CN-121980233-A - Detection method of communication machine room and related equipment

CN121980233ACN 121980233 ACN121980233 ACN 121980233ACN-121980233-A

Abstract

The embodiment of the application provides a detection method and related equipment for a communication machine room, and belongs to the technical field of machine room equipment monitoring. The method comprises the steps of obtaining machine room environment data respectively collected by a plurality of environment sensors, carrying out dynamic weighting fusion calculation on the machine room sensing data to obtain total environment feature vectors, analyzing the total environment feature vectors through a lightweight model to obtain machine room detection results, wherein the lightweight model is obtained by learning a trained large detection model, and the lightweight model is deployed on the edge server. The lightweight model of the embodiment of the application can still maintain higher analysis precision and generalization capability while remarkably reducing the occupation of calculation and storage resources, and finally outputs the accurate detection result of the abnormal environment or equipment state of the machine room, thereby not only effectively improving the real-time performance and reliability of the monitoring of the environment of the machine room, but also enhancing the autonomous decision-making and response capability at the edge side.

Inventors

WEN ZHENSHAN
XIAO YISHAN
LIU JIANFENG
DENG DINGFENG
YE KEFENG
CHEN SIJIONG
LIU ZUOZHENG

Assignees

宜通世纪科技股份有限公司

Dates

Publication Date: 20260505
Application Date: 20251204

Claims (10)

1. The detection method of the communication machine room is characterized by being applied to electronic equipment, wherein the electronic equipment is in communication connection with an edge server, and the method comprises the following steps of: Acquiring machine room environment data respectively acquired by a plurality of environment sensors; carrying out dynamic weighting fusion calculation on the sensing data of each machine room to obtain a total environment feature vector; And analyzing the total environmental feature vector through a lightweight model to obtain a machine room detection result, wherein the lightweight model is obtained by learning a large-scale detection model after training, and the lightweight model is deployed on the edge server.
2. The method of claim 1, wherein before the dynamically weighted fusion calculation is performed on each of the machine room sensor data to obtain a total environmental feature vector, the method further comprises: performing data cleaning, outlier processing, data synchronization processing and data standardization processing on the machine room environment data respectively acquired by the plurality of sensors to obtain machine room sensing data after each processing; The step of carrying out dynamic weighting fusion calculation on the machine room sensing data to obtain a total environment feature vector comprises the following steps: and carrying out dynamic weighted fusion calculation on the machine room sensing data after each treatment to obtain a total environment feature vector.
3. The method of claim 1, wherein the performing a dynamic weighted fusion calculation on each machine room sensor data to obtain a total environmental feature vector includes: Identifying each machine room sensing data, and distributing corresponding initial weight parameters according to the influence degree of each machine room sensing data on the stability of the machine room; Acquiring precision data corresponding to each environmental sensor; according to the precision data corresponding to each environmental sensor, adjusting initial weight parameters corresponding to the machine room sensing data acquired by each environmental sensor to obtain target weight parameters corresponding to each machine room sensing data; And carrying out weighted calculation according to each machine room sensing data and the target weight parameter corresponding to each machine room sensing data to obtain the total environment feature vector.
4. The method of claim 3, wherein the adjusting the initial weight parameter corresponding to the machine room sensing data collected by each environmental sensor according to the accuracy data corresponding to each environmental sensor to obtain the target weight parameter corresponding to each machine room sensing data includes: If the decrease of the precision data corresponding to the first environmental sensor is detected, reducing an initial weight parameter corresponding to the machine room sensing data acquired by the first environmental sensor to obtain the target weight parameter; if the rising of the precision data corresponding to the first environment sensor is detected, increasing an initial weight parameter corresponding to the machine room sensing data acquired by the first environment sensor to obtain the target weight parameter.
5. The method of claim 1, wherein the electronic device is communicatively connected to a backup power source, a fire protection system, and an alarm buzzer, and wherein after the analyzing the total environmental feature vector by the lightweight model, the method further comprises: performing grading detection on the machine room detection result, and if the machine room detection result is detected to be a primary alarm, controlling to start the standby generator and triggering the fire protection system; If the detection result of the machine room is detected to be a secondary alarm, the alarm buzzer is controlled to carry out intermittent alarm; And if the machine room detection result is detected to be three-level alarm, displaying the machine room detection result on a target interface.
6. The method of claim 1, wherein the analyzing the total environmental feature vector by the lightweight model to obtain the machine room detection result comprises: Respectively reasoning through a neural network and a long-term memory network in the lightweight model to obtain a neural network result and a long-term memory network result; and fusing the neural network result and the long-term and short-term memory network result to obtain the machine room detection result.
7. The method according to any one of claims 1 to 6, wherein before the acquiring the machine room environment data respectively acquired by the plurality of environment sensors, the method further comprises: the method comprises the steps of obtaining a sample machine room environment data set, wherein the sample machine room environment data set comprises a plurality of sample machine room environment data and target machine room detection results corresponding to the just-sample machine room environment data; Inputting the sample machine room environment data set into a large detection model to be trained, analyzing the sample machine room environment data through the large detection model to obtain prediction machine room detection results corresponding to the sample machine room environment data, and adjusting model parameters of the large detection model according to errors between the prediction machine room detection results corresponding to the sample machine room environment data and target machine room detection; If the error between the predicted machine room detection result corresponding to the sample machine room environment data and the target machine room detection is smaller than a preset error threshold value, a large-scale detection model after training is obtained; deleting convolution kernels of the trained large detection model with contribution degree lower than a preset contribution threshold value to obtain a processed large detection model; and extracting and learning the processed large detection model to obtain a lightweight model.
8. A detection device for a communication room, the device comprising: the data acquisition module is used for acquiring the machine room environment data respectively acquired by the plurality of environment sensors; the vector calculation module is used for carrying out dynamic weighting fusion calculation on the sensing data of each machine room to obtain a total environment characteristic vector; The vector analysis module is used for analyzing the total environmental feature vector through a lightweight model to obtain a machine room detection result, the lightweight model is obtained by learning a trained large detection model, and the lightweight model is deployed on the edge server.
9. A computer device, characterized in that it comprises a memory storing a computer program and a processor implementing the method according to any one of claims 1 to 7 when executing the computer program.
10. A computer readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the method of any one of claims 1 to 7.

