A communication machine room refrigeration monitoring method and system based on temperature distribution field

Abstract

The application relates to a communication machine room refrigeration monitoring method and system based on a temperature distribution field, which comprises the following steps: step 1, collecting point clouds near the air inlet of each group of cabinets; step 2, obtaining images of the pixel range of each high-precision temperature sensor; step 3, obtaining infrared images of the pixel range of each high-precision temperature sensor; step 4, obtaining the temperature value of each pixel in the infrared image of the pixel range of the air inlet of each group of cabinets; and step 5, respectively calculating the index of the cabinet being higher than the recommended highest temperature range and the index of the cabinet being lower than the recommended lowest temperature range, and combining the index of the cabinet being higher than the recommended highest temperature range and the index of the cabinet being lower than the recommended lowest temperature range to optimize the machine room air conditioner scheme. The application can monitor the air inlet temperature of the cabinet temperature field in the machine room in real time, so as to quickly respond to the refrigeration effect of the machine room air conditioner, improve the stability of the operation of the communication machine room equipment, and guarantee the operation environment of the communication machine room equipment.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of communication dynamic environment, in particular to a communication room refrigeration monitoring method and system based on a temperature distribution field. BACKGROUND

[0002] In recent years, with the rapid development of communication technology in power, the importance of communication rooms is increasingly prominent. As an important channel for power communication data transmission, business access and system operation, the communication room is crucial for business operation. The internal environment temperature of the room is crucial for the normal operation of the internal equipment of the communication room, and the national standard has strict requirements for this.

[0003] For the special situation of the communication room, directly adjusting the air conditioning set parameters of the communication room is not realistic, not only cannot know the local hot spot distribution in the room, but also has certain risk, improper temperature adjustment will cause the temperature of the room to be too high, and damage the equipment in the room, therefore, the environment temperature of the room, especially the cabinet temperature, needs to be dynamically monitored to reflect the air conditioning operation in real time.

[0004] At present, the temperature monitoring of the communication room is mainly realized through the power environment monitoring system, and the traditional power environment monitoring system cannot globally reflect the operation of the room cabinet, and it is more difficult to realize the overall monitoring of the three-dimensional temperature field of the room. Therefore, it is necessary to monitor the refrigeration of the communication room based on the temperature distribution field. SUMMARY

[0005] The purpose of the embodiment of the application is to provide a communication room refrigeration monitoring method and system based on a temperature distribution field, which can monitor the inlet temperature of the cabinet temperature field in the room in real time, so as to quickly reflect the refrigeration effect of the air conditioner in the room, improve the stability of the operation of the equipment in the room, and protect the operation environment of the equipment in the room.

[0006] In order to achieve the above purpose, the application provides the following technical scheme:

[0007] In a first aspect, the application provides a communication room refrigeration monitoring system based on a temperature distribution field, which comprises a microprocessor, a laser point cloud scanning device, a plurality of high-precision temperature sensors, an image sensor, an infrared camera, a warning module and a display module.

[0008] The microprocessor is connected with the laser point cloud scanning device, the plurality of high-precision temperature sensors, the image sensor, the infrared camera, the warning module and the display module in sequence.

[0009] Each high-precision temperature sensor is arranged at the inlet of the cabinet of the communication room.

[0010] The laser point cloud scanning device is deployed on the top of the communication equipment room and faces the air inlet of the communication equipment room cabinet;

[0011] The infrared camera is deployed on the top of the communication equipment room and faces the air inlet of the communication equipment room cabinet;

[0012] The microprocessor, early warning module, and display module are deployed on the top of the cabinet below the air inlet of the communication equipment room.

[0013] The laser point cloud scanning device is a three-dimensional laser point cloud scanning device equipped with a gimbal.

[0014] Secondly, embodiments of this application provide a method for monitoring the cooling of a communication equipment room based on a temperature distribution field, comprising the following steps:

[0015] Step 1: The microprocessor collects point clouds near the air inlets of each rack group using a laser point cloud scanning device, performs Euclidean clustering on the point clouds near the air inlets of each rack group to obtain multiple clusters corresponding to the point clouds near the air inlets of each rack group; performs multi-view projection on each cluster corresponding to the point clouds near the air inlets of each rack group, and uses an MVCNN network to identify the clusters near the air inlets of each rack group; and uses the RANSAC plane fitting set morphological method to extract the planar point clouds of the air inlets of each rack group from the clusters near the air inlets of the rack group.

