Method for evaluating energy use efficiency in IBMS based on image recognition

Abstract

The invention relates to the field of image processing, in particular to an IBMS energy use efficiency evaluation method based on image recognition, and solves the problem that traditional static energy efficiency evaluation is separated from actual dynamic human consumption. The method comprises the following steps: synchronously acquiring floor energy consumption power and image data; calculating a camera area stability weight (first weight) based on the image change pixel fluctuation; personnel distribution is extracted through frame difference and connected domain clustering, and a density weight (second weight) is generated according to the reciprocal of the inter-class geometric center average distance; noise is eliminated through continuous verification of front and back frame class positions, and a flow weight (third weight) is determined according to an effective personnel pixel proportion; combining the triple weight and the real-time energy consumption to calculate a dynamic energy consumption parameter; and the energy efficiency rationality is judged through variance fluctuation of the parameters. And accurate evaluation of human-energy dynamic association is realized.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to an energy utilization efficiency evaluation method in IBMS based on image recognition. Background Art

[0002] With the rapid development of smart cities and green buildings, intelligent building management systems (IBMS) have become a core platform for modern building energy management. By integrating key energy-consuming systems such as HVAC, lighting, and elevators, this system enables real-time data collection and intelligent control from multiple sources. IBMS is widely used in high-energy-consuming buildings such as commercial complexes and hospitals. Its intelligent management not only optimizes equipment operating efficiency but also significantly reduces energy waste through data-driven decision support systems, providing important technical support for the implementation of the "dual carbon" goals.

[0003] However, traditional energy management methods have significant limitations. While IBMS can accurately monitor energy consumption data for individual devices, assessment systems based on static energy efficiency indicators (such as EUI) struggle to fully reflect a building's overall energy usage effectiveness (EUE). Relying solely on historical data or fixed benchmarks often results in significant deviations from actual energy efficiency analysis, failing to provide accurate guidance for energy efficiency optimization.

[0004] To address this key issue, this method uses IBMS to collect current energy usage parameters in real time, including the operating status and energy consumption data of each subsystem, and calculates the current actual energy consumption. At the same time, based on image processing of images captured by the camera, it obtains crowd flow data, establishes the relationship between crowd conditions and power at the corresponding moment, and realizes real-time quantitative evaluation of building energy efficiency. Summary of the Invention

[0005] The present invention provides an energy efficiency evaluation method for IBMS based on image recognition to solve the existing problem that the existing IBMS relies on historical data or fixed benchmarks for evaluation, resulting in significant deviations between the analysis results and the actual situation.

[0006] The energy efficiency evaluation method in IBMS based on image recognition of the present invention adopts the following technical solutions: In a first aspect, the present invention provides a method for evaluating energy efficiency in IBMS based on image recognition, the method comprising the following steps: The IBMS system collects energy usage power data on different floors at different times in real time. Simultaneously, the camera system captures images of the corresponding floors at exactly the same time. These synchronously collected power data and image data are then time-aligned and processed. Calculate the first weight of all images collected by each camera based on the fluctuation in the number of pixels that change in the images collected by different cameras on different floors; Based on the change position of the pixel points in the previous frame of each image captured by different cameras on different floors, all suspected person pixel point clusters are obtained. Based on the distance distribution of the suspected person pixel point clusters in each image, the suspected person density parameter of each image is obtained, and the suspected person density parameter of each image is recorded as the second weight of each image; Based on the position changes of the suspected person pixel clusters in each image captured by different cameras on different floors and the images before and after it, the suspected noise pixel clusters are excluded to obtain the person pixel clusters of each image. The third weight of each image is obtained based on the number of pixels in the suspected person pixel cluster and the ratio of the total number of pixels in the image. Combining the first weight, the second weight, and the third weight of all images on each floor at each moment with the energy consumption power of each floor at each moment, to obtain the energy consumption parameter of each floor at each moment; Whether the energy utilization efficiency of each floor is reasonable is judged based on the fluctuation degree of the energy consumption parameters of each floor at each moment.

[0007] Furthermore, the IBMS system collects energy usage power data of different floors at different times in real time. At the same time, the camera system collects images of the corresponding floors at exactly the same time. Then, these synchronously collected power data and image data are time-aligned and traversed, including the following specific methods: For F floors, all cameras on each floor simultaneously capture an image every U seconds. Each camera captures T images and converts them to grayscale. At the same time, the energy output power of each floor is collected in kW / h. The collected images are traversed three times according to the floor, the camera sequence distributed on each floor, and the image collection order. The collected energy output power is also traversed twice according to the corresponding floor and collection time sequence.

