooling execution unit; wherein the distributed temperature sensing module and / or the cooling execution unit are / is electrically connected with the main control module through an interrupt controller. According to the invention, efficient and real-time temperature control can be realized, energy is saved, and the stability of the system is effectively ensured.

Description

Technical Field

[0001] This application relates to the field of cloud computing services, and more particularly to a cloud computing service platform based on artificial intelligence. Background Technology

[0002] A cloud computing service platform mainly consists of physical server clusters, network connectivity devices, distributed storage units, and cooling systems. With the rapid development of cloud computing and big data, the physical architecture of traditional cloud computing service platforms is increasingly unable to adapt to dynamically changing business load demands, resulting in problems such as low resource utilization and energy waste.

[0003] By introducing artificial intelligence technology to intelligently layout and dynamically adjust the physical architecture of cloud computing service platforms, and creating cloud computing service platforms optimized by artificial intelligence, resource utilization can be improved, energy consumption and operation and maintenance costs can be reduced, and platform operating efficiency and service quality can be enhanced.

[0004] However, the introduction of artificial intelligence technology has led to frequent changes in server load on cloud computing platforms. Existing cloud computing platforms generally employ fixed cooling strategies for server temperature control, such as controlling fan speed or cooling system startup and shutdown based on preset thresholds. While these cooling solutions are simple to implement, they struggle to achieve efficient, real-time temperature control in the face of frequent load changes in AI-powered servers, easily resulting in over- or under-cooling, leading to energy waste or decreased system stability. Summary of the Invention

[0005] This application provides an artificial intelligence-based cloud computing service platform to achieve efficient and real-time temperature control, save energy, and effectively ensure system stability.

[0006] To address the above technical problems, this application provides an artificial intelligence-based cloud computing service platform, including a physical server cluster, network connection devices, an artificial intelligence module, and distributed storage units. The physical server cluster includes multiple distributed servers, and the cloud computing service platform further includes:

[0007] A distributed temperature sensing module is distributed in each of the key heat-generating parts of the server, the key heat-generating parts including the CPU module located on the server motherboard;

[0008] The cooling unit is located near the critical heat-generating part and includes multiple sets of miniature PWM speed-controlled fans and / or miniature liquid cooling devices.

[0009] The PWM dynamic closed-loop speed control circuit includes a main control module, a PWM speed control module, and a status monitoring module connected to the cooling execution unit;

[0010] The distributed temperature sensing module and / or the cooling execution unit are electrically connected to the main control module via an interrupt controller.

[0011] In some possible implementations, the cooling execution unit is connected to the PWM speed control module via a PWM control interface; and / or, the miniature liquid cooling device includes a miniature liquid cooling pump equipped with a flow control valve; and / or, the miniature PWM speed control fan is implemented using a brushless DC fan.

[0012] The micro liquid-cooled pump is implemented using a micro magnetic drive pump, and the flow control valve is implemented using a proportional solenoid valve. The magnetic drive pump and the flow control valve are integrated into an Asetek water-cooled plate structure.

[0013] In some possible implementations, the flow control valve is also provided with a CAN bus interface.

[0014] In some possible implementations, the status monitoring module includes a fan status feedback module connected to the miniature PWM speed-regulating fan and / or a liquid cooling status feedback module connected to the miniature liquid cooling device; and / or

[0015] The key heat-generating components also include a GPU and / or a power module; and / or

[0016] An embedded AI inference acceleration module, which is electrically connected to the PWM dynamic closed-loop speed regulation circuit, is also provided on the main control module.

[0017] In some possible implementations, the fan status feedback module is implemented using a Hall sensor or a current detection module; and / or, the liquid cooling status feedback module is implemented using a pressure sensor and a temperature sensor; and / or, the status monitoring module is connected to the main control module through a comparator circuit.

[0018] In some possible implementations, the main control module is implemented using an FPGA or ASIC chip; and / or, the main control module periodically triggers the distributed temperature sensing module and / or the status monitoring module through a hardware state machine.

[0019] In some possible implementations, the distributed temperature sensing module employs high-precision digital temperature sensors, with a total of 8 high-precision digital temperature sensors, 4 miniature PWM speed-regulating fans, and 2 miniature liquid cooling devices.

[0020] In some possible implementations, a BMC communication interface is also provided, and / or the distributed temperature sensing module communicates via I... 2 The C bus or SPI interface is used to communicate with the main control module.

[0021] In some possible implementations, the PWM closed-loop dynamic speed control circuit adopts the AXI4-Lite bus architecture.

