Building Internet of Things equipment deployment and control management method

Abstract

The invention relates to an Internet of Things equipment deployment and control management method, in particular to a building Internet of Things equipment deployment and control management method. Comprising the following steps: deploying various sensors including an infrared sensor, a camera and an air quality sensor in a building, and collecting personnel flow and environment change data in real time; through data analysis and scene classification, the data are converted into scenes including office peaks, night modes and emergency evacuation of the building; training a scene recognition model by using a random forest classifier; a time sequence prediction model is adopted to predict scene change in a future short period, and a corresponding equipment control instruction is generated; the method comprises the following steps: establishing an incidence matrix of use frequency, spatial position and function interaction between devices by using sensor data, and identifying a device cooperation mode; dynamically generating an equipment deployment scheme through a graph theory algorithm based on the incidence matrix; issuing the generated deployment scheme to each equipment control node; and performing fine adjustment according to real-time data feedback.

Description

Technical Field

[0001] The present invention relates to a method for controlling the deployment of Internet of Things devices, and more particularly to a method for controlling the deployment of Internet of Things devices in buildings. Background Art

[0002] The Internet of Things-based smart building management method and system disclosed in Chinese invention patent 2023117576686, although the patent proposes a management method based on finite element model, temperature field data analysis and fatigue damage prediction, its technical means have improved the accuracy of building management to a certain extent, but there are still some shortcomings and disadvantages, especially when dealing with the actual deployment and management of building Internet of Things equipment, there are several problems that need to be solved and optimized.

[0003] First, this method emphasizes the acquisition of environmental parameters and temperature field data through IoT sensors for stress analysis, and the use of finite element models and bidirectional threshold loop networks for damage prediction and building remaining life assessment. The main problem with this method is that it over-relies on physical state data such as structural data and temperature stress distribution, while ignoring the impact of dynamic factors such as device logical state and task execution on device operation. In the actual management of smart buildings, the operating state of equipment is not only affected by physical factors (such as temperature, current, vibration, etc.), but also includes the performance of equipment in specific tasks (such as task completion time, system load, fault frequency, etc.). The current patent does not fully consider these logical state data, resulting in its health status assessment focusing on physical damage prediction and failing to fully cover the complexity of equipment operation. For example, even if some equipment is in good physical state, it may cause equipment failure or performance degradation due to task overload or unreasonable scheduling in the logical state. Therefore, the accuracy and flexibility of building equipment control management relying solely on physical models are low.

[0004] Secondly, the method described in the patent mainly focuses on the health management and fatigue damage prediction of building structures, and ignores the optimization of the collaborative relationship between devices. IoT devices in modern smart buildings are often the result of the collaborative work of multiple devices, such as air-conditioning systems, lighting systems, security monitoring and smart elevators. The collaborative optimization of these devices is one of the keys to ensuring building energy efficiency and user experience. However, the technical means of this patent are limited to the stress and damage analysis of a single device, and do not involve the dynamic collaborative optimization mechanism between devices. For example, in the operation of building equipment, the collaborative efficiency of air conditioning and lighting systems directly affects the overall energy consumption of the building, and this collaborative efficiency needs to be optimized through the combined analysis of the physical and logical states between devices. Therefore, this method has certain shortcomings in dealing with complex equipment collaboration modes and does not provide an effective solution for multi-device collaborative management.

[0005] In addition, the technical means of this patent are mainly based on finite element models for temperature stress analysis. Although fatigue damage prediction of building structures can be achieved, its core relies on structural design data and temperature field simulation. This method lacks real-time performance and flexibility in dealing with dynamic environments. Building management is not a single structural health issue, but more often it involves dynamic deployment, real-time monitoring and optimized scheduling of equipment. Therefore, the application scope of simple finite element models in building equipment deployment management is limited, especially when faced with real-time data processing and decision-making, traditional model optimization and temperature stress analysis methods have lags and cannot adapt to rapidly changing environmental requirements. Summary of the invention

[0006] The purpose of the present invention is to provide a building Internet of Things equipment deployment control management method, so as to solve some of the disadvantages pointed out in the background technology.

[0007] The present invention solves the above-mentioned technical problems by adopting the following technical solution, which comprises the following steps:

[0008] S1. Intelligent deployment and scheduling based on scenario prediction:

[0009] S1.1. Deploy multiple sensors including infrared sensors, cameras, and air quality sensors in buildings, and collect data on personnel flow and environmental changes in real time;

[0010] S1.2. Through data analysis and scene classification, the data is converted into building scenes including office peak, night mode, and emergency evacuation; and the scene recognition model is trained using a random forest classifier;

[0011] S1.3, using the time series prediction model to predict the scene changes in the short term in the future and generate corresponding equipment control instructions;

[0012] S2. Dynamic deployment based on the correlation between devices:

[0013] S2.1. Using sensor data, establish a correlation matrix of usage frequency, spatial location, and functional interaction between devices to identify device collaboration patterns;

[0014] S2.2 dynamically generates equipment deployment plans based on the association matrix through graph theory algorithms;

[0015] S2.3 sends the generated deployment plan to each device control node and makes fine adjustments based on real-time data feedback;

[0016] S3, fusion of physical and logical states:

[0017] S3.1. Collect physical state data of the equipment including temperature, current, and voltage, and logical state data including task completion rate and response time through the sensor network;

[0018] S3.2. Use data fusion algorithm to analyze physical status and logical status data uniformly to generate comprehensive health status indicators of equipment;

[0019] S3.3, detect abnormal modes in equipment operation in real time through anomaly detection and issue early warning signals;

[0020] S4. Real-time load adjustment based on equipment status:

[0021] S4.1. Collect real-time load data of the equipment, including the number of processing tasks, CPU utilization, and power consumption, and establish a load monitoring system;

[0022] S4.2. Use the health assessment algorithm to analyze the current health status of the equipment and predict the load change trend in the short term;

[0023] S4.3. When the equipment load approaches the warning threshold, the transfer plan is calculated, the equipment is selected to take over part of the load, and the task allocation adjustment is performed.

