Building data intelligent analysis decision method and system based on Internet of Things

Abstract

The invention relates to the technical field of industrial data control, in particular to a building data intelligent analysis decision method and system based on the Internet of Things, and the method comprises the steps: obtaining the space parameters, glass material parameters and real-time environment data of each building; the method comprises the following steps of: constructing a building association map and generating a control task list by combining a sun position and a spatial geometrical relationship, decomposing each control task into a source side shielding sub-task and a receiving side compensation sub-task, distributing priority and setting a constraint condition; querying an equipment capability list and an equipment state incidence matrix of the building according to the subtasks and constraint conditions, and determining equipment control parameters; collecting equipment feedback data and environment change data, and adjusting equipment control parameters and / or regenerating a control scheme when an expected deviation exceeds a threshold value; recording manual intervention behaviors and environment parameters at the same time, and adjusting optimization parameters of the corresponding rooms; and counting a group intervention mode, and when the intervention proportion caused by the same optimization strategy exceeds a preset value, triggering global strategy regulation and control.

Description

Technical Field

[0001] This invention relates to the field of industrial data control technology, specifically to a building data intelligent analysis and decision-making method and system based on the Internet of Things. Background Technology

[0002] With the development of building intelligence technology, a large amount of operational data will be generated during operation. Effective analysis and control of this industrial data is the key to realizing intelligent building management and control.

[0003] Currently, the coordinated control of building equipment mainly relies on preset rules and strategies. This fixed-rules approach cannot adapt to dynamically changing environments and usage needs. When external conditions change, manual adjustments to rule parameters are required, resulting in high maintenance costs. Furthermore, the lack of effective data sharing and collaborative optimization between subsystems often leads to isolated information silos, resulting in low overall energy efficiency. Therefore, traditional methods lack the ability to predict future loads and equipment status, and can only passively respond to current conditions, leading to problems of response lag and over-adjustment. Summary of the Invention

[0004] The purpose of this invention is to provide a building data intelligent analysis and decision-making method and system based on the Internet of Things. Through cross-building impact identification, intelligent task decomposition and equipment collaborative linkage, combined with user behavior self-learning mechanism, a sustainable and optimized intelligent control system for building groups is formed.

[0005] To achieve the above objectives, the present invention provides the following technical solution: IoT-based intelligent analysis and decision-making methods for building data include: Acquire spatial parameters, glass material parameters, and real-time environmental data collected by IoT sensors for each building; By combining the sun's position and spatial geometry, a building association map is constructed to identify the impact transmission chain; A control task list is generated based on the building association map. Each control task is decomposed into a source-side occlusion sub-task and a receiver-side compensation sub-task. Priorities are assigned to the sub-tasks and constraints are set. Based on the sub-tasks and constraints, query the equipment capacity list and equipment status association matrix of this building, determine the equipment control parameters, and use sequential execution control for strongly coupled equipment and parallel execution control for weakly coupled equipment. Collect equipment execution feedback data and environmental change data. When the deviation from the expected value exceeds the threshold, adjust the equipment control parameters and / or regenerate the control scheme. Record manual intervention behaviors and environmental parameters at the same time, and adjust the optimization parameters of the corresponding room; statistically analyze group intervention patterns, and trigger global strategy adjustment when the intervention ratio caused by the same optimization strategy exceeds the preset value.

[0006] Preferably, the spatial parameters include floor height, building spacing, and three-dimensional coordinates; the building association map includes multiple floor nodes and directed influence edges between nodes, and the weight of the directed influence edge represents the influence intensity and influence time period of the source floor on the target floor; The steps to construct a building association map include: The spatial area of ​​each building is divided into grid cells, and nodes are created for each floor of each building. The node attributes include the floor center coordinates, glass area, and orientation angle. The solar altitude angle and azimuth angle are calculated at multiple time points within a preset time window. For each time point, the floor nodes with glass curtain walls are traversed as influence source nodes. The propagation direction of reflected light is calculated based on the solar incident angle and glass reflectivity. The target floor node reached by the reflected light is determined by ray tracing. When the reflected light energy reaching the target node exceeds a preset threshold, a directed edge is established between the influence source node and the target node. The weight of the directed edge is the calculated influence intensity. A set of directed graphs with time series is generated to form a building association map. The identification of the influence transmission chain includes: analyzing the multi-level transmission paths existing in the directed graph, and when the transmission relationship of the first node influencing the second node and the second node influencing the third node is identified, the second node is marked as a key transit node.

[0007] Preferably, the steps for generating a control task list based on the building association map include: Query the set of directed edges in the building association map where the influence intensity is greater than a preset threshold within a preset future time window; extract the source node, target node, influence intensity, and influence time period corresponding to each directed edge; obtain the number of affected rooms and historical user feedback records; calculate the task priority score based on the influence intensity, the number of affected rooms, and the historical user feedback records; sort the influence relationships according to the priority score to generate a task list containing the source node, target node, influence intensity, time window, and priority.