[0016] Step 2: Calculate the rotation matrix between point cloud images using the point cloud coordinates of the laser point cloud scanning device and the image coordinates of the image sensor. Combine the rotation matrix between point cloud images to project the plane point cloud of each rack air inlet onto the corresponding image data to obtain the image of the plane point cloud of each rack air inlet. Use the morphological contour feature extraction method to obtain the pixel range image of the plane point cloud of each rack air inlet. Use the image feature point extraction method to obtain the pixel range image of each high-precision temperature sensor.

[0017] Step 3: Calculate the rotation matrix between the image and infrared coordinates of the image sensor and the infrared coordinates of the infrared camera. Combine the rotation matrix between the image and infrared coordinates to project the pixel range of the air inlet of each cabinet onto the corresponding infrared image data to obtain the infrared image of the pixel range of the air inlet of each cabinet. Combine the rotation matrix between the image and infrared coordinates to project the pixel range of the image of each high-precision temperature sensor onto the corresponding infrared image data to obtain the infrared image of the pixel range of each high-precision temperature sensor.

[0018] Step 4: The microprocessor collects the high-precision temperature of each rack air inlet through each high-precision temperature sensor, and calculates the temperature value of each pixel in the infrared image of the pixel range of each rack air inlet based on the pixel value of the infrared image of the pixel range of each group of rack air inlets, the pixel value of the infrared image of the pixel range of each high-precision temperature sensor, and the high-precision temperature of each rack air inlet.

[0019] Step 5: Calculate the high-precision average temperature of each pixel in the infrared image of the air inlet of each rack group by weighting the temperature values ​​of each pixel. Separate indices for racks exceeding the recommended maximum temperature range and racks falling below the recommended minimum temperature range. Optimize the data center air conditioning solution by combining these indices.

[0020] In step 1, the point cloud near the air inlet of each rack is subjected to Euclidean clustering as follows:

[0021] Based on the scanning resolution of the laser point cloud scanning device, a threshold th1 for the maximum distance between two adjacent points is set.

[0022] Randomly select a laser point as the seed point of the cluster. If the distance between the laser point and the seed point is within the preset threshold range, add the laser point to the cluster and set the laser point as the seed point to continue the search until the number of points in the cluster no longer increases, then the extraction of a cluster is completed.

[0023] Then, select other un-clustered laser points as seed points for new clusters, and repeat the above steps until all points are assigned to different clusters. Let the cluster obtained from the point cloud near the air inlet of the k-th cabinet be:

[0024]

[0025] in, Represents the result obtained from the k-th 3D laser point cloud data. A cluster, The number of clusters obtained from the k-th 3D laser point cloud data;

[0026] The point cloud near the air inlet of each rack is divided into multiple separate clusters, and the point cloud of each rack air inlet appears in the corresponding cluster.

[0027] In step 1, the multi-view projection of the clusters is as follows:

[0028] For a cluster, first calculate its cluster center, and then translate the origin of the point cloud coordinate system to the cluster center;

[0029] Place the first virtual camera at a fixed distance from the cluster on the XOY plane of the point cloud coordinate system, and make the Z-axis of the virtual camera point to the center of the cluster, and the Y-axis of the virtual camera be aligned with the Z-axis of the point cloud coordinate system to generate a 2D projection image from the first perspective.

[0030] Rotate the virtual camera around the Z-axis of the point cloud coordinate system by 30 degrees. o At intervals, a projection is performed once for each rotation, thus generating 2D projected images with 12 viewpoints.

[0031] In step 1, the MVCNN network is used to identify the clusters near the air inlet of each rack as follows:

[0032] The dataset for the MVCNN network is constructed by using the multi-view projection to obtain multi-view projection images of clusters.

[0033] The multi-view projection image first passes through the first stage of the convolutional neural network CNN1, and features are extracted from each image to obtain the multi-view feature map of the air inlet of the cabinet.