[0024] Furthermore, the training scenario recognition model method includes: quantifying the collaboration intensity between devices by comprehensively evaluating the device usage frequency, spatial location and functional interactivity; using sensor data to collect the usage behavior, spatial distribution and interaction characteristics of the device to form the basic data of the association matrix; defining the association intensity calculation formula:

[0025]

[0026] Among them: a ij represents the association strength between device i and device j, which is an element in the association matrix and is used to quantify the closeness of device collaboration; T is the total time period for evaluating device collaboration, and the dynamic changes of device collaboration at different times are captured through time integration; α1, α2, α3 are weight coefficients, which respectively adjust the contribution of joint use frequency, spatial proximity and functional interactivity to the association strength; f u (t) is the joint frequency function, reflecting the synchronous working frequency of devices i and j within time t; g s (x i ,x j ) is the spatial proximity function, based on the spatial positions x of devices i and j i and x j , used to evaluate the impact of physical distance on device collaboration; h f (t,i,j) is the functional interactivity function, which describes the functional complementarity and collaboration of devices at a specific time and situation.

[0027] Furthermore, the training scenario recognition model method includes: identifying the device collaboration mode through graph analysis, clustering algorithm or network analysis; further mining the deep relationship of device collaboration through mathematical models; defining a device collaboration mode recognition function:

[0028]

[0029] Where: M is the global measurement value of the device collaboration mode, which is used to evaluate the efficiency and stability of the overall collaboration of the device group; n is the total number of devices, representing the scale of the association matrix; φ(d ij ) is the distance influence function, d ij represents the physical distance between devices i and j, reflecting the weakening effect of distance on device collaboration; exp(-β·φ(d ij )) represents the attenuation function, and the attenuation factor β regulates the effect of physical distance on the intensity of collaboration; is the second-order derivative function of the association strength, which evaluates the dynamic stability of the equipment collaboration relationship and captures the sudden changes or fluctuations in the relationship between equipment; the integral term It is a global analysis of the changes in collaboration within the space V where the devices are located.

[0030] Furthermore, the training scenario recognition model method includes: further optimizing the equipment deployment strategy based on the identified equipment collaboration mode; adjusting the operating parameters, working sequence and physical location of the equipment through the intelligent scheduling algorithm to maximize the collaboration efficiency between the equipment, and defining the deployment optimization function:

[0031]

[0032] in: is the overall measurement function of control optimization, which evaluates the comprehensive performance of the control strategy in terms of resource balance and conflict resolution. σ(t) is the resource utilization balance function, which measures the rationality of resource allocation within time t. Indicates the rate of change of resource balance, combined with an exponential decay factor Controls the smoothness of resource adjustment. ρ(t) is the device conflict resolution rate function, which describes the efficiency of devices in resolving resource conflicts in high-frequency collaboration.

[0033] is the rate of change of the conflict resolution rate, combined with the exponential decay factor Adjust the response speed of conflict resolution and smoothly adjust the collaborative state of the equipment; γ1, γ2 are optimization weight coefficients used to adjust the system's emphasis on resource balance and conflict resolution; λ1, λ2 are response adjustment factors that control the sensitivity of the control strategy to balance adjustment and conflict resolution.

[0034] Furthermore, the method for generating a comprehensive health status indicator of a device includes:

[0035] Preprocess the physical state data including temperature, current, and wear of the equipment and the logical state data including task completion time, system load, and failure rate; extract key features reflecting the health of the equipment; the features of the physical state data include the equipment temperature fluctuation rate, average current, and equipment wear rate, while the features of the logical state data involve the completion time of task execution, system load pressure, and failure frequency; define the physical state feature extraction formula as:

[0036]

[0037] in: It represents the comprehensive value of the physical status characteristics, quantifies the overall physical health level of the equipment, and serves as the basis for subsequent health status assessment; T(t) is the equipment temperature, It indicates the rate of change of temperature over time t, reflecting the temperature fluctuation of the equipment during operation; I avg (t) is the average current of the device, which indicates the current consumption during the operation of the device; μ(t) is the wear rate of the device, which measures the degree of physical loss of the device over time t; α1, α2, α3 are weight coefficients.

[0038] Furthermore, the method for generating a comprehensive health status indicator of a device includes:

[0039] The physical state data and the logical state data are integrated and analyzed through the data fusion algorithm. The physical state reflects the hardware operation status of the device, while the logical state focuses more on the task execution capability and load status of the device. By fusing the two types of data, potential problems of the device are identified. The data fusion formula for the physical state and the logical state is defined as:

[0040]

[0041] in: It represents the fused health status characteristics of the equipment, integrating the characteristic information of the physical state and the logical state; β1 and β2 are the fusion weight coefficients, which respectively adjust the relative contribution of the physical state and the logical state in the comprehensive evaluation; It is a physical status feature, which indicates the hardware operating status of the device, including the fusion value of temperature, current, and wear data; It is a logical state feature, which contains key data of equipment task execution, including task completion time τ(t), fault occurrence frequency f fail (t) and system load L sys (t); It is the exponential decay function of the system load on the logic state, which controls the influence of the system load on the logic characteristics. γ is the adjustment parameter, which is used to adjust the attenuation amplitude of the load on the logic state.

[0042] Further, the method for generating a comprehensive health status indicator of a device includes: a health status indicator method for quantifying the overall health of the device; combining data at the logical level of task execution efficiency and fault frequency of the device to reflect the actual operating status of the device;

[0043] The formula for defining the comprehensive health status index is:

[0044]

[0045] in: It is a comprehensive health status indicator used to evaluate the health status of the equipment at time t; it is used to reflect the health risk of the equipment; The fused health characteristic value combines the key data of physical state and logical state to provide an assessment of the comprehensive health status of the equipment; fail (t) is the failure frequency, reflecting the failure risk of the equipment within time t; λ1, λ2 are control weights, which are used to adjust the impact of health characteristic value and failure frequency on the comprehensive health status respectively. The size of the weight depends on the system's focus on health status and failure risk.

