Method and system for analyzing and positioning data center by using UWB-RFID tag

Abstract

The invention provides a method and a system for analyzing and positioning a data center by using a UWB-RFID tag, and belongs to the field of data center digital image processing. The method comprises the following steps: configuring a UWB-RFID tag for each target device to be detected in a data center machine room, and configuring a sensor in each area of the machine room; in the machine room inspection process, continuous images of target equipment at different angles are collected, and label images are screened out; a two-dimensional normal probability density function is adopted to retrieve image blocks from the screened label images, the image blocks are input into a convolutional neural network to identify and classify the label images, and the classified label images correspond to each target device; and the position of the target equipment is determined according to the positioning information received by the corresponding UWB-RFID receiving end, data of an area sensor closest to the position of the target equipment is collected, and the environment around the target equipment is detected and scored. According to the method, the image processing information extraction capability and the accuracy under the multi-interference condition can be effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of digital image processing in data centers, and specifically relates to a method and system for analysis and positioning in data centers using UWB-RFID tags. Background Technology

[0002] The rapid development of artificial intelligence has enabled the realization of image-based inspection methods. For complex, variable, and heavily occluded RFID and QR code tags, global perception can effectively improve the model's detection capabilities. However, traditional data center financial inventory and intelligent asset inspection target detection networks cannot adapt to tags with varying angles. Existing technologies for inspection robots or video monitoring data center image tag environments are complex, with severe occlusion, making detection difficult. Furthermore, their image processing information extraction capabilities are insufficient under multiple interference conditions, and the accuracy of analysis and positioning is also lacking. Summary of the Invention

[0003] This invention addresses the problems of complex environments, severe occlusion, and high detection difficulty in existing technologies for inspection robots or video monitoring data center image tags. It provides a method and system for analysis and positioning of data centers using UWB-RFID tags.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for analysis and positioning of a data center using UWB-RFID tags, comprising the following steps: Step 1: Configure UWB-RFID tags for each target device to be inspected in the data center server room, divide the server room into areas, and configure sensors in each area; Step 2: During the inspection of the computer room, collect continuous images of the target equipment from different angles and filter out the labeled images; Step 3: Use a two-dimensional normal probability density function to retrieve image patches from the filtered label images, input the image patches into a convolutional neural network to identify and classify the label images, and assign the classified label images to each target device; Step 4: Determine the location of the target device based on the positioning information received by the UWB-RFID receiver corresponding to the UWB-RFID tag configured on the target device, and collect sensor data from the area closest to the target device to detect and score the environment around the target device.

[0005] Optionally, in step 1, the UWB-RFID tag transmits data according to the IEEE 802.15.3a protocol.

[0006] Optionally, in step 2, for continuous images of the target device from different angles, images whose angle and scale of the target device are within a preset range are selected and sorted according to the time axis.

[0007] Optionally, in step 3, a two-dimensional normal probability density function is used to retrieve image patches located near the image center from the filtered label images. The two-dimensional normal probability density function is as follows:

[0008] In the formula, Represents the probability density of an image patch. Indicates the center of the image. Indicates the center of the image patch. The covariance matrix represents the pixel vectors of an image patch. These are adjustable parameters; The image patches are sorted according to their probability density, and the image patch with the highest probability density is selected as the image patch near the image center.

[0009] Optionally, in step 4, during the process of determining the location of the target device, a cumulative probability distribution function is used to calculate the probability distribution of the positioning error, and positioning information with a probability greater than a set threshold is removed.

[0010] Optionally, in step 4, the sensor data includes air quality, temperature, and humidity. The environment in which the target device is located is scored by combining the data from each sensor, and the target device with a score lower than a set threshold is added to the inspection queue.

[0011] Optionally, the step of scoring the environment in which the target device is located by integrating data from various sensors specifically involves: First, the corresponding indicators represented by various sensor data are scored, and then the scores of various indicators are combined using normalization to obtain the total score of the environment in which the target device is located.