[0034] Among them, CNN1 adopts the VGG-M structure, which consists of 5 convolutional layers, 3 fully connected layers and a Softmax classification layer;

[0035] Then, view pooling is used to synthesize the feature maps of a rack air inlet from multiple views into a shape descriptor. The rack air inlet projection images from 12 views are fused to extract the robust shape features of the cluster.

[0036] The second-stage convolutional network CNN2 is used to identify the synthesized shape descriptors and determine whether the current cluster contains rack air inlet point clouds.

[0037] The network was pre-trained on ImageNet and then fine-tuned on the training set used in this paper.

[0038] After the network training is completed, for a cluster, the multi-view projection method is first used to generate 2D images of 12 views, and then input into the trained MVCNN network to determine whether the cluster contains the point cloud of the air inlet of the cabinet.

[0039] Ultimately from clusters Extract clusters containing the air inlet of the cabinet.

[0040]

[0041] Where K represents the number of identified clusters of air inlets in the server racks.

[0042] In step 1, the RANSAC planar fitting set morphological method is used to extract the planar point cloud of the cabinet air inlet cluster as follows:

[0043] For each rack air inlet cluster Furthermore, the RANSAC method is used to iteratively obtain the plane where the air inlet of the cabinet is located, and the interior points of the plane are obtained to remove most of the interference point cloud at the bottom.

[0044] First, use distance d to perform RANSAC plane fitting to obtain the initial in-plane points. Then, reduce the distance to d / 2 and repeat RANSAC fitting until the angle between the plane normal vectors obtained from the two fittings is less than th2. Then stop the iteration. The in-plane points at this time are considered to be the point cloud of the plane where the air inlet of the cabinet is located.

[0045] The final obtained planar point cloud of the rack air inlet is as follows:

[0046]

[0047] in, This represents the planar point cloud of the air inlet of the k-th rack. The first in K represents the number of laser points identified in the plane point cloud of the server rack air inlet. This represents the number of laser points in the plane point cloud of the air inlet of the k-th cabinet.

[0048] In step 4, the temperature value of each pixel in the infrared image of the pixel range of each group of cabinet air inlets is calculated as follows:

[0049] Obtain the pixel value of each pixel in the infrared image of the air inlet of each rack;

[0050] Calculate the pixel mean of the infrared image of multiple pixels within the pixel range of each high-precision temperature sensor;

[0051] The ratio of each pixel in the infrared image of the pixel range of each rack air inlet is obtained by dividing the temperature value of each pixel in the infrared image of multiple pixels in the pixel range of each high-precision temperature sensor by the average pixel value of the infrared image of each pixel in the pixel range of each rack air inlet. This ratio is then multiplied by the high-precision temperature of each rack air inlet to obtain the pixel value of each pixel in the infrared image of the pixel range of each rack air inlet.

[0052] In step 5, the indices for computer cabinets exceeding the recommended maximum temperature range and for server racks falling below the recommended minimum temperature range are as follows:

[0053]

[0054]

[0055] in, This represents the high-precision average temperature at the air inlet of the x-th cabinet group. This index indicates that the server rack's temperature exceeds the recommended maximum range. This index indicates that the server rack is below the recommended minimum temperature range. Indicates the maximum permissible temperature. Indicates the maximum recommended temperature. Indicates the minimum permissible temperature. Indicates the minimum recommended temperature;

[0056] If the index for the server rack is higher than the recommended maximum temperature range and the index for the server rack is lower than the recommended minimum temperature range, both of which are higher than the temperature threshold, then the data center air conditioning solution does not need to be optimized.

[0057] If the index of the cabinet is higher than the recommended maximum temperature range but lower than the temperature threshold, it means that the temperature of the air inlet of the xth cabinet is too high and the cooling needs to be further reduced.

[0058] If the index of the cabinet's temperature below the recommended minimum range is lower than the temperature threshold, it indicates that the temperature at the air inlet of the xth cabinet group is too low, and further energy-saving cooling is needed.