[0046] The building Internet of Things equipment deployment and management method of the present invention has multiple significant beneficial effects:

[0047] First, the present invention uses data fusion technology to combine physical status data with logical status data to comprehensively evaluate the operating status of the equipment. This fusion not only considers the hardware operating conditions of the equipment (such as temperature, current, wear, etc.), but also combines logical-level data such as task completion time, system load, and fault occurrence frequency, thereby improving the accuracy of equipment health assessment. Compared with traditional analysis methods based only on single physical data, the present invention can identify potential problems of equipment earlier, improve prediction and early warning capabilities, and help building managers take preventive maintenance measures before equipment failures occur, thereby reducing downtime and equipment damage.

[0048] By establishing a correlation matrix between devices, using graph analysis and data mining technology, we can identify device collaboration patterns and optimize device operation and scheduling strategies. This technology makes collaboration between multiple devices more efficient, avoids waste of resources and conflicts between devices, and maximizes energy efficiency. For example, the collaborative management of air conditioning and lighting systems can automatically adjust the operating mode according to the real-time environment and usage, ensuring the optimal allocation of device energy efficiency in different usage scenarios.

[0049] Thirdly, the present invention can adjust the operating parameters, working sequence and physical location of the equipment in real time through the intelligent scheduling algorithm. This means that the scheduling of the equipment can be dynamically adjusted according to the health status indicators and predicted load changes, ensuring the rational use of resources while reducing the occurrence of equipment overload and resource conflicts. This dynamic adjustment mechanism enables the building Internet of Things system to make intelligent decisions based on actual conditions, improve the flexibility and response speed of the system, and optimize the overall operation of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a flow chart of the building Internet of Things equipment deployment and management method of the present invention.

[0051] Figure 2 This is a flow chart of the method for training a scene recognition model of the present invention.

[0052] Figure 3 A flow chart of a method for generating a comprehensive health status indicator of a device according to the present invention. DETAILED DESCRIPTION

[0053] The specific implementation modes of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0054] Combined with Figure 1 The core of the building IoT device control and management method of the present invention is to achieve efficient management and scheduling of equipment in the building through intelligent technical means. The specific process can be implemented through the following steps. First, in S1 intelligent control and scheduling based on scene prediction, the scheduling of equipment is based on intelligent identification and prediction of different scenes to ensure that the equipment can automatically adjust the working mode in different building usage scenarios, thereby improving energy efficiency and user experience.

[0055] S1.1 deploys a variety of sensor devices including infrared sensors, cameras, air quality sensors, etc. These sensors are installed in key locations of the building and can collect real-time personnel flow data, environmental light, temperature changes, air quality and other information. The collected data will serve as the basic input for subsequent scene prediction. These sensors continuously transmit real-time data to the system, allowing the system to understand the dynamic situation in the building in real time, so as to have the ability to adjust in real time. Then, in the S1.2 stage, the system will analyze the collected environmental data and personnel flow data and classify the scenes. Through data mining technology, the system can identify different building usage scenarios, such as office peak, night mode and emergency evacuation. Each scene has its unique environmental characteristics and personnel behavior patterns. In order to achieve automatic classification of scenes, the system uses random forest classifiers for model training. Random forest is an ensemble learning algorithm that can improve classification accuracy by integrating multiple decision tree models. It is suitable for handling complex and diverse scene classification tasks in buildings. Random forest automatically identifies the scene of the building by analyzing the multi-dimensional characteristics of environmental and personnel data, and improves the accuracy and robustness of classification as the amount of data continues to increase.

[0056] On this basis, S1.3 uses a time series prediction model, and the system can predict the scene changes in the building in the short term in the future. The time series model will predict the scene changes that may occur in the building at a certain moment in the future based on the trend of historical data, such as when the office peak will come and when an emergency evacuation may occur. Based on the prediction results, the system generates corresponding equipment control instructions, which can control the lighting, air conditioning, ventilation and other equipment in the building, so that it can automatically adjust its working status according to different scene requirements. For example, during peak office hours, lighting and air conditioning will be enhanced, and in night mode, the equipment will enter energy-saving state.

[0057] In S2, dynamic deployment based on the correlation between devices aims to dynamically optimize the deployment of devices by analyzing the collaboration mode and functional correlation between devices, so as to improve the overall efficiency and resource utilization of the system. S2.1 By using sensor data, the system can comprehensively monitor the usage frequency, spatial location and functional interaction of each device. Sensors collect various data in real time during the operation of devices, including the frequency of device use, their relative position in space, and the functional interaction between them, such as whether some devices are working together or relying on each other. Based on this data, the system can construct an association matrix, in which each element reflects the strength of association between devices. The association matrix is ​​the basis for analyzing the collaboration mode of devices, which quantifies the collaborative relationship between devices in terms of function and location. Devices with higher correlation tend to have higher usage frequency, or they are physically closer and have interdependent characteristics in terms of function. S2.2 After the association matrix is ​​established, the system uses graph theory algorithms to dynamically generate device deployment plans. Graph theory algorithms construct an optimal device deployment graph by analyzing the correlation between devices. In this graph, the edges between device nodes represent the collaborative relationship between them. The graph theory algorithm generates the optimal device deployment strategy through operations such as the shortest path and maximum flow, ensuring that the devices achieve the highest collaboration efficiency in terms of physical location and functional distribution. The system dynamically adjusts the location and distribution of devices to reduce unnecessary communication and energy waste between devices, thereby improving the overall system efficiency.

[0058] The device deployment plan generated by S2.3 will be sent to the control node of each device through the network. After receiving the deployment plan, the control node will adjust the working mode and physical location of the device to make it consistent with the optimized plan. In order to ensure the effectiveness of the deployment plan in actual application, the system will also make fine adjustments based on real-time data feedback. Sensors continuously monitor the operating status of the device and feed back the newly collected data to the system. The system iteratively optimizes the deployment strategy of the device through this feedback data, and further fine-tunes the location, usage mode, etc. of the device.