[0022] The beneficial effects of this application are:

[0023] This application embodiment sets up a distributed temperature sensing module in the key heat-generating parts of the server to collect local temperature data in real time, and realizes a closed-loop control path from temperature sensing to cooling regulation through a PWM dynamic closed-loop speed regulation circuit 7. This ensures low system response latency and does not rely on host CPU resources, thereby significantly improving the real-time performance and energy efficiency of temperature control, reducing the overall power consumption and deployment cost of the system. It is especially suitable for the hardware-level temperature management needs of AI-based cloud computing service platforms with large-scale and frequently changing loads. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the structure of an artificial intelligence-based cloud computing service platform according to an embodiment of this application.

[0026] Explanation of reference numerals in the attached figures:

[0027] 1-Physical server cluster, 11-Servers;

[0028] 2- Network connection devices;

[0029] 3-Artificial Intelligence Module;

[0030] 4-Distributed storage unit;

[0031] 5-Distributed temperature sensing module;

[0032] 6-Cooling actuator, 61-Miniature PWM speed-regulating fan, 62-Miniature liquid cooling device, 63-PWM control interface, 64-CAN bus interface, 621-Miniature liquid cooling pump, 622-Flow control valve;

[0033] 7-PWM dynamic closed-loop speed control circuit, 71-main control module, 72-PWM speed control module, 73-status monitoring module, 74-comparator circuit, 75-BMC communication interface, 710-interrupt controller, 711-AI inference acceleration module, 731-fan status feedback module, 732-liquid cooling status feedback module. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be described in detail below with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0035] It should be noted that: throughout the accompanying drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions; in the description of this application, the terms "center," "longitudinal," "lateral," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the scope of protection of this application; in the description of this application, "first," "second," etc., are only used to distinguish each other, and do not indicate their degree of importance or order, etc.

[0036] In the description of this application, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linkage" should be interpreted broadly. For example, they can refer to fixed connections, movable connections, or detachable connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; they can refer to the internal communication between two components, etc. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0037] Please refer to Figure 1 This application provides an artificial intelligence-based cloud computing service platform, including a physical server cluster 1, a network connection device 2, an artificial intelligence module 3, and a distributed storage unit 4. The physical server cluster 1 includes multiple distributed servers 11. The cloud computing service platform also includes:

[0038] The distributed temperature sensing module 5 is distributed in each of the key heat-generating parts of the server 11 to collect local temperature data of the key heat-generating parts in real time. The key heat-generating parts include a CPU (Central Processing Unit) module located on the motherboard of the server 11.

[0039] The cooling execution unit 6 is located near the key heat-generating part and includes multiple sets of miniature PWM (Pulse Width Modulation) speed-regulating fans 61 and / or miniature liquid cooling devices 62.

[0040] The PWM dynamic closed-loop speed control circuit 7 includes a main control module 71, a PWM speed control module 72, and a status monitoring module 73 connected to the cooling execution unit 6;

[0041] The distributed temperature sensing module 5 and / or the cooling execution unit 6 are electrically connected to the main control module 71 via the interrupt controller 710.

[0042] In some possible implementations, the cooling execution unit 6 is connected to the PWM speed control module 72 via a PWM control interface 63;

[0043] In some possible implementations, the miniature liquid cooling device 62 includes a miniature liquid cooling pump 621 equipped with a flow control valve 622, which is connected to the PWM control interface 63.

[0044] In some possible implementations, the miniature PWM speed-regulating fan 61 is implemented as a brushless DC fan, which is connected to the PWM control interface 63.

[0045] In some possible implementations, the miniature liquid-cooled pump 621 is implemented using a miniature magnetically driven pump, and the flow control valve 622 is implemented using a proportional solenoid valve. In specific implementations, the magnetically driven pump 621 and the flow control valve 622 can be integrated into an Asetek water-cooled plate structure.

[0046] In some possible implementations, the flow control valve 622 is also provided with a CAN bus interface 64.

[0047] In some possible implementations, the status monitoring module 73 includes a fan status feedback module 731 connected to the miniature PWM speed-regulating fan 61 and / or a liquid cooling status feedback module 732 connected to the miniature liquid cooling device 62.

[0048] In some possible implementations, the key heat-generating components may also include a GPU (Graphics Processing Unit), a power supply module, etc.

[0049] In some possible implementations, the main control module 71 is further provided with an embedded AI inference acceleration module 711 that is electrically connected to the PWM dynamic closed-loop speed control circuit 7.

[0050] In some possible implementations, the fan status feedback module 731 is implemented using a Hall sensor or a current detection module, and the liquid cooling status feedback module 732 is implemented using a pressure sensor and a temperature sensor.

[0051] In some possible implementations, the status monitoring module 73 is connected to the main control module 71 via a comparator circuit 74.

[0052] In some possible implementations, the main control module 71 is implemented using an FPGA (Field-Programmable Gate Array) or ASIC (Application Specific Integrated Circuit) chip.

[0053] In some possible implementations, the main control module 71 periodically triggers the distributed temperature sensing module 5 and / or the status monitoring module 73 through a hardware state machine.