[0059] Compared with the prior art, the beneficial effects of this application are: this application can monitor the air inlet temperature of the cabinet temperature field in the computer room in real time, so as to quickly respond to the cooling effect of the computer room air conditioner, improve the stability of the operation of communication equipment room equipment, and protect the operating environment of communication equipment room equipment. Attached Figure Description

[0060] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0061] Figure 1 This is a system diagram of an embodiment of the present invention;

[0062] Figure 2 This is a flowchart of a method according to an embodiment of the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. It should be noted that similar reference numerals and letters in the following drawings indicate similar items; therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0064] The terms “comprising,” “including,” or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0065] The terms “first,” “second,” etc., are used only to distinguish one entity or operation from another, and should not be construed as indicating or implying relative importance, nor as requiring or implying any such actual relationship or order between these entities or operations.

[0066] Reference Figure 1 The following is combined with Figure 1 The specific embodiment of the present invention is a communication equipment room cooling monitoring system based on a temperature distribution field, comprising:

[0067] 1. Microprocessor; 2. Laser point cloud scanning device; 3. Multiple high-precision temperature sensors; 4. Image sensor; 5. Infrared camera; 6. Early warning module; 7. Display module;

[0068] The microprocessor 1 is sequentially connected to the laser point cloud scanning device 2, multiple high-precision temperature sensors 3, image sensor 4, infrared camera 5, early warning module 6, and display module 7.

[0069] Each of the high-precision temperature sensors 3 is deployed at the air inlet of the cabinet in the communication equipment room;

[0070] The laser point cloud scanning device 2 is deployed on the top of the communication equipment room and faces the air inlet of the communication equipment room cabinet;

[0071] The infrared camera 5 is deployed on the top of the communication equipment room and faces the air inlet of the communication equipment room cabinet;

[0072] The microprocessor 1, the early warning module 6, and the display module 7 are deployed on the top of the cabinet below the air inlet of the cabinet in the communication equipment room;

[0073] The laser point cloud scanning device is a three-dimensional laser point cloud scanning device equipped with a gimbal;

[0074] The microprocessor selected is the ARM9 microprocessor;

[0075] The laser point cloud scanning device is a VLP-16 laser scanning radar;

[0076] The high-precision temperature sensor is a DS18B20 temperature sensor.

[0077] The image sensor selected is a gray-dot industrial camera;

[0078] The infrared camera is an infrared thermal imaging sensor;

[0079] The early warning module is a voice alarm module;

[0080] The display module is an LCD screen module;

[0081] like Figure 2 The technical solution of the present invention is a method for monitoring the cooling of a communication equipment room based on a temperature distribution field, which specifically includes the following steps:

[0082] Step 1: The microprocessor collects point clouds near the air inlets of each rack group using a laser point cloud scanning device, performs Euclidean clustering on the point clouds near the air inlets of each rack group to obtain multiple clusters corresponding to the point clouds near the air inlets of each rack group; performs multi-view projection on each cluster corresponding to the point clouds near the air inlets of each rack group, and uses an MVCNN network to identify the clusters near the air inlets of each rack group; and uses the RANSAC plane fitting set morphological method to extract the planar point clouds of the air inlets of each rack group from the clusters near the air inlets of the rack group.

[0083] Step 1 involves performing Euclidean clustering on the point clouds near the air inlets of each rack group:

[0084] Based on the scanning resolution of the laser point cloud scanning device, a threshold th1 for the maximum distance between two adjacent points is set.

[0085] The algorithm randomly selects a laser point as the seed point of the cluster. If the distance between the laser point and the seed point is within the preset threshold of 0.1, the laser point is added to the cluster and set as the seed point to continue the search until the number of points in the cluster no longer increases, thus completing the extraction of a cluster.

[0086] Then, select other un-clustered laser points as seed points for new clusters, and repeat the above steps until all points are assigned to different clusters. Let the cluster obtained from the point cloud near the air inlet of the k-th cabinet be:

[0087]

[0088] in, Represents the result obtained from the k-th 3D laser point cloud data. A cluster, The number of clusters obtained from the k-th 3D laser point cloud data;

[0089] The point cloud near the air inlet of each rack is divided into multiple separate clusters, and the point cloud of the air inlet of each rack appears in the corresponding cluster.