[0059] S3.1 First, through the sensor network, the system collects the physical state data and logical state data of the device in real time. The physical state data includes parameters such as the temperature, current, and voltage of the device. These data reflect the operation of the device at the physical level. For example, temperature changes can indicate whether the device is overheated, and current and voltage data can indicate whether the device is under normal workload. The logical state data includes the task completion rate and response time of the device. The logical state is mainly used to evaluate the task execution capability and performance of the device. The task completion rate can reflect the efficiency of the device in processing tasks, and the response time reflects the delay of the device when executing tasks. These data come from different levels. The physical data reflects the hardware status of the device, and the logical data reflects the performance of the device when performing specific tasks.

[0060] S3.2 Next, using the data fusion algorithm, the system conducts a unified analysis of these physical state and logical state data. Physical and logical data are essentially different types of data information, describing the operation of the device from the perspectives of hardware and task execution respectively. Through the data fusion algorithm, the system can unify and integrate these data to generate a comprehensive health status indicator. This health status indicator is obtained by comprehensively considering the physical performance and logical task performance of the device, and can comprehensively reflect the operating health of the device. The core of data fusion is to associate the relevant information in the physical and logical states. For example, some devices may have potential performance problems due to excessive task load when the physical state seems normal. By fusing physical and logical state data, the system can identify these potential fault hazards.

[0061] S3.3Finally, the system monitors the operating status of the equipment in real time through anomaly detection technology. By analyzing the health status indicators of the equipment, the anomaly detection algorithm can identify abnormal patterns in the operation of the equipment, such as abnormal temperature rise, abnormal task response time, or sudden current fluctuations. Once the system detects that the operating mode of the equipment deviates from the normal range, it will automatically issue an early warning signal to alert managers to possible equipment failures or performance degradation.

[0062] S4 is a real-time load adjustment based on device status. Its core goal is to monitor the load of the device in real time and dynamically adjust the workload of the device according to the health status of the device and the load change trend to ensure the stability and efficiency of the system. First, in S4.1, the system collects the load data of the device in real time through sensors and monitoring tools. These load data include the number of tasks currently processed by the device, CPU utilization, and power consumption. The number of processed tasks reflects the current workload of the device, the CPU utilization indicates the computing resource consumption of the device when processing tasks, and the power consumption is related to the power usage of the device. These data are continuously collected through the sensor network and uploaded to the system's load monitoring platform to build a complete load monitoring system. The system can comprehensively monitor the real-time load status of the device and provide basic data support for subsequent load adjustments.

[0063] Then in S4.2, the system analyzes the current health status of the equipment through the health assessment algorithm. The health assessment algorithm comprehensively considers the historical operating data and current load conditions of the equipment to generate a health status score for the equipment. Based on the current health status of the equipment, the system can also predict the load change trend of the equipment in the short term, and determine whether the equipment can continue to bear the current load in the future, or whether there will be a risk of performance degradation or even failure due to excessive load. The health assessment not only considers the current status of the equipment, but also combines the prediction model to analyze the trend of future loads, thereby helping the system make load adjustment decisions in advance. Finally, in S4.3, when the load of the equipment approaches the warning threshold, the system automatically calculates the load transfer plan.

[0064] The warning threshold refers to a critical point when the device is carrying a task load. When the load of the device approaches this point, continuing to increase the load may cause the device to overload or fail. At this time, based on the results of load monitoring and health assessment, the system will select devices with better health and lower load to take over some tasks to balance the load of the overall system. The load transfer plan includes the specific operation of transferring some tasks from high-load devices to other devices to ensure that the tasks can be smoothly transitioned while ensuring that the overall performance of the system is not affected.

[0065] Embodiment 1:

[0066] See also Figure 2 In this example, a large office building is equipped with a variety of IoT devices, including air conditioning systems, lighting systems, air quality monitors, security cameras, and elevator control systems. The management system hopes to optimize the collaborative management of these devices through scene recognition models, thereby achieving intelligent scheduling of various systems in the building to improve energy efficiency and reduce resource waste. The collaborative optimization of air conditioning and lighting systems will be discussed in detail.

[0067] It is assumed that each floor of the office building has 5 independent air conditioning units and 20 lighting units. The peak office hours are from 9 am to 5 pm, and the night mode is from 8 pm to 7 am. In order to simplify the problem, one day's office hours (24 hours) are selected as the evaluation period, that is, T = 24 hours. The system collects the usage behavior, spatial distribution and functional interaction characteristics of the equipment through sensor data to form the basic data of the association matrix.

[0068] The system evaluates the collaboration intensity between air conditioners and lighting devices by the frequency of use, spatial location, and functional interactivity of the devices. First, the collaboration intensity a between device i and device j is calculated using the following formula: ij :

[0069]

[0070] Among them: a ijrepresents the collaboration intensity between air conditioning equipment i and lighting equipment j; T is the evaluation time of 24 hours; α1, α2, α3 are weight coefficients, which control the influence of each feature on the collaboration intensity. The value range is α1, α2, α3∈[0.1, 1.0]; f u (t) is the frequency of joint use of air conditioning and lighting equipment, which is defined as the probability that the two devices are turned on at the same time in a certain period of time; g s (x i ,x j ) represents the spatial proximity of devices i and j, which is measured here by the distance between devices in the floor plan. The closer the distance, the greater the potential for collaboration; h f (t,i,j) describes the functional interactivity, reflecting whether the air conditioner and lighting equipment are scheduled in the same scenario. The higher the functional complementarity, the larger the value.

[0071] The following data are measured by sensor data:

[0072] 1. Combined use frequency f u (t): The frequency of joint use of air conditioning and lighting equipment during peak office hours is 0.9 (i.e., both are turned on at the same time 90% of the time), and the frequency of joint use in night mode is 0.2 (i.e., both are turned on at the same time 20% of the time).

[0073] 2. Spatial proximity g s (x i ,x j ): Devices i and j are 10 meters apart, according to the set distance function g s (x i ,x j )=1 / d 2 , then the spatial proximity is g s (x i ,x j )=1 / 10 2 =0.01.

[0074] 3. Functional interactivity f (t,i,j): During peak hours, air conditioning and lighting equipment are activated together in the office environment, and the functional interactivity value is set to 0.8; in night mode, the functional interactivity value is 0.3, indicating that the use of air conditioning is more independent of lighting equipment.

