CN-121979743-A - Immersed liquid cooling monitoring system and method for reinforced intelligent computation server

CN121979743ACN 121979743 ACN121979743 ACN 121979743ACN-121979743-A

Abstract

The application belongs to the technical field of computer cooling, and particularly discloses an immersed liquid cooling monitoring system and method of a reinforced intelligent computing server, wherein the system comprises a monitoring module, a monitoring module and a monitoring module, wherein the monitoring module is used for acquiring data to be monitored in an intelligent computing server cabinet, and the data to be monitored comprise flow and temperature of liquid inlet of the intelligent computing server cabinet, flow and temperature of liquid outlet of the intelligent computing server cabinet, liquid inlet pressure of the intelligent computing server cabinet, liquid level and temperature of each computing unit in the intelligent computing server, impurity, color condition, moisture content, pH value, oxidation-reduction potential and conductivity of immersed liquid in the intelligent computing server cabinet; and the analysis module is used for inputting the data to be monitored into a pre-trained decision tree model to obtain a processing scheme output by the decision tree model. The system can enhance the operation stability and reliability of the immersed liquid cooling monitoring system.

Inventors

ZHANG HUAI
ZHANG QI
LI MENG
HE PENGYUAN
LI CAN
XU YUAN
DONG QINGPING

Assignees

中国船舶集团有限公司第七〇九研究所

Dates

Publication Date: 20260505
Application Date: 20260130

Claims (10)

1. An immersion liquid cooling monitoring system of a reinforced intelligent computing server, comprising: The monitoring module is used for acquiring data to be monitored in the intelligent computing server cabinet, wherein the data to be monitored comprises flow and temperature of liquid inlet of the intelligent computing server cabinet, flow and temperature of liquid outlet of the intelligent computing server cabinet, liquid inlet pressure of the intelligent computing server cabinet, liquid level and temperature of each computing unit in the intelligent computing server, impurities, color conditions, moisture content, pH value, oxidation-reduction potential and conductivity of immersion liquid in the intelligent computing server cabinet; And the analysis module is used for inputting the data to be monitored into a pre-trained decision tree model to obtain a processing scheme output by the decision tree model.
2. The system of claim 1, wherein the training process of the decision tree model comprises: Taking the numerical range of the data to be monitored and a corresponding preset processing scheme as a data set, and preprocessing the data set; Training the decision tree model based on the preprocessed data set until the performance of the decision tree model meets the preset requirement.
3. The immersion liquid cooled monitoring system of a ruggedized intelligent computing server according to claim 1, wherein the system further comprises: And the liquid cooling heat dissipation cabinet is connected with the analysis module and is used for responding to the first processing scheme output by the decision tree model and cooling and dissipating heat of the intelligent computation server.
4. The immersion liquid cooled monitoring system of a ruggedized intelligent computing server according to claim 1, wherein the system further comprises: the first output module is used for periodically outputting first reminding information, and the first reminding information is used for reminding a user to observe the impurity and the color condition of the immersion liquid through observing the sight glass valve; the first receiving module is used for receiving first input information input by a user, wherein the first input information is impurities and color conditions of the immersion liquid.
5. The immersion liquid cooled monitoring system of a ruggedized intelligent computing server according to claim 1, wherein the system further comprises: The second output module is used for periodically outputting second reminding information, wherein the second reminding information is used for reminding a user to sample the immersion liquid through the sampling valve and carrying out pH value detection, oxidation-reduction potential detection and conductivity detection on the sampled sample; the second receiving module is used for receiving second input information input by a user, wherein the second input information is the pH value, the oxidation-reduction potential and the conductivity of the immersion liquid obtained through detection.
6. The immersion liquid cooling monitoring method of the reinforced intelligent computing server, which is applied to the immersion liquid cooling monitoring system of the reinforced intelligent computing server as claimed in claim 1, is characterized by comprising the following steps: Obtaining data to be monitored in a smart server cabinet, wherein the data to be monitored comprises flow and temperature of liquid inlet of the smart server cabinet, flow and temperature of liquid outlet of the smart server cabinet, liquid inlet pressure of the smart server cabinet, liquid level and temperature of each computing unit in the smart server, impurities, color conditions, moisture content, pH value, oxidation-reduction potential and conductivity of immersion liquid in the smart server cabinet; and inputting the data to be monitored into a pre-trained decision tree model to obtain a processing scheme output by the decision tree model.
7. The system of claim 6, wherein the training process of the decision tree model comprises: Taking the numerical range of the data to be monitored and a corresponding preset processing scheme as a data set, and preprocessing the data set; Training the decision tree model based on the preprocessed data set until the performance of the decision tree model meets the preset requirement.
8. The method of submerged monitoring of a ruggedized intelligent computing server of claim 6, further comprising: and responding to the first processing scheme output by the decision tree model, and cooling and radiating the intelligent computing server through the liquid cooling radiating cabinet.
9. The method of submerged monitoring of a ruggedized intelligent computing server of claim 6, further comprising: periodically outputting first reminding information, wherein the first reminding information is used for reminding a user to observe impurities and color conditions of the immersion liquid through observing the sight glass valve; And receiving first input information input by a user, wherein the first input information is impurities and color conditions of the immersion liquid.
10. The method of submerged monitoring of a ruggedized intelligent computing server of claim 6, further comprising: periodically outputting second reminding information, wherein the second reminding information is used for reminding a user to sample the immersion liquid through a sampling valve and carrying out pH value detection, oxidation-reduction potential detection and conductivity detection on the sampled sample; And receiving second input information input by a user, wherein the second input information is the pH value, the oxidation-reduction potential and the conductivity of the immersion liquid obtained by detection.