[0090] Step 1, which involves performing multi-view projection on the clusters, is as follows:

[0091] For a cluster, first calculate its cluster center, and then translate the origin of the point cloud coordinate system to the cluster center;

[0092] Place the first virtual camera at a fixed distance from the cluster on the XOY plane of the point cloud coordinate system, and make the Z-axis of the virtual camera point to the center of the cluster, and the Y-axis of the virtual camera be aligned with the Z-axis of the point cloud coordinate system to generate a 2D projection image from the first perspective.

[0093] The virtual camera is rotated around the Z-axis of the point cloud coordinate system, and a projection is performed at 30° intervals, so that 2D projected images with 12 viewpoints can be generated.

[0094] Step 1 describes using the MVCNN network to identify clusters near the air inlets of each rack:

[0095] The dataset for the MVCNN network is constructed by using the multi-view projection to obtain multi-view projection images of clusters.

[0096] The multi-view projection image first passes through the first stage of the convolutional neural network CNN1, and features are extracted from each image to obtain the multi-view feature map of the air inlet of the cabinet.

[0097] Among them, CNN1 adopts the VGG-M structure, which consists of 5 convolutional layers, 3 fully connected layers and a Softmax classification layer;

[0098] Then, view pooling is used to synthesize the feature maps of a rack air inlet from multiple views into a shape descriptor. The rack air inlet projection images from 12 views are fused to extract the robust shape features of the cluster.

[0099] The second-stage convolutional network CNN2 is used to identify the synthesized shape descriptors and determine whether the current cluster contains rack air inlet point clouds.

[0100] The network was pre-trained on ImageNet and then fine-tuned on the training set used in this paper.

[0101] After the network training is completed, for a cluster, the multi-view projection method is first used to generate 2D images of 12 views, and then input into the trained MVCNN network to determine whether the cluster contains the point cloud of the air inlet of the cabinet.

[0102] Ultimately from clusters Extract clusters containing the air inlet of the cabinet.

[0103]

[0104] Where K is the number of identified clusters of air inlets in the cabinet;

[0105] Step 1 describes the extraction of the rack air inlet planar point cloud from the rack air inlet cluster using the RANSAC planar fitting set morphological method:

[0106] For each rack air inlet cluster Furthermore, the RANSAC method is used to iteratively obtain the plane where the air inlet of the cabinet is located, and the interior points of the plane are obtained to remove most of the interference point cloud at the bottom.

[0107] First, perform RANSAC plane fitting using distance d to obtain initial in-plane points. Then, reduce the distance to d / 2 and repeat RANSAC fitting until the angle between the plane normal vectors obtained from the two fittings is less than th2=5. Then stop the iteration. The in-plane points at this point are considered to be the point cloud of the plane where the air inlet of the cabinet is located.

[0108] The final obtained planar point cloud of the rack air inlet is as follows:

[0109]

[0110] in, This represents the planar point cloud of the air inlet of the k-th rack. The first in K represents the number of laser points identified in the plane point cloud of the server rack air inlet. This represents the number of laser points in the planar point cloud of the air inlet of the k-th rack;

[0111] Step 2: Calculate the rotation matrix between point cloud images using the point cloud coordinates of the laser point cloud scanning device and the image coordinates of the image sensor. Combine the rotation matrix between point cloud images to project the plane point cloud of each rack air inlet onto the corresponding image data to obtain the image of the plane point cloud of each rack air inlet. Use the morphological contour feature extraction method to obtain the pixel range image of the plane point cloud of each rack air inlet. Use the image feature point extraction method to obtain the pixel range image of each high-precision temperature sensor.

[0112] Step 3: Calculate the rotation matrix between the image and infrared coordinates of the image sensor and the infrared coordinates of the infrared camera. Combine the rotation matrix between the image and infrared coordinates to project the pixel range of the air inlet of each cabinet onto the corresponding infrared image data to obtain the infrared image of the pixel range of the air inlet of each cabinet. Combine the rotation matrix between the image and infrared coordinates to project the pixel range of the image of each high-precision temperature sensor onto the corresponding infrared image data to obtain the infrared image of the pixel range of each high-precision temperature sensor.