[0075] The weight coefficients are set to α1 = 0.5, α2 = 0.3, and α3 = 0.2. Substitute the usage data of each time period into the formula to calculate the collaboration intensity during the office peak period (9:00-17:00) and the collaboration intensity during the night mode (20:00-7:00).

[0076] 1. Office peak hours (8 hours):

[0077]

[0078] 2. Night mode (11 hours):

[0079]

[0080] 3. Total collaboration intensity a ij :

[0081] a ij =4.904+1.793=6.697

[0082] Through the above calculation method, the cooperation intensity between each pair of air conditioners and lighting equipment can be obtained and stored in the association matrix. Each element a in the matrix ij Represents the degree of collaboration between two devices. After the matrix is ​​generated, it can be further used for device scheduling optimization. For example, devices with higher collaboration intensity can be assigned similar working hours and tasks to maximize energy efficiency; while devices with lower collaboration intensity can be managed separately to avoid resource waste. In this example, by quantitatively analyzing the frequency of device use, spatial proximity, and functional interactivity, the collaboration intensity a between air conditioners and lighting devices was calculated using the formula ij =6.697. The actual data was substituted into the calculation to demonstrate the feasibility of this method in the deployment and control of building IoT devices.

[0083] In order to further optimize the collaborative management of devices, a scene recognition model will be introduced to identify device collaboration patterns using graph analysis, clustering algorithms, or network analysis. Through mathematical model analysis, we can more deeply explore the deep-level collaborative relationships between devices and quantify the global effectiveness of the collaboration pattern.

[0084] Assume that there are a total of 10 air conditioning units and 30 lighting units in the office building, which are distributed on different floors and spatial areas. The sensor system has built an association matrix for these devices, reflecting their usage frequency, physical location, and functional interaction. In order to optimize the deployment and task allocation of devices, the system needs to identify the collaboration modes of the devices through the scene recognition model and evaluate the stability and efficiency of these modes.

[0085] Through graph analysis, each device can be regarded as a node, and the collaboration intensity between devices is ij As the edge weight between nodes. Based on this, the global metric function M of device collaboration pattern recognition is defined to evaluate the efficiency and stability of the overall collaboration of the device group. The calculation formula is as follows:

[0086]

[0087] Where: M is the global metric value, reflecting the efficiency and stability of the equipment collaboration mode; n is the total number of equipment, set to n = 40, indicating the total number of air conditioners and lighting equipment; φ(d ij ) is the distance influence function, d ij represents the physical distance between device i and device j. For devices that are far apart in an office building, the collaboration intensity usually decays, so the effect of distance needs to be considered; exp(-β·φ(d ij )) is the attenuation function, β is the attenuation factor, which adjusts the effect of physical distance on the intensity of collaboration. Setting β = 0.5 represents a moderate degree of distance attenuation; is the second-order derivative function of the association strength, which is used to capture the dynamic changes in the relationship between devices. The changes in the device collaboration relationship may be affected by the external environment, time or task type. This item is used to evaluate the possible mutations or fluctuations in the system; the integral term It is a global analysis of the equipment collaboration relationship within the entire space V. Space V represents the physical area of ​​the office building.

[0088] In order to perform actual calculations, it is assumed that the spatial location information and collaborative relationship data of the devices have been collected through the sensor system. The following is the relevant data:

[0089] 1. Equipment distance d ij : Set the average distance between air conditioning equipment and lighting equipment to 20 meters, and set the distance influence function Then φ(d ij )=1 / 20 2 =0.0025.

[0090] 2. Collaboration intensity a ij :Through the above calculation, the cooperation intensity between air conditioners and lighting equipment is a ij =6.697.

[0091] 3. Calculation of attenuation factor: Substitute into the attenuation function and calculate exp(-0.5·0.0025)≈0.99875.

[0092] Substitute this data into the device cooperation mode recognition function:

[0093]

[0094] By calculating M, we can get the global metric value of the device cooperation mode. Assuming that the integral term is approximated to be 0.01, then:

[0095] M=40×40×6.697×0.99875-0.01=10691.07-0.01=10691.06

[0096] According to the calculation results, the global metric value M = 10691.06 shows that the air conditioners and lighting equipment have high collaboration efficiency and stability under the current deployment scheme. Through this model, we can further analyze which devices collaborate most closely and which devices may have potential lack of collaboration. Since the distance factor is taken into account, the system can optimize the physical deployment plan of the equipment and bring devices that are far away but have frequent functional collaboration as close as possible, thereby improving collaboration efficiency.

[0097] Through the instantiation of the scene recognition model, we can use graph theory analysis and mathematical models to quantify the collaborative relationships of equipment in the building and further evaluate the stability and efficiency of these relationships. The calculation result M = 10691.06 proves that the system can optimize the deployment and scheduling of equipment by analyzing the collaborative mode.

[0098] Next, the system will further optimize the equipment deployment strategy based on these identified equipment collaboration modes to achieve reasonable resource allocation and effective conflict resolution. In order to enable the system to dynamically adjust the working sequence, operating parameters and physical location of the equipment during actual operation, the system adopts an intelligent scheduling algorithm and quantifies the scheduling effect through the following deployment optimization function.

[0099] Define the control optimization function as:

[0100]

[0101] in, It is the overall metric function for control optimization, which evaluates the comprehensive performance of the system in terms of resource balance and conflict resolution. In order to better understand this process, the air conditioning and lighting system in an office building is taken as an example to explain how to optimize the working parameters and scheduling strategies of the equipment through this control optimization function.

[0102] Set in an office building, the management system needs to optimize the management of 10 air conditioning units and 30 lighting units. The peak usage period is from 9:00 to 17:00 every day, and the off-peak period is from 17:00 to 9:00. The system hopes to ensure that the energy consumption of equipment during peak hours is balanced by optimizing resource allocation, while reducing conflicts during off-peak hours and improving the scheduling efficiency of equipment.