[0008] Preferably, the step of decomposing each control task into source-side shading sub-tasks and receiver-side compensation sub-tasks includes: for each task in the control task list, generating a source-side shading sub-task based on the type of the influencing source node, wherein the source-side shading sub-task points to the shading device and / or dimming glass of the influencing source building; generating a receiver-side compensation sub-task based on the type of the target node, wherein the receiver-side compensation sub-task points to the lighting system and air conditioning system of the affected building; setting a first constraint condition for the source-side shading sub-task, wherein the first constraint condition includes that the indoor illuminance of the influencing source building is not lower than the lower limit of comfortable illuminance; setting a second constraint condition for the receiver-side compensation sub-task, wherein the second constraint condition includes that the increase in lighting power does not exceed a preset proportion of the base power and a limit on the increase in air conditioning load; and assigning execution priorities to each sub-task based on the impact intensity and affected range of the task.

[0009] Preferably, the steps for determining equipment control parameters include: reading the equipment state correlation matrix, which records the influence intensity and influence delay time of the first equipment action on the state of the second equipment; extracting the equipment combinations involved according to the equipment control instructions to be executed, and calculating the correlation strength between the equipment; when the correlation strength between the equipment exceeds the strong coupling threshold, it is determined to be a strongly coupled equipment, and the control instructions of the strongly coupled equipment are arranged into sequential execution control, starting the master equipment first and monitoring its state changes, and starting the slave equipment after the master equipment state stabilizes; when the correlation strength between the equipment is lower than the weak coupling threshold, it is determined to be a weakly coupled equipment, and the control instructions of the weakly coupled equipment are arranged into parallel execution control, and control instructions are issued simultaneously; when the correlation strength between the equipment is between the strong coupling threshold and the weak coupling threshold, staggered execution control is adopted, and the execution time interval is set according to the influence delay time.

[0010] Preferably, the steps of adjusting equipment control parameters and / or regenerating control schemes include: collecting position sensor data of the shading device, illuminance sensor data of the lighting system, and temperature sensor data of the air conditioning system as equipment execution feedback data; collecting environmental sensor data of the affected area as environmental change data, wherein the environmental sensor data includes indoor illuminance, temperature, and user activity status; calculating the deviation between the equipment execution feedback data and the expected target value, and calculating the deviation between the environmental change data and the desired environmental parameters; when any deviation exceeds a preset threshold, determining the type of deviation; if the deviation is caused by inadequate equipment execution, adjusting the control parameters of the corresponding equipment to increase and / or decrease the equipment action amplitude; if the deviation is caused by environmental interference and / or mutual influence of multiple devices, recalculating the influence intensity, updating the building association map, and regenerating the task decomposition scheme and equipment control parameters based on the updated map.

[0011] Preferably, the steps for triggering global strategy control include: counting the number of manual interventions by users in each room unit within a preset time period; classifying user manual intervention behaviors according to control strategy type, and calculating the number of interventions and intervention ratios triggered by each control strategy; when the intervention ratio of a certain control strategy exceeds a preset intervention ratio threshold, extracting the triggering conditions, control parameters, and execution result data corresponding to that control strategy; analyzing the environmental parameter characteristics and time distribution characteristics that trigger the intervention, and adjusting the global trigger threshold, control intensity, and / or execution logic of the control strategy; conducting trial runs of the adjusted control strategy in some buildings, and collecting user intervention data and energy consumption data during the trial run; when the performance is better than the original strategy, extending the adjusted control strategy to all buildings.

[0012] The IoT-based building data intelligent analysis and decision-making system includes: The data acquisition module is used to acquire spatial parameters, glass material parameters, and real-time environmental data collected by IoT sensors for each building. The influence correlation modeling module is used to combine the sun's position and spatial geometric relationships to construct a building correlation map and identify influence transmission chains; The task decomposition module queries the building's equipment capacity list and equipment status association matrix based on sub-tasks and constraints to determine equipment control parameters. The scheme adjustment module is used to collect equipment execution feedback data and environmental change data. When the deviation from the expected value exceeds the threshold, the equipment control parameters are adjusted and / or the control scheme is regenerated. The behavior learning module is used to record user manual intervention behaviors and group intervention patterns, and dynamically adjust control parameters and global strategies.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. The building association map construction method proposed in this invention combines multi-source parameters such as floor 3D coordinates, glass area, orientation angle, solar altitude angle, and azimuth angle. It utilizes ray tracing to calculate the propagation path of reflected light from glass curtain walls, thereby automatically identifying directional influence relationships across buildings and constructing a weighted influence map of reflected light energy. Furthermore, this invention identifies multi-level influence transmission chains, marks intermediate nodes, and identifies multi-level implicit influences from "source floor → intermediate floor → affected floor." This overcomes the limitations of traditional building control systems that can only identify direct influences, enabling the system to discover potential influence chains in advance and perform predictive analysis of influences within future time windows. This significantly improves the accuracy and foresight of overall building complex control, reducing the impact of sudden light interference and temperature increases on user comfort and energy consumption.

[0014] 2. The control task decomposition mechanism proposed in this invention can automatically break down each control task into "source-side shading sub-tasks" and "received-side compensation sub-tasks" based on cross-building influence relationships, and set differentiated constraints such as illuminance lower limit, power upper limit, and air conditioning load limit for each. By assigning different priorities to the influence intensity and scope of the tasks, the system can dynamically select the most effective intervention path. At the device execution level, this invention identifies strongly coupled and weakly coupled devices through a device state association matrix, and adopts sequential, parallel, or staggered execution control respectively, effectively avoiding mutual interference between devices or energy consumption superposition. This invention constructs a cross-device and cross-building collaborative control system, enabling the system to achieve the best environmental improvement effect with the lowest energy consumption, significantly improving overall energy efficiency and user comfort.