[0012] Secondly, the present invention provides a system for data center analysis and positioning using UWB-RFID tags, comprising: UWB-RFID tags are installed on every target device to be inspected in the data center computer room; The UWB-RFID receiver corresponds to the UWB-RFID tag configured on the target device and receives positioning information based on the UWB-RFID tag. Sensors are installed in various areas of the computer room; The acquisition unit is used to acquire continuous images of the target equipment from different angles during the inspection of the computer room; The processing unit is used to filter out label images from the acquired continuous images, retrieve image blocks from the filtered label images using a two-dimensional normal probability density function, input the image blocks into a convolutional neural network to identify and classify the label images, and assign the classified label images to each target device. The evaluation unit is used to determine the location of the target device based on the positioning information received by the UWB-RFID receiver, and to collect sensor data from the area closest to the target device to detect and score the environment around the target device.

[0013] The beneficial effects of this invention are as follows: During video / image inspection in a computer room, this invention assists in equipment inventory and positioning by recognizing tag image features; it maps the segmented image block information to each asset target, and uses a cumulative probability distribution function (CDF) probability distribution chart to calculate the positioning error of the positioning information received by the corresponding UWB-RFID receiver; the smaller the value, the more accurate the positioning. Qualitative evaluation indicators and normalization processing are used to assess and score the equipment environment, facilitating environmental monitoring and management. Through these strategies, this invention effectively improves image processing information extraction capabilities and accuracy under multi-interference conditions. Attached Figure Description

[0014] Figure 1 This is a flowchart of a method for data centers to use UWB-RFID tags for analysis and positioning.

[0015] Figure 2 This is a block diagram of a UWB-RFID transponder. Detailed Implementation

[0016] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0017] In one embodiment, the present invention proposes a method for analysis and positioning of data centers using UWB-RFID tags, the process of which is as follows: Figure 1 As shown, firstly, during video / image inspection and asset inventory in the computer room, tag image feature recognition is used to assist in equipment inventory and location. Secondly, the image block information contained in the image is mapped to each asset target, and the cumulative probability distribution function (CDF) is used to calculate the probability distribution of the location error for the location information received by the UWB-RFID receiver corresponding to the target equipment; the smaller the value, the more accurate the location. Then, the environmental conditions of the equipment are evaluated and scored based on qualitative evaluation indicators (environmental indicators such as air quality, temperature, and humidity) and normalization processing. The method specifically includes the following steps: Step 1: Configure UWB-RFID tags for each target device to be inspected in the data center server room, divide the server room into areas, and configure sensors in each area.

[0018] In this embodiment, the UWB-RFID transponder structural block diagram is as follows: Figure 2 As shown, UWB-RFID tags transmit data using the IEEE 802.15.3a protocol, which meets the following five conditions: 1. It can achieve sub-meter accuracy even in indoor scenes or in situations with obstacles; 2. Small (maximum area is about a few square centimeters) and lightweight (without bulky batteries); 3. Environmentally friendly (made from recyclable materials); 4. Conformable; 5. Low cost to allow deployment of multiple tags in an environment. The design is suitable for backscattered operating conditions and utilizes R-UWB signals (European low ultra-wideband 3.14 GH) for ultra-low power communication and positioning.

[0019] IEEE 802.15.3a is a channel model for short-range (within 10m), high-speed data transmission (110~480Mbit / s), which evolved from the SV model. This model retains the characteristics of multipath clustering and energy following a double exponential distribution found in the SV model, while making certain modifications based on actual measurement results. The statistical characteristics of multipath attenuation have been changed from a Rayleigh distribution to a log-normal distribution, and the channel coefficients are now represented by real numbers.

[0020] Step 2: During the inspection of the computer room, collect continuous images of the target equipment from different angles and filter out the labeled images.