[0103] Define resource utilization balance function σ(t)

[0104] The system first defines the resource utilization balance function σ(t) to measure the resource utilization of air conditioning and lighting systems at different times. It is assumed that during peak hours, the resource utilization of air conditioning and lighting equipment tends to be balanced. Based on historical data, the system derives the resource utilization balance σ(t) = 0.85 during peak hours and σ(t) = 0.6 during off-peak hours, which represents the reasonable allocation of resources at different times. In order to control the smoothness of resource adjustment, the system introduces an exponential decay factor Set λ1=0.3 to mitigate drastic changes in resources and avoid excessive device switching.

[0105] Optimize resource utilization during computing peak hours:

[0106]

[0107] The calculation shows that during peak hours, the utilization of equipment resources is more balanced and the adjustment is smoother.

[0108] Define the device conflict resolution rate function ρ(t)

[0109] For resource conflicts that are prone to occur when devices collaborate at high frequencies, the system defines a device conflict resolution rate function ρ(t) to describe how the air conditioning and lighting systems resolve conflicts in collaboration. During peak hours, the conflict rate is high, and the conflict resolution efficiency ρ(t) = 0.5, while during off-peak hours, there are fewer conflicts and the resolution efficiency ρ(t) = 0.9. In order to smoothly adjust the response speed of conflict resolution, the system uses an exponential decay factor Set λ2=0.4 to adjust the sensitivity of device conflict handling.

[0110] Calculate the conflict resolution optimization effect during peak hours:

[0111]

[0112] This result shows that although there are higher conflicts during peak hours, the response speed of conflict resolution is relatively stable through optimized scheduling.

[0113] Substitute the above calculation results into the control optimization function and integrate the entire time period (24 hours a day), and the system can obtain the optimization effect of the overall control. Optimization evaluation is performed during peak hours (8 hours) and off-peak hours (16 hours):

[0114] 1. Peak hours (9:00-17:00):

[0115]

[0116] 2. Non-peak hours (17:00-9:00):

[0117] The resource balance and conflict resolution efficiency are set to be higher during off-peak hours, σ(t) = 0.6 and ρ(t) = 0.9 respectively. The optimization effect during off-peak hours is calculated:

[0118]

[0119] The final control optimization effect is:

[0120]

[0121] Step 4: Optimization analysis and adjustment

[0122] Calculated The results show that the system performs well in resource allocation and conflict resolution within 24 hours, especially during off-peak hours, where the system effectively reduces resource conflicts and improves balance through intelligent scheduling algorithms. During peak hours, although conflicts are high, the system achieves reasonable resource allocation by adjusting the working sequence and physical location of the equipment, ensuring the maximum collaborative efficiency of the equipment.

[0123] This example optimizes the function through deployment control The calculation shows the feasibility of the building IoT equipment deployment management method in practical applications. Combined with resource utilization balance and conflict resolution mechanism, the system can dynamically adjust the working mode of the equipment to ensure efficient collaboration and optimal resource allocation.

[0124] Embodiment 2:

[0125] Combination Figure 3 In this embodiment, in the application of the building Internet of Things equipment deployment and management method, generating comprehensive health status indicators of equipment is a key step to ensure stable equipment operation, predict potential failures, and optimize equipment maintenance. By analyzing the physical state data (such as temperature, current, wear) and logical state data (such as task completion time, system load, failure rate) of the equipment, the health characteristic indicators of the equipment can be obtained. This solution will combine actual data and demonstrate the specific method of generating comprehensive health status indicators of equipment through a detailed calculation process, and prove the feasibility of this method.

[0126] In a central air conditioning system in an office building, the management system wants to monitor the health status of the air conditioning equipment and predict its future maintenance needs. Sensor data is used to collect the physical status (temperature, current, wear rate) and logical status (task completion time, system load, and failure frequency) of the air conditioning equipment. The system needs to generate comprehensive health status indicators for the air conditioning equipment based on this data to ensure the stability of the equipment under high load conditions and prevent failures in advance.

[0127] First, the system collects the physical status data of the air conditioning equipment through the sensor network, including temperature fluctuations during equipment operation, current consumption, and equipment wear rate. Through the preprocessing step, the data is cleaned and standardized to ensure its accuracy and consistency. Then, the system extracts the physical status characteristics that reflect the health of the equipment according to the following formula.

[0128] The physical state feature extraction formula is defined as:

[0129]

[0130] in:

[0131] The comprehensive value representing the physical status characteristics quantifies the overall physical health level of the device. This value will be used for subsequent health status assessment; T(t) is the device temperature, is the rate of change of temperature, reflecting the temperature fluctuation of the equipment during operation; I avg (t) is the average current of the device, indicating the current consumption of the device during operation; μ(t) is the wear rate of the device, measuring the degree of physical loss of the device over time t; α1, α2, α3 are weight coefficients, which respectively control the impact of temperature fluctuation, current consumption and wear rate on the comprehensive health status, and the value range is set to α1, α2, α3∈[0.1, 1.0].

[0132] In addition to physical status data, the system also collects logical status data of air conditioning equipment, including task execution time (such as the time to complete the cooling process), system load (pressure during air conditioning operation), and fault frequency. The system pre-processes this data and extracts logical status features as another important indicator reflecting equipment performance.

[0133] Task completion time: indicates the average time it takes for the air conditioning equipment to complete a temperature adjustment task, which is set to 20 minutes;

[0134] System load: reflects the load pressure during the operation of the equipment, and the load of the equipment during peak hours is set to 70%;

[0135] Failure frequency: The number of failures that occur in air-conditioning equipment per month is set at 0.05 times / day.

[0136] The comprehensive physical health value of the air conditioning equipment can be calculated by the physical state feature extraction formula. The evaluation period T is set to 24 hours in a day, and the equipment temperature change rate collected by the system is Average current I avg (t) = 10 A , equipment wear rate μ(t)=0.001. The weight coefficients are set as α1=0.5, α2=0.3, and α3=0.2.

[0137] Calculate the physical health characteristics of air conditioning equipment

[0138]

[0139] The comprehensive value of the physical state Reflects the overall physical health level of the device. This value will be combined with the logical state data to further generate a comprehensive health status index. The system combines the physical state characteristics with the logical state characteristics to generate a comprehensive health status index for the device. Set the weight of the logical state characteristic value to 0.6 and the weight of the physical state characteristic value to 0.4. The system generates a comprehensive health status index according to the following formula:

[0140] Task completion time + 1 system load + 1 fault occurrence frequency

[0141] The task completion time is 20 minutes (converted to hours). ), the system load is 70% (converted to ), the failure frequency is 0.05 times / day (i.e. ).