[0015] 3. This invention records user manual interventions and their corresponding environmental parameters, extracts individual user preference rules, and automatically adjusts the optimization parameters of corresponding rooms, forming a personalized adaptation mechanism for the user end. Simultaneously, this invention statistically analyzes the group intervention ratio, performs a global evaluation of frequently questioned or resisted local strategies, and automatically identifies the irrationality of the triggering conditions and control parameters of a strategy when the intervention ratio exceeds a threshold, and executes dynamic adjustments to the global strategy. By piloting the adjusted strategy in selected buildings, this invention employs a closed-loop mechanism of "verification-optimization-promotion" to ensure that the new control strategy has higher user acceptance and better energy efficiency. Attached Figure Description

[0016] Figure 1 A flowchart illustrating the IoT-based intelligent analysis and decision-making method for building data provided by this invention; Figure 2 The structural diagram of the building data intelligent analysis and decision-making system based on the Internet of Things provided by this invention; Figure 3 The device control parameter flowchart provided by this invention; Figure 4 A flowchart for global strategy adjustment provided by the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0018] Example 1: Please see Figure 1 and Figure 2 This invention provides an IoT-based intelligent analysis and decision-making method for building data, applicable to an IoT-based intelligent analysis and decision-making system for building data. The method includes a data acquisition module, an impact correlation modeling module, a task decomposition module, a scheme adjustment module, and a behavior learning module. The technical solution is as follows: Acquire spatial parameters, glass material parameters, and real-time environmental data collected by IoT sensors for each building; The main parameters of glass material include the glass type designation and the corresponding refractive index value. For ordinary flat glass, the refractive index is 1.52, while for low-emissivity coated glass, the refractive index ranges from 1.50 to 1.55. In addition, glass material parameters also involve geometric properties such as the effective reflective area of ​​the glass curtain wall, the total glass area, and the main orientation angles.

[0019] Real-time environmental data covers multi-dimensional environmental monitoring information for both indoor and outdoor environments; indoor environmental data includes indoor illuminance, indoor temperature, and user activity status; outdoor environmental data includes outdoor temperature, humidity, solar radiation intensity, and cloud cover.

[0020] By combining the sun's position and spatial geometry, a building association map is constructed to identify the impact transmission chain; Furthermore, the spatial parameters include floor height, building spacing, and three-dimensional coordinates; the solar position includes solar altitude angle and azimuth angle; the building association map includes multiple floor nodes and directed influence edges between nodes, and the weight of the directed influence edge represents the influence intensity and influence time period of the source floor on the target floor; The steps to construct a building association map include: The spatial area of ​​each building is divided into grid cells, and nodes are created for each floor of each building. The node attributes include the floor center coordinates, glass area, and orientation angle. The solar altitude angle and azimuth angle are calculated at multiple time points within a preset time window. For each time point, the floor nodes with glass curtain walls are traversed as influence source nodes. The propagation direction of reflected light is calculated based on the solar incident angle and the glass reflectivity. The glass reflectivity is dynamically calculated based on the Fresnel equation according to the incident angle. The target floor node reached by the reflected light is determined by ray tracing. The reflected light energy reaching the target node is calculated based on the initial energy of the reflected light, the propagation distance, and the air attenuation coefficient. When the reflected light energy reaching the target node exceeds a preset threshold (initial value is 50 lux), a directed edge is established between the influence source node and the target node. The weight of the directed edge is the calculated influence intensity. A set of directed graphs with time series is generated to form a building association map. The identification of the influence transmission chain includes: analyzing the multi-level transmission paths existing in the directed graph, and when the transmission relationship of the first node influencing the second node and the second node influencing the third node is identified, the second node is marked as a key transit node.

[0021] To calculate the initial energy of the reflected light, the direct solar irradiance at the current time is first obtained. This irradiance is determined based on the solar altitude angle and atmospheric mass coefficient. Under clear weather conditions, the ground solar irradiance is in the range of 800 to 1000 watts per square meter. The incident irradiance reaching the surface of the glass curtain wall at the source node is then calculated. This is equal to the direct solar irradiance multiplied by the cosine of the incident angle, where the incident angle is the angle between the direction of solar incidence and the normal direction of the glass curtain wall. The incident irradiance is then multiplied by the glass reflectivity and then by the effective reflective area of ​​the glass curtain wall to obtain the total power of the reflected beam, i.e., the initial energy of the reflected light, in watts.

[0022] When calculating the energy attenuation of reflected light, it is assumed that the glass curtain wall is a uniform reflective surface and the reflected light propagates in a divergent manner. Based on the three-dimensional spatial distance between the source node and the target node, the geometric diffusion attenuation factor is calculated using the inverse square law. This factor is equal to the ratio of the luminous flux received per unit area at the target node to the luminous flux reflected at the source. At the same time, atmospheric transmission loss is considered. The atmospheric transmittance is calculated based on the propagation path length and the atmospheric extinction coefficient. The extinction coefficient is taken as 0.1 per kilometer in clear weather and 0.5 per kilometer in hazy weather. The atmospheric transmittance is equal to the negative extinction coefficient with the base of the natural logarithm multiplied by the exponent of the distance. The reflected light energy reaching the target node is equal to the initial energy of the reflected light multiplied by the geometric diffusion attenuation factor and then multiplied by the atmospheric transmittance.