[0054] In some possible implementations, the distributed temperature sensing module 5 employs high-precision digital temperature sensors, with a quantity of 8 high-precision digital temperature sensors, a quantity of 4 miniature PWM speed-regulating fans 61, and a quantity of 2 miniature liquid cooling devices 62.

[0055] In some possible implementations, a BMC (Baseboard Management Controller) communication interface 75 is also provided;

[0056] In some possible implementations, the distributed temperature sensing module 5 uses I 2 The main control module 71 is connected to the C bus or SPI (Serial Peripheral Interface) for communication.

[0057] In some possible implementations, the PWM closed-loop dynamic speed control circuit 7 adopts an AXI4-Lite bus architecture.

[0058] This application embodiment employs a distributed temperature sensing architecture, deploying multiple miniature high-precision temperature sensors in key heat-generating areas of the server motherboard (such as near the CPU, GPU, and power module). The sensors utilize I0... 2 The sensor node communicates with the main control module via C or SPI interface to collect local temperature data in real time; the sensor node integrates low-power ADC and wireless transmission function, supports multi-point synchronous sampling, and ensures high spatiotemporal resolution of temperature data; the main control unit receives temperature data in real time through hardware interrupt mechanism, avoiding reliance on operating system scheduling and improving system response speed.

[0059] The main control module in this embodiment can also be built into an embedded AI inference acceleration module based on FPGA or ASIC. This module integrates a lightweight neural network inference engine to execute a pre-trained temperature prediction model. The model input includes current temperature data, historical load information, ambient temperature and airflow status, and the output is a predicted temperature curve and cooling strategy suggestions. The module supports online model updates and dynamic parameter adjustments to adapt to different server configurations and operating environments. The inference process is completed entirely at the hardware layer, without relying on host CPU resources, thus reducing system overhead.

[0060] The cooling execution unit in this embodiment consists of multiple sets of PWM speed-regulating fans and micro liquid cooling devices. Each cooling device is equipped with an independent PWM control interface and flow rate adjustment module. The fan speed and liquid cooling flow rate are dynamically adjusted according to the cooling strategy output by the AI ​​inference acceleration module to achieve refined and localized cooling control. At the same time, a feedback mechanism is integrated to monitor the actual operating status through Hall sensors or current detection modules to ensure accurate execution of control commands. The cooling strategy supports hierarchical control and redundancy switching to ensure that the system can still maintain basic cooling functions when some hardware fails.

[0061] The PWM dynamic closed-loop speed control circuit in this embodiment constructs a closed-loop control path through a dedicated hardware logic circuit. Temperature data acquisition, AI inference, and cooling adjustment command output are all automatically completed by the hardware without software intervention. The control logic unit integrates a timer and an interrupt controller to ensure that temperature sensing and cooling adjustment operations are completed within a fixed time window, improving the system's real-time performance. The closed-loop control path supports multi-level priority scheduling to ensure that the system can still respond quickly to temperature changes under high load or abnormal operating conditions.

[0062] This application embodiment can adopt a low-power heterogeneous computing architecture. The temperature sensing, AI inference, and cooling control modules all have independent power management units, supporting dynamic voltage and frequency scaling (DVFS) and deep sleep mode. Inter-module communication adopts high-speed low-power LVDS (Low-Voltage Differential Signaling) interface or TSV (Through-Silicon Via) interconnect technology to reduce signal delay and electromagnetic interference. It also provides standardized hardware interfaces to support seamless integration with existing server motherboards, BMC, or DCIM (Data Center Infrastructure management) systems, facilitating large-scale deployment and unified management.

[0063] The following is an exemplary description of an implementation scheme for a medium-sized artificial intelligence-based cloud computing service platform provided by an embodiment of this application. The core objective of this embodiment is to minimize energy consumption and cooling costs while ensuring stable server operation through hardware-level closed-loop control.

[0064] In this embodiment, eight high-precision digital temperature sensors (model TMP102) are arranged in key areas of the server motherboard (such as near the CPU, GPU, and power module). The sensors have a sampling frequency of 1Hz, a temperature resolution of 0.0625℃, and an error of ±0.5℃. The sensors are connected via I... 2 The C-bus connects to the main control module, which is implemented using an FPGA (Field-Programmable Gate Array) chip and is responsible for data acquisition and preprocessing. To improve system response speed, temperature data is not processed by the operating system but is directly sent to the AI ​​inference acceleration module via a hardware interrupt mechanism.

[0065] In this implementation, the cooling unit consists of four PWM speed-controlled fans and two sets of miniature liquid cooling devices. The fans are Delta BFB1206M models with a speed range of 0-6000 RPM. The liquid cooling devices adopt an Asetek water-cooled plate design, supporting 0-100% flow rate adjustment. A dedicated state machine can be implemented via FPGA to receive the cooling strategy output from the AI ​​inference acceleration module. After the cooling strategy is output, the FPGA controls the duty cycle of the PWM signal to achieve dynamic adjustment of the fans and liquid cooling devices. The actual fan speed is monitored by Hall sensors and fed back to the control logic to form a closed-loop regulation.