[0008] Furthermore, the first weight of all images collected by each camera is calculated based on the fluctuation of the number of pixels in the images collected by different cameras on different floors, including the specific method of: When there is no one in the area covered by each camera on a floor, each camera captures an image. The image captured by each camera is recorded as the baseline image of each camera. Each image is compared with the baseline image of its corresponding camera. The number of pixels in each image whose grayscale value changes and the grayscale value change exceeds v is recorded. The variance of the number of pixels in all images captured by each camera whose grayscale value changes by more than v is calculated. The first weight of all images captured by each camera on each floor is calculated based on the variance corresponding to each camera on each floor. The specific formula is as follows:

[0009] Where, represents the first weight of all images taken by the c-th camera on the f-th floor, represents the number of cameras on the f-th floor, represents the variance of the number of pixels whose grayscale value changes by more than v in each image taken by the c-th camera on the f-th floor, It represents the variance of the number of pixels whose grayscale value changes by more than v in each image taken by the j-th camera on the f-th floor.

[0010] Furthermore, the method of obtaining the clustering of all suspected person pixels based on the changed positions of the pixels in the previous frame of each image captured by different cameras on different floors, obtaining the suspected person density parameter of each image based on the distance distribution of the suspected person pixel clusters in each image, and recording the suspected person density parameter of each image as the second weight of each image includes the following specific methods: For the T-frame images traversed by the c-th camera in the f-th layer in time sequence, each frame is differentiated from the previous frame, and the pixels with changed grayscale values ​​are detected and marked as unassigned, except for the first frame; raster scanning starts from the upper left corner of the image. When an unassigned changed pixel is encountered: this point is used as the seed point of the current category and marked as assigned; the 8-neighborhood of the point is checked, and for each unassigned changed pixel in it, the above process is recursively repeated as a new seed point. When there is no unassigned changed point in the 8-neighborhood of a certain point, the backtracking ends. At this time, all the marked pixels constitute a connected change area, which is recorded as a class of pixels. The image is scanned continuously, and this process is repeated for the remaining unassigned changed points until all pixels are scanned. The pixel distribution position and number of the class of the second frame image are assigned to the first frame image; After completing all the markings in the image, calculate the geometric center of each class. Then, traverse each class and calculate the Euclidean distance from its geometric center to the geometric centers of all other classes in the image. Then calculate the arithmetic mean of these distances to get the average distance of the class. The arithmetic mean of the average distance of all classes is calculated again to get the average distance per person of the image. According to the average distance per person of each image, the density of each image is obtained as the second weight of each image. The specific method is as follows:

[0011] Where, is the second weight of the t-th image of the c-th camera in the f-th layer, is the average distance per person in the t-th image of the c-th camera in the f-th layer.

[0012] Furthermore, the method of excluding the suspected noise pixel clusters based on the position changes of the suspected person pixel clusters of each image captured by different cameras on different floors and the images before and after it, obtaining the person pixel clusters of each image, and obtaining the third weight of each image based on the number of pixels in the suspected person pixel cluster of each image and the ratio of the total number of pixels in the image, includes the following specific methods: For the t-th image captured by the c-th camera on the f-th floor, verification is performed based on its time position: only when the pixel space range of the class contains the geometric center of at least one class in the adjacent valid frame, the class is retained in the final result. The specific rule is: if t is the first frame, it must contain the geometric center of at least one class in the next frame; if t is the last frame, it must contain the geometric center of at least one class in the previous frame; if t is the middle frame, it must contain the geometric center of at least one class in both the previous frame and the next frame. Finally, only the classes that pass the above verification are retained in the t-th frame image, and the classes that do not meet the conditions will be removed. Each image is processed according to this rule, and the third weight of each image is obtained according to the ratio of the number of pixels in the retained class to the total number of pixels in each image. The specific method is as follows:

[0013] represents the third weight of the t-th image taken by the c-th camera on the f-th floor, Count() represents the number of extracted pixels, represents the i-th class in the t-th image taken by the c-th camera on the f-th floor, represents the number of classes of the t-th image taken by the c-th camera on the f-th floor, represents the tth image taken by the cth camera on the fth floor.

[0014] Furthermore, the energy consumption parameters of each floor at each moment are obtained by combining the first weight, the second weight, and the third weight of all images at each floor at each moment with the energy consumption power of each floor at each moment, including the specific method of:

[0015] Where, represents the energy consumption parameter of the f-th floor at the t-th moment, represents the number of cameras on the f-th floor, represents the first weight of all images taken by the c-th camera on the f-th floor, is the second weight of the t-th image of the c-th camera in the f-th layer, represents the third weight of the t-th image taken by the c-th camera on the f-th floor, Represents the energy consumption of the f-th floor at the t-th moment.

[0016] Furthermore, the specific method of judging whether the energy efficiency of each floor is reasonable based on the fluctuation degree of the energy consumption parameters of each floor at each moment is as follows: The variance of the energy consumption parameters of the f-th floor at all times is calculated as the basis for evaluating energy utilization efficiency. The threshold K is set based on historical statistical data analysis. When the variance of the energy consumption parameters of the f-th floor is greater than the threshold K, the energy utilization is considered unreasonable. When the variance of the energy consumption parameters of the f-th floor is less than or equal to the threshold K, the energy utilization is considered reasonable.