[0023] When converting the reflected light energy reaching the target node into indoor illuminance impact, first calculate the irradiance of the reflected light illuminating the window on the target floor, which is equal to the light energy reaching the target floor divided by the irradiance area. Considering the transmittance of the window glass, which is generally between 0.6 and 0.8, the luminous flux transmitted into the room is equal to the irradiance received by the window multiplied by the window area and then multiplied by the transmittance. Based on the location and area of ​​the indoor work surface, assuming that the light diffuses uniformly indoors, calculate the illuminance increment generated by this luminous flux on the work surface. This illuminance increment is the impact intensity, measured in lux. When the impact intensity exceeds 50 lux, it is considered a significant impact.

[0024] For cases where the same target node receives reflected light from multiple source nodes, the illuminance effects of each source node are linearly superimposed to calculate the total illuminance effect on the target node.

[0025] The glass material parameters include the glass type identifier and the corresponding refractive index value. The refractive index of ordinary flat glass is 1.52, while the refractive index of low-emissivity coated glass is a specific value within the range of 1.50 to 1.55. The system selects the corresponding refractive index based on the glass type recorded in the building archive. When calculating the reflection coefficient using Fresnel's equations, since sunlight is unpolarized, the reflection coefficients of s-polarized and p-polarized light are calculated separately. The s-polarized reflection coefficient is calculated by squared the ratio of the incident angle and the refraction angle using a sine function, and the p-polarized reflection coefficient is calculated by squared the ratio of the incident angle and the refraction angle using a tangent function. The refraction angle is determined by the incident angle and the refractive indices of the two media according to the law of refraction. The average of the s-polarized and p-polarized reflection coefficients is taken as the equivalent reflection coefficient of unpolarized sunlight. For multi-layered glass structures such as insulated glass, to simplify the calculation, only the reflection from the outermost glass surface is considered, ignoring the contribution of the inner layer reflection, because the light reflected from the inner layer needs to pass through multiple transmissions and reflections from the outer glass, resulting in significant energy attenuation.

[0026] The target floor node reached by reflected light is determined by ray tracing. Specifically, each floor node is represented by its center coordinates and a representative window surface, which is represented by a rectangular plane. The direction of the rectangle is determined by the orientation angle. Reflected light is emitted from the center of the glass curtain wall at the source node. The direction of the light is determined according to the law of specular reflection, i.e., the reflection angle equals the incident angle, and the reflected ray, incident ray, and mirror normal are coplanar. During the ray's propagation in three-dimensional space, it is checked whether it intersects with the outer envelope boxes of other buildings. The outer envelope box is represented by a cuboid determined by the building's three-dimensional coordinates and floor height. If the ray reaches the target node before... If the ray intersects with the outer envelope of other buildings, it is considered an obstruction, and the influence relationship is not valid. If the ray is not obstructed, it is further determined whether it is directed towards the window surface of the target node. By calculating the intersection point of the ray and the rectangular plane of the target node window, if the intersection point is within the rectangular area, it is determined that the ray has reached the target node. Further, the incident angle of the ray relative to the target window surface is determined. If the incident angle is greater than 90 degrees, it means that the light is shining from the back. In this case, the light cannot enter the room, and the influence relationship is not valid. Only when the ray is not obstructed and the incident angle is less than 90 degrees is it confirmed that the reflected light can reach the target node and may have an impact on the room.

[0027] The building association map is stored in the system in the form of a graph database or adjacency list. The node data includes attribute fields such as a unique node identifier, the building identifier, the floor number, the three-dimensional coordinates of the floor center, the total area of ​​the glass curtain wall, and the main orientation angle. The directed edge data includes attribute fields such as the source node identifier, the target node identifier, the influence intensity value, the influence effective time, and the influence end time. To support time window queries, a time index is established for the effective time and end time of the edges. When it is necessary to query the influence relationship within a future time window, the time index is used to quickly locate the set of edges that are effective within that time window and filter out the edges whose influence intensity exceeds a preset threshold. The map data is updated periodically as the preset time window moves, adding newly calculated influence relationships for future time points.

[0028] The calculation steps for the reflected light energy are as follows: First, based on the sun's position and solar radiation intensity data, calculate the incident irradiance reaching the surface of the glass curtain wall on the source floor. This incident irradiance is the direct solar irradiance multiplied by the cosine of the incident angle. Then, calculate the reflection coefficient of the glass curtain wall using the Fresnel equation and the incident angle. Multiply the incident irradiance by the reflection coefficient and then by the effective reflective area of ​​the glass curtain wall to obtain the initial energy of the reflected light. Based on the direction of light propagation and the three-dimensional distance from the source node to the target node, apply the inverse square law to calculate the energy attenuation caused by geometric diffusion, while also considering atmospheric transmission. The energy attenuation caused by the rate is calculated, where atmospheric transmittance is calculated based on the propagation distance and the extinction coefficient under current weather conditions. The extinction coefficient is taken as 0.1 per kilometer in clear weather and 0.5 per kilometer in hazy weather. The attenuated reflected light energy is divided by the window area of ​​the target floor and multiplied by the glass transmittance to obtain the luminous flux entering the room. Then, the illuminance increment generated by this luminous flux on the working surface is calculated according to the geometric relationship of the indoor space. This illuminance increment is the influence intensity, and the unit is lux. When the influence intensity exceeds 50 lux, it is judged as a significant influence, and a directed edge is established.