[0021] In continuous video / images, many consecutive images of an asset from different angles can be tracked on a timeline. When the asset images reach the appropriate angle and scale, these image sets are stored for subsequent recognition and classification.

[0022] Step 3: Use a two-dimensional normal probability density function to retrieve image blocks from the filtered label images, input the image blocks into a convolutional neural network to identify and classify the label images, and assign the classified label images to each target device.

[0023] After the aforementioned screening, the qualified image patches will be located in or near the center of an image. In other words, the most key information and the greatest potential value are found near the center of the image. Therefore, a two-dimensional normal probability density function is used to extract a series of image patches near the center of asset images and feed them into a convolutional neural network. The probability density function is as follows:

[0024] In the formula, Represents the probability density of an image patch. Indicates the center of the image. Indicates the center of the image patch. The covariance matrix represents the pixel vectors of an image patch. This is an adjustable parameter.

[0025] Image patches are sorted according to their probability density, and the top-ranked patches are selected as those near the image center. Using this probability density function, a large number of image patches near the image center are randomly and quickly extracted, which has a high probability of covering the identifiable key locations of the image patches. This filters out many background pixels, avoiding interference and redundancy from irrelevant pixels, making image features more prominent, and improving feature extraction performance to a certain extent.

[0026] Step 4: Determine the location of the target device based on the positioning information received by the UWB-RFID receiver corresponding to the UWB-RFID tag configured on the target device, and collect sensor data from the area closest to the target device to detect and score the environment around the target device.

[0027] In this embodiment, the cumulative probability distribution function (CDF) is used to plot the probability distribution of positioning error for the positioning information received by the UWB-RFID receiver corresponding to the target. The smaller the CDF value, the more accurate the positioning. The cumulative probability distribution function describes the probability distribution of a real random variable. In a UWB positioning system, it can produce a probability distribution of positioning error, thereby analyzing the positioning accuracy and the degree to which measurement data is affected by the human body or the environment. If the positioning error exceeds a selected interference threshold, the positioning is considered unsuccessful. The cumulative probability distribution function can intuitively reflect the positioning performance.

[0028] This embodiment evaluates the equipment's environment based on qualitative evaluation indicators and a normalization process. The qualitative evaluation indicators selected include environmental factors such as air quality, temperature, and humidity. On one hand, the air quality in the computer room affects the operation of the equipment within the rack; dust, odors, and other pollutants may enter the rack, affecting heat dissipation and performance. On the other hand, the rack's operating environment needs to maintain suitable temperature and humidity to ensure normal equipment operation and extend its lifespan. Excessively high or low temperatures and humidity can both affect the performance and stability of the equipment within the rack.

[0029] Then, based on the data from various sensors, the corresponding indicators they represent are scored. Next, a normalization process is used to synthesize the scores of all indicators to obtain the total score of the environment in which the target device is located. The formula is as follows:

[0030] In the formula, , , This indicates the scores for various indicators such as air quality, temperature, and humidity. , , For the corresponding weights, For target devices with scores below the set threshold, they are added to the inspection queue for subsequent manual inspection.

[0031] In another embodiment, the present invention proposes a system for data center analysis and positioning using UWB-RFID tags, used to implement the method for data center analysis and positioning using UWB-RFID tags as described in the foregoing embodiments, including: UWB-RFID tags are installed on every target device to be inspected in the data center computer room; The UWB-RFID receiver corresponds to the UWB-RFID tag configured on the target device and receives positioning information based on the UWB-RFID tag. Sensors are installed in various areas of the computer room; The acquisition unit is used to acquire continuous images of the target equipment from different angles during the inspection of the computer room; The processing unit is used to filter out label images from the acquired continuous images, retrieve image blocks from the filtered label images using a two-dimensional normal probability density function, input the image blocks into a convolutional neural network to identify and classify the label images, and assign the classified label images to each target device. The evaluation unit is used to determine the location of the target device based on the positioning information received by the UWB-RFID receiver, and to collect sensor data from the area closest to the target device to detect and score the environment around the target device.