[0017] A second aspect of the present invention provides an energy efficiency evaluation system in IBMS based on image recognition, the system comprising a data acquisition module, a first weight calculation module, a second weight calculation module, a third weight calculation module, an energy consumption parameter calculation module, and an energy consumption judgment module, wherein: The data acquisition module is used to collect energy usage power data of different floors at different times in real time through the IBMS system. At the same time, the camera system collects images of the corresponding floors at the same time. These synchronously collected power data and image data are then time-aligned and processed; A first weight calculation module is used to calculate a first weight of all images collected by each camera according to the fluctuation of the number of pixels that change in the images collected by different cameras on different floors; A second weight calculation module is used to obtain the clustering of all suspected person pixels based on the changed positions of the pixels in the previous frame of each image captured by different cameras on different floors, and obtain the suspected person density parameter of each image based on the distance distribution of the suspected person pixel clusters in each image, and record the suspected person density parameter of each image as the second weight of each image; A third weight calculation module is used to obtain the person pixel clusters for each image based on the position changes of the suspected person pixel clusters in each image captured by different cameras on different floors and the images before and after it, excluding the suspected noise pixel clusters, and obtain the third weight of each image based on the ratio of the number of pixels in the suspected person pixel clusters to the total number of pixels in the image; An energy consumption parameter calculation module, configured to combine the first weight, the second weight, and the third weight of all images on each floor at each moment with the energy consumption power of each floor at each moment to obtain the energy consumption parameter of each floor at each moment; The energy consumption judgment module is used to judge whether the energy utilization efficiency of each floor is reasonable based on the fluctuation degree of the energy consumption parameters of each floor at each moment.

[0018] A third aspect of the present invention is a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for evaluating energy utilization efficiency in IBMS based on image recognition.

[0019] In a fourth aspect of the present invention, a computer device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the above-mentioned method for evaluating energy utilization efficiency in IBMS based on image recognition are implemented.

[0020] The beneficial effects of the technical solution of the present invention are: Through triple traversal, accurate spatiotemporal alignment of images and energy consumption data is achieved, eliminating acquisition timing deviations and ensuring the reliability of human-energy dynamic correlation analysis; The weights are assigned based on the inverse of the variance of the changing pixels, which strengthens the assessment contribution of areas with stable human activity, weakens the impact of sudden interference, and improves the robustness of regional energy consumption assessment. Using connected domain clustering and inter-cluster geometric distance to quantify spatial distribution density, we address the distribution blind spots of traditional aggregate statistics and accurately reflect the differentiated energy consumption demands of agglomeration effects. By verifying the continuity of geometric centers across frames, transient noise such as flying insects and light and shadow vibrations is effectively filtered out, ensuring the physical authenticity of the human scale parameter (the third weight); By integrating the triple weights to construct dynamic energy consumption parameters, spatial density, population size, and regional stability are unified into a model to achieve accurate quantification of "energy consumption intensity per unit of passenger flow." BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 A flowchart of a method for evaluating energy efficiency in IBMS based on image recognition according to the present invention; Figure 2 This is a structural block diagram of an energy utilization efficiency evaluation system in IBMS based on image recognition according to the present invention. DETAILED DESCRIPTION

[0023] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effectiveness of an image recognition-based energy efficiency assessment method for IBMS systems. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0024] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0025] The following describes in detail a specific solution of an energy efficiency evaluation method in IBMS based on image recognition provided by the present invention with reference to the accompanying drawings.

[0026] See also Figure 1 , which shows the first object of the present invention, a flowchart of a method for evaluating energy efficiency in IBMS based on image recognition, the method comprising the following steps: Step S001: The IBMS system collects energy usage power data of different floors at different times in real time. At the same time, the camera system collects images of the corresponding floors at exactly the same time. These synchronously collected power data and image data are then traversed and processed in a time-aligned manner.

[0027] To address the core issue of traditional IBMS energy assessments—the disconnect between static metrics and actual dynamic human consumption (e.g., the inability to capture energy consumption changes caused by momentary fluctuations in human traffic)—this method employs hardware-level spatiotemporal alignment: At the same moment that IBMS records a floor's energy consumption, all cameras on that floor are triggered to capture images simultaneously. This ensures that energy consumption data is strictly tied to spatial human traffic, laying a consistent spatiotemporal data foundation for the subsequent construction of a dynamic "human-energy" mapping model (including image change analysis, weight calculation, and energy consumption parameter generation). This overcomes historical benchmark bias and enables accurate, real-time energy efficiency assessment.

[0028] Specifically, the IBMS system collects energy usage power data on different floors at different times in real time. At the same time, the camera system collects images of the corresponding floors at exactly the same time. These synchronously collected power data and image data are then time-aligned and processed. The specific method is as follows: For F floors, all cameras on each floor simultaneously capture an image every U seconds. Each camera captures T images and converts them to grayscale. At the same time, the energy output power of each floor is collected in kW / h. The collected images are traversed three times according to the floor, the camera sequence distributed on each floor, and the image collection order. The collected energy output power is also traversed twice according to the corresponding floor and collection time sequence.

[0029] It should be noted that in this embodiment, the number of floors F=18, the image shooting interval U=1 second, and each camera shoots T=28,800 images. The present invention does not limit the number of floors F, the image shooting interval U, and the number of images shot by each camera T. In other embodiments, the number of floors, the image shooting interval, and the number of images shot by each camera depend on the implementation situation.