[0029] The purpose of identifying multi-level transmission chains is to discover latent influence relationships. However, considering the significant energy attenuation of light after multiple reflections, the system only identifies second-level transmission chains and does not consider third-level or higher transmissions. When calculating the influence intensity of second-level transmission, the system considers the re-reflection loss of the relay node based on the first-level influence intensity. That is, after the beam reaches the second node, it is assumed that 15% to 25% of its energy is specularly reflected again by the glass curtain wall of the second node. The reflection ratio is determined according to the reflection coefficient of the glass of the second node. The re-reflected light is used to calculate the influence intensity reaching the third node according to the aforementioned ray tracing and energy attenuation methods. When the influence intensity of the second-level transmission still exceeds 10 lux, a second-level influence edge is established between the first and third nodes. This edge is marked as a second-level transmission type in the directed graph, and the relay node identifier is recorded. When the control task is decomposed, the relay node, as both the upstream influence receiving node and the downstream influence source node, needs compensation measures. The system coordinates these two types of measures to ensure that the relay node's shading action does not exacerbate its own insufficient illumination problem. Secondary transfers with energy below 10 lux are considered to have no practical impact value, and no impact edge is established.

[0030] By constructing a building association graph containing floor nodes and directed influence edges, the system can accurately identify and quantify the relationship between light effects across floors and buildings. It can calculate the sun's position, simulate reflection paths, and generate a set of time-series directed graphs, revealing the multi-level transmission chain of light in the building complex and key transfer nodes. This provides high-precision data support for environmental control, energy-saving optimization, and equipment linkage decision-making, thereby improving the intelligence and response efficiency of the overall management of the building complex.

[0031] A control task list is generated based on the building association map. Each control task is decomposed into a source-side occlusion sub-task and a receiver-side compensation sub-task. Priorities are assigned to the sub-tasks and constraints are set. Furthermore, the steps for generating a control task list based on the building association map include: Query the set of directed edges in the building association map where the influence intensity is greater than a preset threshold within a preset future time window; extract the source node, target node, influence intensity, and influence time period corresponding to each directed edge; obtain the number of affected rooms and historical user feedback records; calculate the task priority score based on the influence intensity, the number of affected rooms, and the historical user feedback records; sort the influence relationships according to the priority score to generate a task list containing the source node, target node, influence intensity, time window, and priority.

[0032] The task priority score is calculated as follows: First, the impact intensity is normalized to the zero-to-one range using a logarithmic scale. Specifically, the minimum significant impact intensity is set to 50 lux, corresponding to a normalized value of 0.1, and the maximum expected impact intensity is set to 5000 lux, corresponding to a normalized value of 1. The median value is calculated using logarithmic interpolation to obtain the impact intensity normalized score. Second, the number of affected rooms is normalized linearly, with 0.1 for a single room and 1 for ten or more rooms. Third, historical user feedback records are processed to count the manual interventions triggered by the user in the past thirty days. The user dissatisfaction rate is calculated by multiplying the number of manual interventions by 0.5 and summing the result with the number of complaints, then dividing by the total number of impacts. This rate is directly used as the feedback weighting factor. Finally, the three normalized scores are weighted and summed, with weights of 0.5 for impact intensity, 0.3 for the number of affected rooms, and 0.2 for user dissatisfaction, to obtain the final priority score. The score ranges from 0.1 to 1, with a higher score indicating a higher task priority and requiring priority handling. For impact relationships that occur for the first time and for which there is no historical feedback data, the user dissatisfaction item uses a default value of 0.5.

[0033] By filtering relationships where the intensity of impact exceeds a threshold within a future time window, and combining this with the number of affected rooms and historical user feedback, a priority-quantified control task list is generated. This list accurately identifies the sources of impact, target nodes, and time periods of effect, enabling the system to implement more efficient environmental control strategies based on task priorities. This allows for early response to key impact scenarios, significantly improving the proactivity and comfort of building complex environmental management.

[0034] Further, the step of decomposing each control task into source-side shading sub-tasks and receiver-side compensation sub-tasks includes: for each task in the control task list, generating a source-side shading sub-task based on the type of the influencing source node, wherein the source-side shading sub-task targets the shading devices and / or dimming glass of the influencing source building; generating a receiver-side compensation sub-task based on the type of the target node, wherein the receiver-side compensation sub-task targets the lighting system and air conditioning system of the influencing building; setting a first constraint for the source-side shading sub-task, the first constraint including that the indoor illuminance of the influencing source building is not lower than the lower limit of comfortable illuminance; setting a second constraint for the receiver-side compensation sub-task, the second constraint including that the increase in lighting power does not exceed a preset proportion of the base power and a limit on the increase in air conditioning load; and assigning execution priorities to each sub-task based on the impact intensity and affected range of the task. The lower limit of illuminance is determined according to the building use, set at 300 lux for office buildings and 100 lux for residential buildings.