Description

Detection method of communication machine room and related equipment Technical Field The application relates to the technical field of machine room equipment monitoring, in particular to a detection method of a communication machine room and related equipment. Background The communication machine room is used as a core infrastructure for supporting the operation of key industries such as the Internet, telecommunication, finance and the like, a server, network equipment, a storage system, a matched series of complex systems such as power supply, refrigeration, security and the like are integrated in the communication machine room, the data transmission, storage and processing tasks of key businesses are born, and the decisive influence on business continuity and service quality is realized. However, such facilities face various operational risks, including mainly environmental factors, power system problems, network equipment anomalies, human factors, and other factors, which, once they occur, may cause service interruption, thereby causing serious economic and social impacts. Disclosure of Invention The embodiment of the application mainly aims to provide a detection method and related equipment for a communication machine room, which not only effectively improves the real-time performance and reliability of machine room environment monitoring, but also enhances the autonomous decision making and response capability at the edge side. In order to achieve the above object, an aspect of an embodiment of the present application provides a method for detecting a communication room, where the method includes: Acquiring machine room environment data respectively acquired by a plurality of environment sensors; carrying out dynamic weighting fusion calculation on the sensing data of each machine room to obtain a total environment feature vector; And analyzing the total environmental feature vector through a lightweight model to obtain a machine room detection result, wherein the lightweight model is obtained by learning a large-scale detection model after training, and the lightweight model is deployed on the edge server. In some embodiments, before the dynamically weighted fusion calculation is performed on each machine room sensing data to obtain a total environmental feature vector, the method further includes: performing data cleaning, outlier processing, data synchronization processing and data standardization processing on the machine room environment data respectively acquired by the plurality of sensors to obtain machine room sensing data after each processing; The step of carrying out dynamic weighting fusion calculation on the machine room sensing data to obtain a total environment feature vector comprises the following steps: and carrying out dynamic weighted fusion calculation on the machine room sensing data after each treatment to obtain a total environment feature vector. In some embodiments, the performing a dynamic weighted fusion calculation on each machine room sensing data to obtain a total environmental feature vector includes: Identifying each machine room sensing data, and distributing corresponding initial weight parameters according to the influence degree of each machine room sensing data on the stability of the machine room; Acquiring precision data corresponding to each environmental sensor; according to the precision data corresponding to each environmental sensor, adjusting initial weight parameters corresponding to the machine room sensing data acquired by each environmental sensor to obtain target weight parameters corresponding to each machine room sensing data; And carrying out weighted calculation according to each machine room sensing data and the target weight parameter corresponding to each machine room sensing data to obtain the total environment feature vector. In some embodiments, the adjusting, according to the accuracy data corresponding to each environmental sensor, an initial weight parameter corresponding to the machine room sensing data collected by each environmental sensor to obtain a target weight parameter corresponding to each machine room sensing data includes: If the decrease of the precision data corresponding to the first environmental sensor is detected, reducing an initial weight parameter corresponding to the machine room sensing data acquired by the first environmental sensor to obtain the target weight parameter; if the rising of the precision data corresponding to the first environment sensor is detected, increasing an initial weight parameter corresponding to the machine room sensing data acquired by the first environment sensor to obtain the target weight parameter. In some embodiments, the electronic device is in communication connection with a standby power supply, a fire protection system and an alarm buzzer, and after the analysis of the total environmental feature vector by the lightweight model, the method further comprises: performing grading det