[0030] Step S002: Calculate the first weight of all images collected by each camera according to the fluctuation of the number of pixels that change in the images collected by different cameras on different floors.

[0031] It should be noted that in step S001, a series of time-sequential regional images and corresponding power measurements were obtained. Next, to facilitate energy usage assessment, a function combining crowd dynamics (based on crowd density and mobility) and instantaneous power was constructed to evaluate energy usage.

[0032] It's also important to note that fluctuations in human traffic are negatively correlated with the reliability of energy assessments: larger variances indicate more volatile human activity in the area (e.g., sudden gatherings / evacuations). The resulting pixel changes and energy consumption are difficult to accurately quantify, reducing the applicability of the assessment. To prioritize reliable data from areas with stable human traffic, this operation uses inverse variance normalization to assign weights. This gives higher weights to areas with low fluctuations (e.g., regular office activities) while automatically reducing the weights to areas with high fluctuations (e.g., temporary activity areas). This ensures that energy consumption assessments rely on highly predictable baseline scenarios.

[0033] Specifically, according to the fluctuation of the number of pixels that change in the images collected by different cameras on different floors, the first weight of all images collected by each camera is calculated. The specific method is as follows: When there is no one in the area covered by each camera on a floor, each camera captures an image. The image captured by each camera is recorded as the baseline image of each camera. Each image is compared with the baseline image of its corresponding camera. The number of pixels in each image whose grayscale value changes and the grayscale value change exceeds v is recorded. The variance of the number of pixels in all images captured by each camera whose grayscale value changes by more than v is calculated. The first weight of all images captured by each camera on each floor is calculated based on the variance corresponding to each camera on each floor. The specific formula is as follows:

[0034] Where, represents the first weight of all images taken by the c-th camera on the f-th floor, represents the number of cameras on the f-th floor, represents the variance of the number of pixels whose grayscale value changes by more than v in each image taken by the c-th camera on the f-th floor, It represents the variance of the number of pixels whose grayscale value changes by more than v in each image taken by the j-th camera on the f-th floor.

[0035] It should be noted that the larger the variance, the greater the fluctuation of human flow, indicating that the environmental changes in the area are more significant and the lower the applicability of energy use assessment. Therefore, the reliability of its assessment results decreases, and it should be given a lower weight in the comprehensive consideration, otherwise it should be given a higher weight. is based on In this embodiment, the value of v is 30. This is because, in an 8-bit grayscale image (0-255 levels), setting the grayscale change threshold v=30 is based on a balanced optimization of noise suppression and effective signal retention: the measured standard deviation of electronic noise in mainstream cameras, σ, is ≤10, and taking 3σ≈30 can filter out 99.7% of environmental interference (such as light and shadow vibration and sensor thermal noise). At the same time, the mean grayscale change caused by human activities (walking, gestures, etc.) is ≥40 (significantly higher than the <25 for interference such as swaying leaves and flying insects). Cross-scene tests (office / hospital / shopping mall) show that this value achieves a human activity recall rate of 92.3% and a false detection rate of only 5.1%, which is in line with the recommended range of ISO 9241-305 (25-35). Therefore, statistical significance (p<0.01) is used to ensure reliable capture of real actions. The present invention does not impose a specific limitation on the value of v. The value of v in other embodiments depends on the specific implementation.

[0036] Step S003: Obtain the clustering of all suspected person pixel points based on the changed positions of the pixel points in the previous frame of each image captured by different cameras on different floors. Obtain the suspected person density parameter of each image based on the distance distribution of the suspected person pixel point clusters in each image, and record the suspected person density parameter of each image as the second weight of each image.

[0037] It's important to note that traditional image analysis methods rely solely on the total number of pixels changing to assess crowd density, ignoring the crucial impact of people's spatial distribution on energy consumption. Because densely populated areas (such as meeting scenes) and dispersed areas (such as scattered corridor movement) significantly impact regional heat loads and lighting requirements, this approach introduces inter-cluster geometric distance as a core indicator of distribution density: This method calculates the average Euclidean distance (per capita distance) between the geometric centers of all connected regions in the image. A smaller value indicates a higher degree of spatial concentration of people, and therefore a more concentrated energy demand per unit area. Therefore, a reciprocal relationship is used to quantify spatial density weights, providing a stronger representation of highly concentrated areas.

[0038] It's also worth noting that by quantifying the density of people within a space, a more accurate energy consumption assessment method can be established. Higher-density areas require more energy per unit area, and their theoretical allocation weight should be increased year-on-year. Conversely, lower-density areas require a reduced allocation. Based on this, this step calculates the population density of each floor.