[0035] By breaking down control tasks into source-side shading sub-tasks and receiver-side compensation sub-tasks, cross-building illumination impacts can be refined into executable device-level control commands. Appropriate shading, dimming, lighting, or air conditioning equipment is automatically matched based on the source and target node types, and constraints such as illuminance lower limits, power amplification ratios, and air conditioning loads are set to ensure a balance between comfort and energy consumption in control measures. Execution priorities are allocated based on impact intensity and affected range, allowing for the priority handling of high-performing control tasks. This achieves closed-loop management from impact identification to equipment control, improving the execution efficiency of building cluster environmental control.

[0036] Based on the subtasks and constraints, query the building's equipment capacity list and equipment status correlation matrix to determine the equipment control parameters, referring to... Figure 3 For strongly coupled devices, sequential execution control is used; for weakly coupled devices, parallel execution control is used. Further, the steps for determining equipment control parameters include: reading the equipment state correlation matrix, which records the influence intensity and delay time of the first equipment action on the state of the second equipment; extracting the equipment combinations involved according to the equipment control instructions to be executed, and calculating the correlation strength between the equipment; when the correlation strength between the equipment exceeds the strong coupling threshold, it is determined to be a strongly coupled equipment, and the control instructions of the strongly coupled equipment are arranged into a sequence execution control, first starting the master equipment and monitoring its state changes, and when the rate of change of the output parameters of the master equipment within a preset time window is less than the stability threshold, the master equipment is determined to be stable, and the slave equipment is started after the master equipment is stable; when the correlation strength between the equipment is lower than the weak coupling threshold, it is determined to be a weakly coupled equipment, and the control instructions of the weakly coupled equipment are arranged into parallel execution control, and control instructions are issued simultaneously; when the correlation strength between the equipment is between the strong coupling threshold and the weak coupling threshold, staggered execution control is adopted, and the execution time interval is set according to the influence delay time.

[0037] The equipment capability list refers to a data table recording the specifications and parameters of each device in the system (such as the shading angle range of the sunshade device, the power adjustment range of the lighting system, and the cooling and heating capacity of the air conditioning system). The calculation of the correlation strength between devices is based on the equipment state correlation matrix, which is a square matrix where rows and columns correspond to each device in the system. The matrix element M(i,j) represents the influence strength of the action of device i on the state of device j, with a value range of 0 to 1, where 0 indicates no influence and 1 indicates a strong direct influence. When it is necessary to calculate the combination involving multiple devices, the sub-matrices corresponding to these devices in the matrix are extracted. The matrix is ​​used to calculate the maximum value of all off-diagonal elements in the submatrix as the overall correlation strength of the device combination. A strong coupling threshold of 0.7 and a weak coupling threshold of 0.3 are set. When the overall correlation strength is greater than 0.7, it is determined to be strong coupling, and sequential execution control is required. When the overall correlation strength is less than 0.3, it is determined to be weak coupling, and parallel execution control can be used. When the overall correlation strength is between 0.3 and 0.7, it is determined to be medium coupling, and staggered execution control is used. The influence strength value is obtained through statistical analysis of running data. The specific calculation method is described in the subsequent description of the device status correlation matrix update steps.

[0038] The equipment state correlation strength M(i,i) is based on data-driven principles and incorporates the constraints of the physical model. Specifically, for equipment pairs with clear physical relationships (such as indoor shading devices and indoor illuminance, indoor illuminance and lighting system power), the correlation strength is calculated based on the following two components: (1) the statistical correlation component S_stat(ij), which is derived from the time-amplitude correlation index between the action of equipment i and the state change of equipment j in the operating data, with a range of 0-1; (2) the physical constraint component S_phys(ij), which is calculated based on the technical specifications and spatial layout of the equipment. For example, the influence of shading devices on indoor illuminance can be calculated using the shading coefficient of the building physics model, and the influence of the lighting system on the air conditioning system can be estimated using the heat load data in the equipment specifications.

[0039] The strong coupling threshold of 0.7 is determined based on the following principle: when the correlation of state changes between two devices exceeds 0.7, it means that a change in the state of one device has a greater than 70% probability of triggering a significant state change in the other device within a preset delay time. In control theory, this high correlation indicates that the dynamic responses of the two devices are strongly coupled, making parallel control unsuitable to avoid mutual interference. Correspondingly, a weak coupling value of 0.3 indicates that the correlation of state changes between two devices is less than 30%, meaning that the control of one device has no statistically significant impact on the other, and parallel execution is safe.

[0040] By introducing a device state correlation matrix, the influence intensity and delay characteristics between different devices are quantified, enabling intelligent orchestration of device control parameters. Based on the correlation strength of device combinations, strong coupling, weak coupling, and intermediate coupling relationships are automatically distinguished, and sequential execution, parallel execution, or off-peak execution control are adopted respectively. This ensures that the device linkage process avoids mutual interference while significantly improving response efficiency. By monitoring the stability of the master device state in strong coupling scenarios, synchronously issuing commands in weak coupling scenarios, and setting reasonable time intervals in intermediate coupling scenarios, the safety, stability, and efficiency of building environment equipment in linkage control can be ensured, significantly enhancing overall control capabilities and execution reliability.

