Anomaly detection and repair system for liquid cooling systems

Abstract

An anomaly detection and repair system for a liquid cooling system includes at least a sensing module, an edge processing unit, a determination unit, and a control unit. The sensing module includes at least one ultrasonic sensor disposed within the liquid cooling system, used to actively emit an ultrasonic signal and receive a corresponding detection spectrum to obtain a raw signal from the liquid cooling system, wherein the raw signal includes at least an acoustic signal and at least one physical parameter. The edge processing unit is electrically connected to the sensing module and is used to extract features from the raw signal to establish a multi-dimensional feature vector, wherein the multi-dimensional feature vector includes at least a group of acoustic features and a group of physical parameters acquired by the sensing module. The determination unit is electrically connected to the edge processing unit and is used to perform a physical consensus determination based on the multi-dimensional feature vector to identify a stable state transition of the liquid cooling system and determine a corresponding anomaly type. The control unit is electrically connected to the determination unit and is used to consult a preset comparison relationship based on the anomaly type and output a corresponding active repair control signal to trigger the liquid cooling system to execute a corresponding active repair control action.

Claims

1. An anomaly detection and repair system for a liquid cooling system, comprising: A sensing module, including at least one ultrasonic sensor disposed within the liquid cooling system, is used to actively emit an ultrasonic signal and receive a corresponding detection spectrum to obtain a raw signal from the liquid cooling system, wherein the raw signal includes at least an acoustic signal and at least one physical parameter; an edge processing unit, electrically connected to the sensing module, is used to extract features from the raw signal to establish a multi-dimensional feature vector, wherein the multi-dimensional feature vector includes at least an acoustic feature group and a physical parameter group acquired by the sensing module; a determination unit, electrically connected to the edge processing unit, is used to perform a physical consensus determination based on the multi-dimensional feature vector to identify a stable state transition of the liquid cooling system and determine a corresponding anomaly type; and a control unit, electrically connected to the determination unit, is used to consult a preset comparison relationship based on the anomaly type and output a corresponding active repair control signal to trigger the liquid cooling system to perform a corresponding active repair control action.

2. The anomaly detection and repair system for a liquid cooling system as described in claim 1, wherein the preset comparison relationship is a lookup table method, based on the multidimensional feature vector matching a key feature combination, the control unit automatically compares and triggers the corresponding active repair control action.

3. The anomaly detection and repair system for a liquid cooling system as described in claim 2, wherein the key feature combination includes at least one of the corresponding multidimensional feature vectors: a first combination, a second combination, a third combination, and a fourth combination, and the anomaly type includes at least a bubble, a destructive cavitation, a critical dry burning, and a structural microcrack; and when the multidimensional feature vector matches the first combination, the anomaly type is determined to be a bubble, and the active repair control action executes a pulse width modulation frequency conversion to increase the fluid flow rate in the liquid cooling system; when the multidimensional feature vector matches... In the second combination, the anomaly type is determined to be destructive cavitation and the active repair control action is to adjust the operating frequency of a pump. When the multidimensional feature vector meets the third combination, the anomaly type is determined to be critical dry burning and the active repair control action is to forcibly reduce the frequency or cut off the power supply within a preset time range. When the multidimensional feature vector meets the fourth combination, the anomaly type is determined to be structural microcracks and the active repair control action is to close the solenoid valve corresponding to the specific node of the structural microcrack in the liquid cooling system to isolate the leakage.

4. The anomaly detection and repair system for a liquid cooling system as described in claim 3, wherein the at least one ultrasonic sensor includes a plurality of ultrasonic sensors and the determination unit further uses an array of ultrasonic sensors to spatially locate the microcrack in the structure at the centimeter level using triangulation or beamforming technology to determine the location of the specific node of the microcrack.

5. The anomaly detection and repair system for a liquid cooling system as described in claim 1, wherein the sensing module further includes a pressure sensor, a flow sensor and a dielectric parameter sensor, the pressure sensor being used to generate a pressure change parameter, the flow sensor being used to generate a flow change parameter, and the dielectric parameter sensor being used to generate a dielectric parameter.

6. The anomaly detection and repair system for a liquid cooling system as described in claim 1, wherein the at least one ultrasonic sensor includes a plurality of ultrasonic sensors distributed at different nodes in the flow path of the liquid cooling system to form a spatial monitoring network.

7. The anomaly detection and repair system for a liquid cooling system as described in claim 1, wherein the group of physical parameters includes: At least one or a combination of a pressure change parameter, a flow rate change parameter, a dielectric parameter, a time difference parameter between sensors, and a spatial correlation parameter.

8. The anomaly detection and repair system for a liquid cooling system as described in claim 1, wherein the raw signal acquired by the at least one ultrasonic sensor further includes a passive acoustic signal naturally generated by bubble generation, bubble collapse, cavitation, flow field disturbance, structural cracks, structural microcracks, leakage or phase change events within the liquid cooling system, and the edge processing unit further performs feature extraction on the passive acoustic signal to establish the multidimensional feature vector.

Images