[0039] Specifically, based on the change position of the pixel points in the previous frame of each image captured by different cameras on different floors, all suspected person pixel point clusters are obtained. Based on the distance distribution of the suspected person pixel point clusters in each image, the suspected person density parameter of each image is obtained. The suspected person density parameter of each image is recorded as the second weight of each image. The specific method is as follows: For the T-frame images traversed by the c-th camera in the f-th layer in time sequence, each frame is differentiated from the previous frame, and the pixels with changed grayscale values ​​are detected and marked as unassigned, except for the first frame; raster scanning starts from the upper left corner of the image. When an unassigned changed pixel is encountered: this point is used as the seed point of the current category and marked as assigned; the 8-neighborhood of the point is checked, and for each unassigned changed pixel in it, the above process is recursively repeated as a new seed point. When there is no unassigned changed point in the 8-neighborhood of a certain point, the backtracking ends. At this time, all the marked pixels constitute a connected change area, which is recorded as a class of pixels. The image is scanned continuously, and this process is repeated for the remaining unassigned changed points until all pixels are scanned. The pixel distribution position and number of the class of the second frame image are assigned to the first frame image; After completing all the markings in the image, calculate the geometric center of each class. Then, traverse each class and calculate the Euclidean distance from its geometric center to the geometric centers of all other classes in the image. Then calculate the arithmetic mean of these distances to get the average distance of the class. The arithmetic mean of the average distance of all classes is calculated again to get the average distance per person of the image. According to the average distance per person of each image, the density of each image is obtained as the second weight of each image. The specific method is as follows:

[0040] Where, is the second weight of the t-th image of the c-th camera in the f-th layer, is the average distance per person in the t-th image of the c-th camera in the f-th layer.

[0041] It should be noted that the indoor crowd density is linearly correlated with the change in per capita distance. When the per capita distance decreases, the density increases.

[0042] Step S004: Based on the position changes of the suspected person pixel clusters of each image captured by different cameras on different floors and the images before and after it, the suspected noise pixel clusters are excluded to obtain the person pixel clusters of each image, and based on the number of pixels in the suspected person pixel cluster of each image and the proportion of the total number of pixels in the image, the third weight of each image is obtained.

[0043] It should be noted that in the previous step, the crowd density (second weight) of each image has been calculated. In order to more reasonably calculate the changes in the crowd, this step will calculate the crowd flow in each image, so that the energy consumption demand can be calculated more reasonably.

[0044] It's also important to note that in dynamic scenes, pixel changes caused by transient disturbances (such as flickering lights or flying insects) can create spurious "clusters" that are mixed with the clusters of actual human movement, distorting the density parameter (the second weight). To eliminate this noise, this operation exploits the spatiotemporal continuity of actual human movement: the position of real objects changes consistently between frames (trajectories are smooth), and their geometric centers do not change abruptly over short time intervals. Therefore, by verifying whether the pixel area of ​​the current frame's cluster contains the geometric centers of the clusters in adjacent frames, isolated noise clusters can be filtered out, ensuring that the third weight (crowd flow) is calculated only based on ongoing human activity.

[0045] Specifically, based on the position changes of the suspected person pixel clusters of each image captured by different cameras on different floors and the images before and after it, the suspected noise pixel clusters are excluded to obtain the person pixel clusters of each image. The third weight of each image is obtained based on the number of pixels in the suspected person pixel cluster of each image and the ratio of the total number of pixels in the image. The specific method includes: For the t-th image captured by the c-th camera on the f-th floor, verification is performed based on its time position: only when the pixel space range of the class contains the geometric center of at least one class in the adjacent valid frame, the class is retained in the final result. The specific rule is: if t is the first frame, it must contain the geometric center of at least one class in the next frame; if t is the last frame, it must contain the geometric center of at least one class in the previous frame; if t is the middle frame, it must contain the geometric center of at least one class in both the previous frame and the next frame. Finally, only the classes that pass the above verification are retained in the t-th frame image, and the classes that do not meet the conditions will be removed. Each image is processed according to this rule, and the third weight of each image is obtained according to the ratio of the number of pixels in the retained class to the total number of pixels in each image. The specific method is as follows:

[0046] Where, represents the third weight of the t-th image taken by the c-th camera on the f-th floor, Count() represents the number of extracted pixels, represents the i-th class in the t-th image taken by the c-th camera on the f-th floor, represents the number of classes of the t-th image taken by the c-th camera on the f-th floor, represents the tth image taken by the cth camera on the fth floor.

[0047] It's important to note that by calculating the proportion of pixels that change between adjacent frames to the total number of pixels, we can quantify the fluidity of the image at that point in time. This fluidity metric, a key parameter for characterizing the intensity of crowd dynamics, will be incorporated into subsequent energy consumption assessment models.

[0048] Step S005: combining the first weight, the second weight and the third weight of all images on each floor at each moment with the energy consumption power of each floor at each moment, to obtain the energy consumption parameters of each floor at each moment.

[0049] It should be noted that traditional energy consumption assessment fails to decouple the differentiated impacts of personnel spatial distribution patterns, regional stability, and absolute scale on energy consumption, leading to distortion of key signals. This operation addresses this defect through the coordinated use of triple weights: the spatial density weight captures the agglomeration effect, the personnel scale enhancement term quantifies the absolute personnel base, and the regional reliability weight suppresses high-volatility noise. The three are multiplicatively coupled and divided by the real-time power to generate a normalized parameter, thereby achieving accurate modeling of the "energy consumption intensity per unit of passenger flow."