[0142] Calculate the logical state eigenvalue:

[0143] Logical state characteristic = 3 + 1.43 + 20 = 24.43

[0144] Substitute into the comprehensive health status formula:

[0145]

[0146] According to the calculation results, the comprehensive health status index of the air conditioning equipment is This indicator combines the physical and logical status characteristics of the device, indicating that the overall health status of the device is in a relatively stable state. However, as the task execution time increases or the system load increases, the comprehensive health status indicator may decrease, so regular monitoring and maintenance are required.

[0147] In this office building, the central air conditioning system is responsible for cooling different areas. The system collects the physical status data (including temperature, current and equipment wear rate) and logical status data (including task completion time, system load, and fault frequency) of the air conditioner through a sensor network. By integrating this data, the system can generate a comprehensive health status indicator for each air conditioner to predict equipment maintenance needs and optimize operation.

[0148] Set the temperature fluctuation rate of a certain air conditioning device within 24 hours 0.02℃ / hour, average current I avg (t) = 9 A , wear rate μ(t) = 0.0015. The physical state feature has been calculated by the previous physical state feature extraction formula It is 72.

[0149] In terms of logical state data, the air conditioning system task completion time τ(t) is set to 18 minutes, and the system load L sys (t) = 0.75 (i.e. 75% load), fault occurrence frequency f fail (t) = 0.03 times / day. These logical data reflect the task execution and failure frequency of the air-conditioning equipment.

[0150] In order to integrate the data of physical state and logical state, the equipment health state feature fusion formula is defined as:

[0151]

[0152] in:

[0153] Indicates the fused device health status characteristics, which integrates the characteristic information of physical status and logical status;

[0154] β1, β2 are weight coefficients, which adjust the relative contribution of physical state and logical state in comprehensive evaluation. Set β1 = 0.6, β2 = 0.4; is the physical state characteristic value, set to 72;

[0155] is the logical state characteristic, calculated as follows:

[0156]

[0157] Substitute the data into:

[0158]

[0159] is the exponential decay function of the system load to the logic state, and γ = 0.5 is set, then:

[0160]

[0161] Now, we can substitute the above values ​​into the data fusion formula to calculate the comprehensive health status characteristics of the equipment:

[0162]

[0163] Calculate the contributions:

[0164] 0.6·72=43.2,0.4·38·0.687≈10.45

[0165] The comprehensive results are:

[0166]

[0167] Step 4: Analysis and Optimization

[0168] Calculation results Represents the comprehensive health status characteristic value of the air conditioning equipment within 24 hours. This value combines the physical state (temperature, current, wear) and logical state (task completion time, system load, failure rate) of the equipment to provide the system with a comprehensive view of the overall health of the equipment. If the value is high, it means that the equipment is running stably and both the physical and logical states are within the normal range. However, if the value continues to drop, especially due to excessive system load or longer task execution time, the system can issue an early warning to indicate possible equipment aging or impending failure.

[0169] This example demonstrates how to comprehensively evaluate the physical and logical status of a device and generate a comprehensive health status indicator for the device through the actual application of the data fusion formula. By substituting actual data into the formula for calculation, the comprehensive health characteristic value is obtained. Clearly reflects the operating status of the equipment.

[0170] In an office building, the central air conditioning system runs 24 hours a day. In order to evaluate the health of the air conditioning equipment, the system monitors its physical state (temperature, current, wear) and logical state (task completion time, system load, fault frequency) in real time through a sensor network. By integrating this data, the system generates the health characteristic value of the air conditioning equipment. Now, the system needs to further evaluate the overall health of the air conditioning equipment through a comprehensive health status indicator formula and predict its future operation risks.

[0171] To calculate the health status of the air conditioner, the system uses the following comprehensive health status indicator formula:

[0172]

[0173] in:

[0174] It is a comprehensive health status indicator used to evaluate the health status of the equipment at time t, reflecting the overall health level and health risk of the equipment;

[0175] It is the fused health characteristic value, which is obtained by combining the physical state and logical state data, reflecting the operating status of the device. Its value is set to This is calculated previously by fusion of physical and logical state data; f fail (t) is the failure frequency, which indicates the failure risk of the equipment in a unit of time. The historical failure frequency of the air conditioner is set to once a month, that is, f fail (t) = 0.033 times / day; λ1, λ2 are control weights used to adjust the impact of health characteristic values ​​and fault frequency on the comprehensive health status. Setting λ1 = 0.7 means that the system pays more attention to the health characteristics of the equipment, and λ2 = 0.3 means that the weight of the fault risk is relatively low.

[0176] In order to calculate the comprehensive health status index of the equipment within 24 hours, the known data is substituted into the formula. Assume that the time period of one day is T = 24 hours and the frequency of failure is f fail (t) = 0.033 times / day, the fused health characteristic value is The control weights are λ1=0.7 and λ2=0.3.

[0177] The calculation formula is:

[0178]

[0179] First calculate the value of each item:

[0180] 0.7 1287.6 = 901.32

[0181] 0.3 0.033 = 0.0099

[0182] So the formula becomes:

[0183]

[0184] The final calculation is:

[0185]

[0186] Calculation results Indicates the comprehensive health status indicator of the air conditioner within 24 hours. A high value of this indicator indicates that the air conditioner is in good health during the day's operation, the physical and logical states are at a stable level, and the possibility of failure is low. If this indicator gradually decreases in future monitoring, it may indicate that the health of the equipment is deteriorating and inspection or maintenance needs to be arranged.

[0187] The system can continuously monitor the health status of the air conditioning equipment through the health status indicator, keep track of the equipment's operating conditions at any time, and detect potential risks of the equipment in a timely manner. When the value is lower than a preset threshold, the system will trigger a maintenance plan or issue a fault warning to ensure that the equipment can operate in an efficient and stable state.