[0113] Step 4: The microprocessor collects the high-precision temperature of each rack air inlet through each high-precision temperature sensor, and calculates the temperature value of each pixel in the infrared image of the pixel range of each rack air inlet based on the pixel value of the infrared image of the pixel range of each group of rack air inlets, the pixel value of the infrared image of the pixel range of each high-precision temperature sensor, and the high-precision temperature of each rack air inlet.

[0114] Step 4 involves calculating the temperature value of each pixel in the infrared image of the pixel range of each rack air inlet, as detailed below:

[0115] Obtain the pixel value of each pixel in the infrared image of the air inlet of each rack;

[0116] Calculate the pixel mean of the infrared image of multiple pixels within the pixel range of each high-precision temperature sensor;

[0117] The ratio of each pixel in the infrared image of the pixel range of each rack air inlet is obtained by dividing the temperature value of each pixel in the infrared image of multiple pixels in the pixel range of each high-precision temperature sensor by the average pixel value of the infrared image of each rack air inlet. This ratio is then multiplied by the high-precision temperature of each rack air inlet to obtain the pixel value of each pixel in the infrared image of the pixel range of each rack air inlet.

[0118] Step 5: Calculate the high-precision average temperature of each pixel in the infrared image of the air inlet of each rack by weighting the temperature values ​​of each pixel. Separately calculate the index of the computer rack being higher than the recommended maximum temperature range and the index of the rack being lower than the recommended minimum temperature range. Optimize the data center air conditioning solution by combining the indices of the racks being higher than the recommended maximum temperature range and the racks being lower than the recommended minimum temperature range.

[0119] The indices for computer cabinets exceeding the recommended maximum temperature range and for server racks falling below the recommended minimum temperature range mentioned in step 5 are as follows:

[0120]

[0121]

[0122] in, This represents the high-precision average temperature at the air inlet of the x-th cabinet group. This index indicates that the server rack's temperature exceeds the recommended maximum range. This index indicates that the server rack is below the recommended minimum temperature range. Indicates the maximum permissible temperature. Indicates the maximum recommended temperature. Indicates the minimum permissible temperature. Indicates the minimum recommended temperature;

[0123] If the index for the server rack is higher than the recommended maximum temperature range and the index for the server rack is lower than the recommended minimum temperature range, both of which are higher than the temperature threshold, then the data center air conditioning solution does not need to be optimized.

[0124] If the index of the cabinet is higher than the recommended maximum temperature range but lower than the temperature threshold, it means that the temperature at the air inlet of the xth cabinet is too high and the cooling needs to be further reduced.

[0125] If the index of the cabinet's temperature below the recommended minimum range is lower than the temperature threshold, it indicates that the temperature at the air inlet of cabinet group x is too low, and further energy-saving cooling is needed.

[0126] It should be understood that any parts not described in detail in this specification belong to the prior art.

[0127] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for monitoring the cooling of a communication equipment room based on a temperature distribution field, characterized in that, Includes the following steps: Step 1: The microprocessor collects point clouds near the air inlets of each rack group using a laser point cloud scanning device, performs Euclidean clustering on the point clouds near the air inlets of each rack group to obtain multiple clusters corresponding to the point clouds near the air inlets of each rack group; performs multi-view projection on each cluster corresponding to the point clouds near the air inlets of each rack group, and uses an MVCNN network to identify the clusters near the air inlets of each rack group; and uses the RANSAC plane fitting set morphological method to extract the planar point clouds of the air inlets of each rack group from the clusters near the air inlets of the rack group. Step 2: Calculate the rotation matrix between point cloud images using the point cloud coordinates of the laser point cloud scanning device and the image coordinates of the image sensor. Combine the rotation matrix between point cloud images to project the plane point cloud of each rack air inlet onto the corresponding image data to obtain the image of the plane point cloud of each rack air inlet. Use the morphological contour feature extraction method to obtain the pixel range image of the plane point cloud of each rack air inlet. Use the image feature point extraction method to obtain the pixel range image of each high-precision temperature sensor. Step 3: Calculate the rotation matrix between the image and infrared coordinates of the image sensor and the infrared coordinates of the infrared camera. Combine the rotation matrix between the image and infrared coordinates to project the pixel range of the air inlet of each cabinet onto the corresponding infrared image data to obtain the infrared image of the pixel range of the air inlet of each cabinet. Combine the rotation matrix between the image and infrared coordinates to project the pixel range of the image of each high-precision temperature sensor onto the corresponding infrared image data to obtain the infrared image of the pixel range of each high-precision temperature sensor. Step 4: The microprocessor collects the high-precision temperature of each rack air inlet through each high-precision temperature sensor. Based on the pixel values ​​of the infrared image of the pixel range of each rack air inlet, the pixel values ​​of the infrared image of the pixel range of each high-precision temperature sensor, and the high-precision temperature of each rack air inlet, the microprocessor calculates the temperature value of each pixel in the infrared image of the pixel range of each rack air inlet. Step 5: Calculate the high-precision average temperature of each pixel in the infrared image of the air inlet of each rack group by weighting the temperature values ​​of each pixel. Separate indices for racks exceeding the recommended maximum temperature range and racks falling below the recommended minimum temperature range. Optimize the data center air conditioning solution by combining these indices.