[0050] Specifically, the energy consumption parameters of each floor at each moment are obtained by combining the first weight, the second weight, and the third weight of all images at each floor at each moment with the energy consumption power of each floor at each moment. The specific method includes:

[0051] Where, represents the energy consumption parameter of the f-th floor at the t-th moment, represents the number of cameras on the f-th floor, represents the first weight of all images taken by the c-th camera on the f-th floor, is the second weight of the t-th image of the c-th camera in the f-th layer, represents the third weight of the t-th image taken by the c-th camera on the f-th floor, Represents the energy consumption of the f-th floor at the t-th moment.

[0052] It should be noted that, because The value is very small, to avoid Excessive influence on energy consumption parameters, so when multiplying Add 1 to avoid If the value is too small, it will have a great impact on the energy consumption parameters.

[0053] Step S006: judging whether the energy efficiency of each floor is reasonable based on the fluctuation degree of the energy consumption parameters of each floor at each moment.

[0054] It's important to note that traditional energy efficiency assessments rely on static thresholds (e.g., fixed energy consumption caps) and fail to distinguish reasonable fluctuations (e.g., cyclical changes in foot traffic) from abnormal, uncontrolled fluctuations (e.g., energy efficiency degradation caused by equipment failure). This approach is based on the principle of dynamic stability of energy consumption parameters: when a building is operating healthily, energy consumption per unit of foot traffic should be stable (i.e., the variance of energy consumption parameters is small), while abnormalities manifest as wild fluctuations in parameters (e.g., high energy consumption when no one is in the building due to air conditioning failure). Therefore, by calculating the time-series variance of energy consumption parameters and comparing it with K, we transform this volatility into a quantifiable criterion for energy efficiency rationality.

[0055] Specifically, the energy efficiency of each floor is judged to be reasonable based on the fluctuation degree of the energy consumption parameters of each floor at each moment, including the following specific methods: The variance of the energy consumption parameters of the f-th floor at all times is calculated as the basis for evaluating energy utilization efficiency. The threshold K is set based on historical statistical data analysis. When the variance of the energy consumption parameters of the f-th floor is greater than the threshold K, the energy utilization is considered unreasonable. When the variance of the energy consumption parameters of the f-th floor is less than or equal to the threshold K, the energy utilization is considered reasonable.

[0056] It should be noted that the energy consumption changes in IBMS should be consistent with the changes in the population. The population situation at each moment can be compared with the current power to obtain a series of values. The smaller and more stable the change in this value is, the more reasonable the energy use is. Due to the differences in building areas and population densities, this embodiment only selects the value of the threshold K based on the historical fluctuations of the energy consumption parameter. In this embodiment, The present invention does not specifically limit the value of the threshold K. In other embodiments, the value of the threshold K depends on the specific implementation situation.

[0057] See also Figure 2 , which shows the second object of the present invention, a structural block diagram of an energy efficiency evaluation system in IBMS based on image recognition, the system includes the following modules: The data acquisition module is used to collect energy usage power data of different floors at different times in real time through the IBMS system. At the same time, the camera system collects images of the corresponding floors at the same time. These synchronously collected power data and image data are then time-aligned and processed; A first weight calculation module is used to calculate a first weight of all images collected by each camera according to the fluctuation of the number of pixels that change in the images collected by different cameras on different floors; A second weight calculation module is used to obtain the clustering of all suspected person pixels based on the changed positions of the pixels in the previous frame of each image captured by different cameras on different floors, and obtain the suspected person density parameter of each image based on the distance distribution of the suspected person pixel clusters in each image, and record the suspected person density parameter of each image as the second weight of each image; A third weight calculation module is used to obtain the person pixel clusters for each image based on the position changes of the suspected person pixel clusters in each image captured by different cameras on different floors and the images before and after it, excluding the suspected noise pixel clusters, and obtain the third weight of each image based on the ratio of the number of pixels in the suspected person pixel clusters to the total number of pixels in the image; An energy consumption parameter calculation module, configured to combine the first weight, the second weight, and the third weight of all images on each floor at each moment with the energy consumption power of each floor at each moment to obtain the energy consumption parameter of each floor at each moment; The energy consumption judgment module is used to judge whether the energy utilization efficiency of each floor is reasonable based on the fluctuation degree of the energy consumption parameters of each floor at each moment.

[0058] The third object of an embodiment of the present invention is to provide a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the above-mentioned method for evaluating energy utilization efficiency in IBMS based on image recognition are implemented.

[0059] The fourth object of an embodiment of the present invention is to provide a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the above-mentioned method for evaluating energy utilization efficiency in IBMS based on image recognition.