[0041] The device status association matrix is ​​established and updated through the following steps: During system initialization, an initial correlation matrix is ​​preset based on the physical characteristics of the equipment and the control logic. This initial correlation matrix records the theoretical correlation strength between the equipment. During system operation, the execution time, equipment type, and control parameters of each equipment control command are recorded. Multiple equipment state change data are collected after the execution of the equipment control commands. These state change data include the amplitude of the state change and the time to reach a stable state. The time correlation and amplitude correlation between the action of the first equipment and the state change of the second equipment are calculated. When both the time correlation and amplitude correlation exceed preset thresholds, it is determined that there is an actual correlation between the first and second equipment. The correlation strength value is updated according to the amplitude of the state change, and the influence delay time value is updated according to the time to reach a stable state. The equipment state correlation matrix is ​​periodically corrected using actual operating data, so that the correlation matrix gradually approximates the actual influence relationship between the equipment.

[0042] To avoid misinterpreting statistical correlation as causal association, the system employs controlled experiments when analyzing inter-device correlations. This involves actively triggering the action of the first device after excluding other interfering factors, and observing whether the state of the second device changes accordingly. Specifically, during a relatively stable period of the external environment (such as nighttime or cloudy days), the system actively adjusts the control parameters of the first device, recording the adjustment time and magnitude. It continuously monitors the state changes of the second device, recording the time and magnitude of these changes. The system calculates the time difference between the action of the first device and the state change of the second device; this time difference is the impact delay time. The system also calculates the ratio of the magnitude of the state change of the second device to the magnitude of the action of the first device; this ratio reflects the intensity of the impact. The experiment is repeated multiple times, and the stability of the impact delay time and intensity is statistically analyzed. When the same device exhibits consistent delay time and impact intensity in multiple experiments, a causal relationship is determined to exist.

[0043] The determination of causal direction is based on the initiative of the action. The device that actively adjusts is the cause, and the device that responds to the change is the effect. Therefore, the element M(i,j) in the correlation matrix explicitly represents the impact of device i's active adjustment on the state of device j. For possible bidirectional effects, experiments are conducted in both directions, and M(i,j) and M(j,i) are recorded in the matrix respectively, although their values ​​may differ. To eliminate confounding variables, the system locks out other external factors that may affect the two devices during the experiment, such as fixing the outdoor temperature sampling value and prohibiting other devices from operating, to ensure that the observed state changes are mainly caused by the actions of the experimental devices.

[0044] The method for updating the correlation strength value is as follows: calculate the ratio of the magnitude of the change in the state of the second device after the first device takes action to the normal operating range of the second device. This ratio is the observed influence strength in this event. Perform an exponential moving average on the influence strength of multiple observations with a smoothing coefficient of 0.2. That is, the new correlation strength is equal to the original correlation strength multiplied by 0.8 plus the current observed influence strength multiplied by 0.2. This updating method enables the correlation matrix to gradually approximate the actual coupling relationship between devices while maintaining the ability to adapt to system changes.

[0045] By learning online and continuously updating the device status correlation matrix, the system can adaptively identify the actual correlation between devices, making up for the coupling effects between devices that may be missed when relying solely on theoretical design. This approach enhances the system's adaptability, and as the operating time increases, the accuracy and stability of device linkage control continuously improve, reducing system maintenance costs.

[0046] Collect equipment execution feedback data and environmental change data. When the actual effect deviates from the expected value by more than a threshold, adjust the equipment control parameters and / or regenerate the control scheme. Further, the steps of adjusting equipment control parameters and / or regenerating control schemes include: collecting position sensor data of the shading device, illuminance sensor data of the lighting system, and temperature sensor data of the air conditioning system as equipment execution feedback data; collecting environmental sensor data of the affected area as environmental change data, including indoor illuminance, temperature, and user activity status; calculating the deviation between the equipment execution feedback data and the expected target value, and calculating the deviation between the environmental change data and the desired environmental parameters; when any deviation exceeds a preset threshold, determining the type of deviation; if the deviation is caused by inadequate equipment execution, adjusting the control parameters of the corresponding equipment to increase and / or decrease the equipment's action amplitude; if the deviation is caused by environmental interference and / or mutual influence of multiple devices, recalculating the influence intensity, updating the building association map, and regenerating the task decomposition scheme and equipment control parameters based on the updated map. The determination of the deviation type includes: comparing the consistency between the equipment feedback status and the control command; when the consistency deviation exceeds the equipment execution deviation threshold, determining that the equipment is not executing properly; when the consistency deviation does not exceed the equipment execution deviation threshold but the environmental parameter deviation exceeds the environmental deviation threshold, determining that it is environmental interference and / or mutual influence of multiple devices.