2. The communication equipment room cooling monitoring method based on temperature distribution field according to claim 1, characterized in that, In step 1, the point cloud near the air inlet of each rack is subjected to Euclidean clustering as follows: Based on the scanning resolution of the laser point cloud scanning device, a threshold th1 for the maximum distance between two adjacent points is set. Randomly select a laser point as the seed point of the cluster. If the distance between the laser point and the seed point is within the preset threshold range, add the laser point to the cluster and set the laser point as the seed point to continue the search until the number of points in the cluster no longer increases, then the extraction of a cluster is completed. Then, select other un-clustered laser points as seed points for new clusters, and repeat the above steps until all points are assigned to different clusters. Let the cluster obtained from the point cloud near the air inlet of the k-th cabinet be: ， in, Represents the result obtained from the k-th 3D laser point cloud data. A cluster, The number of clusters obtained from the k-th 3D laser point cloud data; The point cloud near the air inlet of each rack is divided into multiple separate clusters, and the point cloud of each rack air inlet appears in the corresponding cluster.

3. The communication equipment room cooling monitoring method based on temperature distribution field according to claim 2, characterized in that, In step 1, the multi-view projection of the clusters is as follows: For a cluster, first calculate its cluster center, and then translate the origin of the point cloud coordinate system to the cluster center; Place the first virtual camera at a fixed distance from the cluster on the XOY plane of the point cloud coordinate system, and make the Z-axis of the virtual camera point to the center of the cluster, and the Y-axis of the virtual camera be aligned with the Z-axis of the point cloud coordinate system to generate a 2D projection image from the first perspective. Rotate the virtual camera around the Z-axis of the point cloud coordinate system by 30 degrees. o At intervals, a projection is performed once for each rotation, thus generating 2D projected images with 12 viewpoints.

4. The communication equipment room cooling monitoring method based on temperature distribution field according to claim 3, characterized in that, In step 1, the MVCNN network is used to identify the clusters near the air inlet of each rack as follows: The dataset for the MVCNN network is constructed by using the multi-view projection to obtain multi-view projection images of clusters. The multi-view projection image first passes through the first stage of the convolutional neural network CNN1, and features are extracted from each image to obtain the multi-view feature map of the air inlet of the cabinet. Among them, CNN1 adopts the VGG-M structure, which consists of 5 convolutional layers, 3 fully connected layers and a Softmax classification layer; Then, view pooling is used to synthesize the feature maps of a rack air inlet from multiple views into a shape descriptor. The rack air inlet projection images from 12 views are fused to extract the robust shape features of the cluster. The second-stage convolutional network CNN2 is used to identify the synthesized shape descriptors and determine whether the current cluster contains rack air inlet point clouds. The network was pre-trained on ImageNet and then fine-tuned on the training set used in this paper. After the network training is completed, for a cluster, the multi-view projection method is first used to generate 2D images of 12 views, and then input into the trained MVCNN network to determine whether the cluster contains the point cloud of the air inlet of the cabinet. Ultimately from clusters Extract clusters containing the air inlet of the cabinet. ， Where K represents the number of identified clusters of air inlets in the server racks.