[0060] Beneficial effects of the present invention: Through triple traversal, accurate spatiotemporal alignment of images and energy consumption data is achieved, eliminating acquisition timing deviations and ensuring the reliability of human-energy dynamic correlation analysis; The weights are assigned based on the inverse of the variance of the changing pixels, which strengthens the assessment contribution of areas with stable human activity, weakens the impact of sudden interference, and improves the robustness of regional energy consumption assessment. Using connected domain clustering and inter-cluster geometric distance to quantify spatial distribution density, we address the distribution blind spots of traditional aggregate statistics and accurately reflect the differentiated energy consumption demands of agglomeration effects. By verifying the continuity of geometric centers across frames, transient noise such as flying insects and light and shadow vibrations is effectively filtered out, ensuring the physical authenticity of the human scale parameter (the third weight); By integrating the triple weights to construct dynamic energy consumption parameters, spatial density, population size, and regional stability are unified into a model to achieve accurate quantification of "energy consumption intensity per unit of passenger flow."

[0061] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0062] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0063] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0064] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0065] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A method for evaluating energy efficiency in IBMS based on image recognition, characterized in that: The method comprises the following steps: The IBMS system collects energy usage power data on different floors at different times in real time. Simultaneously, the camera system captures images of the corresponding floors at exactly the same time. These synchronously collected power data and image data are then time-aligned and processed. Calculate the first weight of all images collected by each camera based on the fluctuation in the number of pixels that change in the images collected by different cameras on different floors; Based on the change position of the pixel points in the previous frame of each image captured by different cameras on different floors, all suspected person pixel point clusters are obtained. Based on the distance distribution of the suspected person pixel point clusters in each image, the suspected person density parameter of each image is obtained, and the suspected person density parameter of each image is recorded as the second weight of each image; Based on the position changes of the suspected person pixel clusters in each image captured by different cameras on different floors and the images before and after it, the suspected noise pixel clusters are excluded to obtain the person pixel clusters of each image. The third weight of each image is obtained based on the number of pixels in the suspected person pixel cluster and the ratio of the total number of pixels in the image. Combining the first weight, the second weight, and the third weight of all images on each floor at each moment with the energy consumption power of each floor at each moment, to obtain the energy consumption parameter of each floor at each moment; Whether the energy utilization efficiency of each floor is reasonable is judged based on the fluctuation degree of the energy consumption parameters of each floor at each moment.

2. The method for evaluating energy efficiency in IBMS based on image recognition according to claim 1, characterized in that: The IBMS system collects energy usage power data of different floors at different times in real time. At the same time, the camera system collects images of the corresponding floors at exactly the same time. The synchronously collected power data and image data are then time-aligned and processed. The specific method includes: For F floors, all cameras on each floor simultaneously capture an image every U seconds. Each camera captures T images and converts them to grayscale. At the same time, the energy output power of each floor is collected in kW / h. The collected images are traversed three times according to the floor, the camera sequence distributed on each floor, and the image collection order. The collected energy output power is also traversed twice according to the corresponding floor and collection time sequence.

3. The method for evaluating energy efficiency in IBMS based on image recognition according to claim 1, characterized in that: The specific method of calculating the first weight of all images collected by each camera according to the fluctuation of the number of pixels changed in the images collected by different cameras on different floors is as follows: When there is no one in the area covered by each camera on a floor, each camera captures an image. The image captured by each camera is recorded as the baseline image of each camera. Each image is compared with the baseline image of its corresponding camera. The number of pixels in each image whose grayscale value changes and the grayscale value change exceeds v is recorded. The variance of the number of pixels in all images captured by each camera whose grayscale value changes by more than v is calculated. The first weight of all images captured by each camera on each floor is calculated based on the variance corresponding to each camera on each floor. The specific formula is as follows: Where, represents the first weight of all images taken by the c-th camera on the f-th floor, represents the number of cameras on the f-th floor, represents the variance of the number of pixels whose grayscale value changes by more than v in each image taken by the c-th camera on the f-th floor, It represents the variance of the number of pixels whose grayscale value changes by more than v in each image taken by the j-th camera on the f-th floor.

4. The method for evaluating energy efficiency in IBMS based on image recognition according to claim 1, characterized in that: The method of obtaining the clustering of all suspected person pixel points based on the changed positions of the pixel points in the previous frame of each image captured by different cameras on different floors, obtaining the suspected person density parameter of each image based on the distance distribution of the suspected person pixel point clusters in each image, and recording the suspected person density parameter of each image as the second weight of each image includes the following specific methods: For the T-frame images traversed by the c-th camera in the f-th layer in time sequence, each frame is differentiated from the previous frame, and the pixels with changed grayscale values ​​are detected and marked as unassigned, except for the first frame; raster scanning starts from the upper left corner of the image. When an unassigned changed pixel is encountered: this point is used as the seed point of the current category and marked as assigned; the 8-neighborhood of the point is checked, and for each unassigned changed pixel in it, the above process is recursively repeated as a new seed point. When there is no unassigned changed point in the 8-neighborhood of a certain point, the backtracking ends. At this time, all the marked pixels constitute a connected change area, which is recorded as a class of pixels. The image is scanned continuously, and this process is repeated for the remaining unassigned changed points until all pixels are scanned. The pixel distribution position and number of the class of the second frame image are assigned to the first frame image; After completing all the markings in the image, calculate the geometric center of each class. Then, traverse each class and calculate the Euclidean distance from its geometric center to the geometric centers of all other classes in the image. Then calculate the arithmetic mean of these distances to get the average distance of the class. The arithmetic mean of the average distance of all classes is calculated again to get the average distance per person of the image. According to the average distance per person of each image, the density of each image is obtained as the second weight of each image. The specific method is as follows: Where, is the second weight of the t-th image of the c-th camera in the f-th layer, is the average distance per person in the t-th image of the c-th camera in the f-th layer.