[0032] In this system, the functions and workflows of each component are the same as those of the steps in the aforementioned embodiments, so they will not be repeated here.

[0033] In the embodiments disclosed in this application, a computer storage medium may be a tangible medium that may contain or store programs for use by or in conjunction with an instruction execution system, apparatus, or device. The computer storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of computer storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CDROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0034] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0035] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A method for analysis and positioning in a data center using UWB-RFID tags, characterized in that, Includes the following steps: Step 1: Configure UWB-RFID tags for each target device to be inspected in the data center server room, divide the server room into areas, and configure sensors in each area; Step 2: During the inspection of the computer room, collect continuous images of the target equipment from different angles and filter out the labeled images; Step 3: Use a two-dimensional normal probability density function to retrieve image patches from the filtered label images, input the image patches into a convolutional neural network to identify and classify the label images, and assign the classified label images to each target device; Step 4: Determine the location of the target device based on the positioning information received by the UWB-RFID receiver corresponding to the UWB-RFID tag configured on the target device, and collect sensor data from the area closest to the target device to detect and score the environment around the target device.

2. A method for analysis and positioning of a data center using UWB-RFID tags as described in claim 1, characterized in that: In step 1, the UWB-RFID tag transmits data according to the IEEE 802.15.3a protocol.

3. The method for analysis and positioning of a data center using UWB-RFID tags as described in claim 1, characterized in that: In step 2, for continuous images of the target device from different angles, images whose angles and proportions are within a preset range are selected and sorted according to the time axis.

4. The method for analysis and positioning of a data center using UWB-RFID tags as described in claim 3, characterized in that: In step 3, a two-dimensional normal probability density function is used to retrieve image patches located near the image center from the filtered label images. The two-dimensional normal probability density function is as follows: In the formula, Represents the probability density of an image patch. Indicates the center of the image. Indicates the center of the image patch. The covariance matrix represents the pixel vectors of an image patch. These are adjustable parameters; The image patches are sorted according to their probability density, and the image patch at the top of the sort is selected as the image patch near the image center.

5. A method for analysis and positioning of a data center using UWB-RFID tags as described in claim 1, characterized in that: In step 4, during the process of determining the location of the target device, the cumulative probability distribution function is used to calculate the probability distribution of the positioning error, and positioning information with a probability greater than a set threshold is eliminated.

6. A method for analysis and positioning of a data center using UWB-RFID tags as described in claim 5, characterized in that: In step 4, the sensor data includes air quality, temperature, and humidity. The environment in which the target device is located is scored by combining the data from various sensors, and the target devices with scores below a set threshold are added to the inspection queue.

7. A method for analysis and positioning of a data center using UWB-RFID tags as described in claim 6, characterized in that: The process of comprehensively evaluating the environment in which the target device is located by integrating data from various sensors is as follows: First, the corresponding indicators represented by various sensor data are scored, and then the scores of various indicators are combined using normalization to obtain the total score of the environment in which the target device is located.

8. A system for analysis and positioning of a data center using UWB-RFID tags, characterized in that, include: UWB-RFID tags are installed on every target device to be inspected in the data center computer room; The UWB-RFID receiver corresponds to the UWB-RFID tag configured on the target device and receives positioning information based on the UWB-RFID tag. Sensors are installed in various areas of the computer room; The acquisition unit is used to acquire continuous images of the target equipment from different angles during the inspection of the computer room; The processing unit is used to filter out label images from the acquired continuous images, retrieve image blocks from the filtered label images using a two-dimensional normal probability density function, input the image blocks into a convolutional neural network to identify and classify the label images, and assign the classified label images to each target device. The evaluation unit is used to determine the location of the target device based on the positioning information received by the UWB-RFID receiver, and to collect sensor data from the area closest to the target device to detect and score the environment around the target device.

Images