Description

Immersed liquid cooling monitoring system and method for reinforced intelligent computation server Technical Field The application belongs to the technical field of computer cooling, and particularly relates to an immersed liquid cooling monitoring system and method of a reinforced intelligent computing server. Background Compared with general air cooling, cold plate type liquid cooling and other non-contact indirect heat exchange modes, for the reinforced intelligent computing server with high heat flux density, the immersed liquid cooling directly contacts the cooling liquid with the electronic component, heat is directly transferred to the cooling liquid in contact with the cooling liquid, a relatively high heat transfer coefficient can be achieved, the cooling liquid has relatively high heat conductivity and specific heat capacity, the operating temperature change rate is relatively small, the adaptability is high, the heat radiation advantage is obvious, the deployment of the artificial intelligent (ARTIFICIAL INTELLIGENCE, AI) computing units such as a high-density graphics processor (Graphics Processing Unit, a GPU), a neural network processor (Natural Processing Unit, NPU), a central processing unit (Central Processing Unit, a CPU) can be supported, and the requirements of high-density deployment, high-efficiency heat radiation and low noise of the reinforced intelligent computing server in a narrow space are met. The existing monitoring method of the immersed liquid cooling of the reinforced intelligent computing server generally monitors the temperature, the pressure and the flow of the key points of the immersed liquid cooling system in real time, however, only the temperature, the pressure and the flow of the key points of the immersed liquid cooling system are considered, only part of working conditions to be treated can be considered, and the immersed liquid cooling system is treated by relying on experience of maintenance personnel when the treatment is needed, so that the stability and the reliability of the immersed liquid cooling system are lower. Disclosure of Invention Aiming at the defects of the prior art, the application aims to provide an immersed liquid cooling monitoring system and method of a reinforced intelligent computing server, and aims to solve the problem that the immersed liquid cooling system is low in stability and reliability due to the fact that only the temperature, the pressure and the flow of key points of the immersed liquid cooling system are considered and the immersed liquid cooling system is processed by relying on experience of maintenance personnel when the immersed liquid cooling system needs to be processed in the prior art. To achieve the above object, in a first aspect, the present application provides an immersion liquid cooling monitoring system of a reinforced intelligent computing server, including: The monitoring module is used for acquiring data to be monitored in the intelligent computing server cabinet, wherein the data to be monitored comprises flow and temperature of liquid inlet of the intelligent computing server cabinet, flow and temperature of liquid outlet of the intelligent computing server cabinet, liquid inlet pressure of the intelligent computing server cabinet, liquid level and temperature of each computing unit in the intelligent computing server, impurities, color conditions, moisture content, pH value, oxidation-reduction potential and conductivity of immersion liquid in the intelligent computing server cabinet; And the analysis module is used for inputting the data to be monitored into a pre-trained decision tree model to obtain a processing scheme output by the decision tree model. The application not only carries out stable real-time monitoring on the temperature, pressure, flow of key points of the immersed liquid cooling system, but also brings the temperature, liquid level and other data of the key calculation units into the acquisition range, increases the impurity, color condition, pH value, oxidation-reduction potential and conductivity of the immersed liquid in the intelligent calculation server cabinet, inputs the monitored data into the pre-trained decision tree model to obtain a processing scheme of model output, including maintenance period suggestion, dehydration treatment and the like, thereby improving the processing efficiency, overcoming the defect of relying on the experience of maintenance personnel, taking more working conditions to be processed into consideration, simplifying and guaranteeing the maintenance operation links, and enhancing the stability and reliability of the operation of the immersed liquid cooling monitoring system. According to the immersion liquid cooling monitoring system of the reinforced intelligent computing server provided by the application, the training process of the decision tree model comprises the following steps: Taking the numerical range of the data to be monitored and a corre