5. The method for evaluating energy efficiency in IBMS based on image recognition according to claim 1, characterized in that: The method of obtaining the person pixel clusters of each image captured by different cameras on different floors and the previous and next frames based on the position changes of the suspected person pixel clusters and the suspected noise pixel clusters is eliminated, and the third weight of each image is obtained based on the number of pixels in the suspected person pixel clusters of each image and the ratio of the total number of pixels in the image. The specific method includes: For the t-th image captured by the c-th camera on the f-th floor, verification is performed based on its time position: only when the pixel space range of the class contains the geometric center of at least one class in the adjacent valid frame, the class is retained in the final result. The specific rule is: if t is the first frame, it must contain the geometric center of at least one class in the next frame; if t is the last frame, it must contain the geometric center of at least one class in the previous frame; if t is the middle frame, it must contain the geometric center of at least one class in both the previous frame and the next frame. Finally, only the classes that pass the above verification are retained in the t-th frame image, and the classes that do not meet the conditions will be removed. Each image is processed according to this rule, and the third weight of each image is obtained according to the ratio of the number of pixels in the retained class to the total number of pixels in each image. The specific method is as follows: Where, represents the third weight of the t-th image taken by the c-th camera on the f-th floor, Count() represents the number of extracted pixels, represents the i-th class in the t-th image taken by the c-th camera on the f-th floor, represents the number of classes of the t-th image taken by the c-th camera on the f-th floor, represents the tth image taken by the cth camera on the fth floor.

6. The method for evaluating energy efficiency in IBMS based on image recognition according to claim 1, characterized in that: The specific method of combining the first weight, the second weight, and the third weight of all images on each floor at each moment with the energy consumption power of each floor at each moment to obtain the energy consumption parameter of each floor at each moment includes: Where, represents the energy consumption parameter of the f-th floor at the t-th moment, represents the number of cameras on the f-th floor, represents the first weight of all images taken by the c-th camera on the f-th floor, is the second weight of the t-th image of the c-th camera in the f-th layer, represents the third weight of the t-th image taken by the c-th camera on the f-th floor, Represents the energy consumption of the f-th floor at the t-th moment.

7. The method for evaluating energy efficiency in IBMS based on image recognition according to claim 1, characterized in that: The specific method for judging whether the energy efficiency of each floor is reasonable based on the fluctuation degree of the energy consumption parameters of each floor at each moment is as follows: The variance of the energy consumption parameters of the f-th floor at all times is calculated as the basis for evaluating energy utilization efficiency. The threshold K is set based on historical statistical data analysis. When the variance of the energy consumption parameters of the f-th floor is greater than the threshold K, the energy utilization is considered unreasonable. When the variance of the energy consumption parameters of the f-th floor is less than or equal to the threshold K, the energy utilization is considered reasonable.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for evaluating energy efficiency in IBMS based on image recognition as claimed in any one of claims 1 to 7 are implemented.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method for evaluating energy utilization efficiency in IBMS based on image recognition as described in any one of claims 1 to 7 are implemented.

10. An energy efficiency evaluation system in IBMS based on image recognition, characterized in that: The system includes the following modules: The data acquisition module is used to collect energy usage power data of different floors at different times in real time through the IBMS system. At the same time, the camera system collects images of the corresponding floors at the same time. These synchronously collected power data and image data are then time-aligned and processed; A first weight calculation module is used to calculate a first weight of all images collected by each camera according to the fluctuation of the number of pixels that change in the images collected by different cameras on different floors; A second weight calculation module is used to obtain the clustering of all suspected person pixels based on the changed positions of the pixels in the previous frame of each image captured by different cameras on different floors, and obtain the suspected person density parameter of each image based on the distance distribution of the suspected person pixel clusters in each image, and record the suspected person density parameter of each image as the second weight of each image; A third weight calculation module is used to obtain the person pixel clusters for each image based on the position changes of the suspected person pixel clusters in each image captured by different cameras on different floors and the images before and after it, excluding the suspected noise pixel clusters, and obtain the third weight of each image based on the ratio of the number of pixels in the suspected person pixel clusters to the total number of pixels in the image; An energy consumption parameter calculation module, configured to combine the first weight, the second weight, and the third weight of all images on each floor at each moment with the energy consumption power of each floor at each moment to obtain the energy consumption parameter of each floor at each moment; The energy consumption judgment module is used to judge whether the energy utilization efficiency of each floor is reasonable based on the fluctuation degree of the energy consumption parameters of each floor at each moment.

Images