5. A method for monitoring the cooling of a communication equipment room based on a temperature distribution field according to claim 4, characterized in that, In step 1, the RANSAC planar fitting set morphological method is used to extract the planar point cloud of the cabinet air inlet cluster as follows: For each rack air inlet cluster Furthermore, the RANSAC method is used to iteratively obtain the plane where the air inlet of the cabinet is located, and the interior points of the plane are obtained to remove most of the interference point cloud at the bottom. First, use distance d to perform RANSAC plane fitting to obtain the initial in-plane points. Then, reduce the distance to d / 2 and repeat RANSAC fitting until the angle between the plane normal vectors obtained from the two fittings is less than th2. Then stop the iteration. The in-plane points at this time are considered to be the point cloud of the plane where the air inlet of the cabinet is located. The final obtained planar point cloud of the rack air inlet is as follows: ， in, This represents the planar point cloud of the air inlet of the k-th rack. The first in K represents the number of laser points identified in the plane point cloud of the server rack air inlet. This represents the number of laser points in the plane point cloud of the air inlet of the k-th cabinet.

6. The method for monitoring the cooling of a communication equipment room based on a temperature distribution field according to claim 1, characterized in that, In step 4, the temperature value of each pixel in the infrared image of the pixel range of each group of cabinet air inlets is calculated as follows: Obtain the pixel value of each pixel in the infrared image of the air inlet of each rack; Calculate the pixel mean of the infrared image of multiple pixels within the pixel range of each high-precision temperature sensor; The ratio of each pixel in the infrared image of the pixel range of each rack air inlet is obtained by dividing the temperature value of each pixel in the infrared image of multiple pixels in the pixel range of each high-precision temperature sensor by the average pixel value of the infrared image of each pixel in the pixel range of each rack air inlet. This ratio is then multiplied by the high-precision temperature of each rack air inlet to obtain the pixel value of each pixel in the infrared image of the pixel range of each rack air inlet.

7. A method for monitoring the cooling of a communication equipment room based on a temperature distribution field according to claim 1, characterized in that, In step 5, the indices for computer cabinets exceeding the recommended maximum temperature range and for server racks falling below the recommended minimum temperature range are as follows: ， ， in, This represents the high-precision average temperature at the air inlet of the x-th rack group. This index indicates that the server rack's temperature exceeds the recommended maximum range. This index indicates that the server rack's temperature is below the recommended minimum range. Indicates the maximum permissible temperature. Indicates the maximum recommended temperature. Indicates the minimum permissible temperature. Indicates the minimum recommended temperature; If the index for the server rack is higher than the recommended maximum temperature range and the index for the server rack is lower than the recommended minimum temperature range, both of which are higher than the temperature threshold, then the data center air conditioning solution does not need to be optimized. If the index of the cabinet is higher than the recommended maximum temperature range but lower than the temperature threshold, it means that the temperature of the air inlet of the xth cabinet is too high and the cooling needs to be further reduced. If the index of the cabinet's temperature below the recommended minimum range is lower than the temperature threshold, it indicates that the temperature at the air inlet of the xth cabinet group is too low, and further energy-saving cooling is needed.

8. A communication equipment room cooling monitoring system based on a temperature distribution field, used to implement the method of any one of claims 1-7, characterized in that, include: Microprocessor, laser point cloud scanning equipment, multiple high-precision temperature sensors, image sensors, infrared cameras, early warning modules, and display modules; The microprocessor is sequentially connected to the laser point cloud scanning device, multiple high-precision temperature sensors, image sensors, infrared cameras, early warning modules, and display modules. Each of the high-precision temperature sensors is deployed at the air inlet of the cabinet in the communication equipment room; The laser point cloud scanning device is deployed on the top of the communication equipment room and faces the air inlet of the communication equipment room cabinet; The infrared camera is deployed on the top of the communication equipment room and faces the air inlet of the communication equipment room cabinet; The microprocessor, early warning module, and display module are deployed on the top of the cabinet below the air inlet of the communication equipment room. The laser point cloud scanning device is a three-dimensional laser point cloud scanning device equipped with a gimbal.

Images