[0066] The PWM dynamic closed-loop speed control is implemented by a hardware state machine, which periodically (once per second) triggers the temperature acquisition, AI inference, and cooling adjustment process. The system also supports priority scheduling. When any sensor temperature exceeds a set threshold (e.g., 75°C), a high-priority interrupt is immediately triggered, forcibly increasing the fan speed and liquid cooling flow until the temperature returns to a safe range.

[0067] In addition, this embodiment also provides a hardware interface with the BMC (Baseboard Management Controller), which supports reporting temperature and cooling status information via the IPMI protocol, facilitating unified monitoring and management in the data center.

[0068] This embodiment employs a low-power design, with the total power consumption of the FPGA and sensor being less than 3W. The cooling adjustment process does not require the involvement of the host CPU, significantly reducing software overhead and response latency.

[0069] In actual operation, this embodiment successfully maintained the server temperature within the range of 70±2℃, and the cooling energy consumption was reduced by about 23% compared with the traditional fan control strategy. At the same time, it has good real-time performance and stability, and is suitable for large-scale data center deployment.

[0070] Note that the above are merely preferred embodiments and the technical principles employed in this application. Those skilled in the art will understand that this application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application, the scope of which is determined by the scope of the appended claims.

Claims

1. An artificial intelligence-based cloud computing service platform comprising a physical server cluster, a network connection device, an artificial intelligence module, and a distributed storage unit, wherein the physical server cluster comprises a plurality of distributed servers. The cloud computing service platform also includes: A distributed temperature sensing module is distributed in each of the key heat-generating parts of the server, the key heat-generating parts including the CPU module located on the server motherboard; The cooling unit is located near the critical heat-generating part and includes multiple sets of miniature PWM speed-controlled fans and / or miniature liquid cooling devices. The PWM dynamic closed-loop speed control circuit includes a main control module, a PWM speed control module, and a status monitoring module connected to the cooling execution unit; The distributed temperature sensing module and / or the cooling execution unit are electrically connected to the main control module via an interrupt controller. 2.The artificial intelligence-based cloud computing service platform of claim 1, wherein: The cooling execution unit is connected to the PWM speed control module via a PWM control interface; and / or, the micro liquid cooling device includes a micro liquid cooling pump equipped with a flow control valve; and / or, the micro PWM speed control fan is implemented using a brushless DC fan.

3. The cloud computing service platform based on artificial intelligence according to claim 2, characterized in that: The micro liquid-cooled pump is implemented using a micro magnetic drive pump, and the flow control valve is implemented using a proportional solenoid valve. The magnetic drive pump and the flow control valve are integrated into an Asetek water-cooled plate structure. 4.The artificial intelligence-based cloud computing service platform according to claim 2, characterized in that: The flow control valve is also equipped with a CAN bus interface.

5. The cloud computing service platform based on artificial intelligence according to any one of claims 1-4, characterized in that: The status monitoring module includes a fan status feedback module connected to the miniature PWM speed-regulating fan and / or a liquid cooling status feedback module connected to the miniature liquid cooling device; and / or The key heat-generating components also include a GPU and / or a power module; and / or An embedded AI inference acceleration module, which is electrically connected to the PWM dynamic closed-loop speed regulation circuit, is also provided on the main control module.

6. The cloud computing service platform based on artificial intelligence according to claim 5, characterized in that: The fan status feedback module is implemented using a Hall sensor or a current detection module; and / or, the liquid cooling status feedback module is implemented using a pressure sensor and a temperature sensor; and / or, the status monitoring module is connected to the main control module through a comparator circuit.

7. The cloud computing service platform based on artificial intelligence according to claim 5, characterized in that: The main control module is implemented using an FPGA or ASIC chip; and / or, the main control module periodically triggers the distributed temperature sensing module and / or the status monitoring module through a hardware state machine.

8. The cloud computing service platform based on artificial intelligence according to any one of claims 1-4, characterized in that: The distributed temperature sensing module uses high-precision digital temperature sensors, and the number of high-precision digital temperature sensors is 8. The number of micro PWM speed-regulating fans is 4, and the number of micro liquid cooling devices is 2.

9. The cloud computing service platform based on artificial intelligence according to any one of claims 1-4, characterized in that: A BMC communication interface is further provided, and / or the distributed temperature sensing module is connected with the main control module through I 2 A C bus or an SPI interface is in communication connection with the main control module.

10. The cloud computing service platform based on artificial intelligence according to any one of claims 1-4, characterized in that: The PWM dynamic closed-loop speed control circuit adopts the AXI4-Lite bus architecture.

Images