Claims

1. A method for controlling and managing the deployment of IoT devices in buildings, characterized in that The steps include: S1. Intelligent deployment and scheduling based on scenario prediction: S1.

1. Deploy multiple sensors including infrared sensors, cameras, and air quality sensors in buildings, and collect data on personnel flow and environmental changes in real time; S1.

2. Through data analysis and scene classification, the data is converted into building scenes including office peak, night mode, and emergency evacuation; and the scene recognition model is trained using a random forest classifier; S1.3, using the time series prediction model to predict the scene changes in the short term in the future and generate corresponding equipment control instructions; S2. Dynamic deployment based on the correlation between devices: S2.

1. Using sensor data, establish a correlation matrix of usage frequency, spatial location, and functional interaction between devices to identify device collaboration patterns; S2.2 dynamically generates equipment deployment plans based on the association matrix through graph theory algorithms; S2.3 sends the generated deployment plan to each device control node and makes fine adjustments based on real-time data feedback; S3, fusion of physical and logical states: S3.

1. Collect physical state data of the equipment including temperature, current, and voltage, and logical state data including task completion rate and response time through the sensor network; S3.

2. Use data fusion algorithm to analyze physical status and logical status data uniformly to generate comprehensive health status indicators of equipment; S3.3, detect abnormal modes in equipment operation in real time through anomaly detection and issue early warning signals; S4. Real-time load adjustment based on equipment status: S4.

1. Collect real-time load data of the equipment, including the number of processing tasks, CPU utilization, and power consumption, and establish a load monitoring system; S4.

2. Use the health assessment algorithm to analyze the current health status of the equipment and predict the load change trend in the short term; S4.

3. When the equipment load approaches the warning threshold, the transfer plan is calculated, the equipment is selected to take over part of the load, and the task allocation adjustment is performed.

2. The building Internet of Things equipment deployment and management method according to claim 1 is characterized in that The training scenario recognition model method includes: quantifying the collaboration intensity between devices by comprehensively evaluating the device usage frequency, spatial location and functional interactivity; using sensor data to collect the usage behavior, spatial distribution and interaction characteristics of the device to form basic data of the association matrix; defining the association intensity calculation formula: Among them: a ij represents the association strength between device i and device j, which is an element in the association matrix and is used to quantify the closeness of device collaboration; T is the total time period for evaluating device collaboration, and the dynamic changes of device collaboration at different times are captured through time integration; α1, α2, α3 are weight coefficients, which respectively adjust the contribution of joint use frequency, spatial proximity and functional interactivity to the association strength; f u (t) is the joint frequency function, reflecting the synchronous working frequency of devices i and j within time t; g s (x i ,x j ) is the spatial proximity function, based on the spatial positions x of devices i and j i and x j , used to evaluate the impact of physical distance on device collaboration; h f (t,i,j) is the functional interactivity function, which describes the functional complementarity and collaboration of devices at a specific time and situation.

3. The building Internet of Things equipment deployment and management method according to claim 1 is characterized in that The training scenario recognition model method includes: identifying the device collaboration mode through graph analysis, clustering algorithm or network analysis; and further mining the deep-level relationship of device collaboration through mathematical models.

4. The building Internet of Things equipment deployment and management method according to claim 1 is characterized in that The training scenario recognition model method includes: further optimizing the equipment deployment strategy based on the identified equipment collaboration mode; adjusting the equipment's operating parameters, working sequence and physical location through an intelligent scheduling algorithm to maximize the collaboration efficiency between the equipment, and defining a deployment optimization function: in: is the overall measurement function of control optimization, which evaluates the comprehensive performance of the control strategy in terms of resource balance and conflict resolution. σ(t) is the resource utilization balance function, which measures the rationality of resource allocation within time t. Indicates the rate of change of resource balance, combined with an exponential decay factor Controls the smoothness of resource adjustment. ρ(t) is the device conflict resolution rate function, which describes the efficiency of devices in resolving resource conflicts in high-frequency collaboration. is the rate of change of the conflict resolution rate, combined with the exponential decay factor Adjust the response speed of conflict resolution and smoothly adjust the collaborative state of the equipment; γ1, γ2 are optimization weight coefficients used to adjust the system's emphasis on resource balance and conflict resolution; λ1, λ2 are response adjustment factors that control the sensitivity of the control strategy to balance adjustment and conflict resolution.

5. The building Internet of Things equipment deployment and management method according to claim 1 is characterized in that The method for generating a comprehensive health status indicator of a device includes: The physical state data including temperature, current, and wear of the equipment and the logical state data including task completion time, system load, and failure rate are preprocessed to extract key features reflecting the health of the equipment; the features of the physical state data include the equipment temperature fluctuation rate, average current, and equipment wear rate, while the features of the logical state data involve the completion time of task execution, the load pressure of the system, and the frequency of failures.

6. The building Internet of Things equipment deployment and management method according to claim 5 is characterized in that The method for generating a comprehensive health status indicator of a device includes: The physical state data and the logical state data are integrated and analyzed through the data fusion algorithm. The physical state reflects the hardware operation status of the device, while the logical state focuses more on the task execution capability and load status of the device. By fusing the two types of data, potential problems of the device are identified. The data fusion formula for the physical state and the logical state is defined as: in: It represents the fused health status characteristics of the equipment, integrating the characteristic information of the physical state and the logical state; β1 and β2 are the fusion weight coefficients, which respectively adjust the relative contribution of the physical state and the logical state in the comprehensive evaluation; It is a physical status feature, which indicates the hardware operating status of the device, including the fusion value of temperature, current, and wear data; It is a logical state feature, which contains key data of equipment task execution, including task completion time τ(t), fault occurrence frequency f fail (t) and system load It is the exponential decay function of the system load on the logic state, which controls the influence of the system load on the logic characteristics. γ is the adjustment parameter, which is used to adjust the attenuation amplitude of the load on the logic state.

7. The building Internet of Things equipment deployment and management method according to claim 6 is characterized in that The method for generating a comprehensive health status index of a device includes: a health status index method for quantifying the overall health of the device; combining data at the task execution efficiency and fault frequency logic level of the device to reflect the actual operating status of the device.

Images