[0047] The consistency deviation is calculated as follows: for positional devices such as sunshades, the consistency deviation is the absolute value of the difference between the actual position and the commanded position divided by the percentage of the device's travel; for output devices such as lighting and air conditioning, the consistency deviation is the absolute value of the difference between the actual output power and the commanded power divided by the percentage of the rated power; the device execution deviation threshold is uniformly set at 10%, that is, when the relative deviation between the actual state and the commanded state exceeds 10%, it is determined that the device is not executing properly; the environmental deviation threshold is set according to the type of environmental parameters, with the illuminance deviation threshold at 100 lux and the temperature deviation threshold at 1 degree Celsius; When the equipment is determined to be malfunctioning, the method for adjusting the equipment control parameters is as follows: Calculate the current deviation rate, which is the difference between the actual environmental value and the target value divided by the target value. Calculate the adjustment increment based on the deviation rate. The adjustment increment is equal to the deviation rate multiplied by the current output value of the equipment multiplied by the gain coefficient, with the gain coefficient set to 1.2. Add the adjustment increment to the original control command to form a new control command. However, the new command must not exceed the maximum output capacity of the equipment; if it does, it is limited to the maximum value. To prevent adjustment oscillations, the time interval between two consecutive adjustments must not be less than five minutes, and the magnitude of a single adjustment must not exceed 20% of the equipment's rated output. When the issue is determined to be environmental interference and / or mutual influence of multiple devices, the method for recalculating the influence intensity is as follows: extract the actual solar position and cloud cover data at the current moment, recalculate the reflected light energy, update the influence intensity value of the affected floor, and if the relative deviation between the updated influence intensity and the original value exceeds 30%, it is determined that the influence relationship has changed significantly, triggering the regeneration of the task list and device control parameters; if the relative deviation does not exceed 30%, only the influence intensity value is updated but the scheme is not regenerated, and the current scheme continues to be executed.

[0048] By collecting real-time feedback data from devices such as shading, lighting, and air conditioning, and combining this with environmental change data such as indoor illuminance, temperature, and user activity, the system achieves dynamic evaluation of control effects. When device execution deviations and / or environmental deviations exceed thresholds, the system can automatically identify the source of the deviation: if it is due to insufficient device execution, the system will immediately adjust control parameters to strengthen and / or weaken device actions; if it is due to environmental disturbances and / or coupling effects between devices, the system will recalculate the intensity of light impact, update the building association map, and generate a new task decomposition scheme and device control parameters based on the latest map. This creates an adaptive closed loop in the control process, significantly improving the stability, accuracy, and robustness of building environment control.

[0049] Record manual intervention behaviors and simultaneous environmental parameters, extract user preference rules, and adjust the optimization parameters of the corresponding room; statistically analyze group intervention patterns, and when the intervention ratio triggered by the same optimization strategy exceeds a preset value, trigger global strategy adjustment, referring to... Figure 4 .

[0050] Furthermore, the steps for triggering global strategy control include: counting the number of manual interventions by users in each room unit within a preset time period; classifying user manual intervention behaviors according to control strategy type, and calculating the number of interventions and intervention ratios triggered by each control strategy; when the intervention ratio of a certain control strategy exceeds a preset intervention ratio threshold, extracting the triggering conditions, control parameters, and execution result data corresponding to that control strategy; analyzing the environmental parameter characteristics and time distribution characteristics that trigger the intervention, and adjusting the global trigger threshold, control intensity, or execution logic of the control strategy; conducting trial runs of the adjusted control strategy in some buildings, and collecting user intervention data and energy consumption data during the trial run; when the performance is better than the original strategy, extending the adjusted control strategy to all buildings.

[0051] The control strategies are classified according to the control scenarios and triggering conditions, mainly including: light interference response strategies, which are triggered when reflected light between buildings causes indoor illuminance to exceed the standard; temperature regulation strategies, which are triggered when indoor temperature deviates from the comfort range; and energy-saving operation strategies, which are triggered during non-working periods and / or low-load periods. Each strategy includes elements such as triggering conditions, control equipment combinations, and control parameter ranges. User manual intervention behaviors are categorized according to the type of control strategy being executed by the system at the time of intervention. The specific method is as follows: record the time of intervention, query the system execution log at that time, extract the type of control strategy being executed and / or just completed, and categorize the intervention behavior into the corresponding strategy; if the system does not execute any control strategy at the time of intervention, it is categorized as user active adjustment and is not included in the strategy evaluation. The specific methods for adjusting the control strategy are as follows: Analyze the characteristics of the environmental parameters that trigger intervention to identify whether the triggering conditions are too sensitive or too sluggish; if the intervention behavior shows that the user tends to regulate earlier or later, adjust the trigger threshold by 10% to 20% of the original threshold; if the intervention behavior shows that the user tends to regulate more or less, adjust the target value output by the device by adjusting the magnitude based on the statistical average of the intervention behavior; if the intervention behavior shows that there is a problem with the execution logic of the strategy, such as an unreasonable device startup sequence, adjust the timing arrangement of device linkage. The performance evaluation standard is a comprehensive score, calculated as follows: the user intervention reduction rate multiplied by 0.6 plus the energy consumption reduction rate multiplied by 0.4. When the comprehensive score is positive, it is considered to be better than the original strategy. The trial operation period is set to seven calendar days. Complete user intervention data and energy consumption data are collected and compared with historical data of the same period type before the implementation of the new strategy (such as working days compared with working days) to calculate the reduction rate or reduction rate.

[0052] Before triggering global policy adjustments, a two-layer causal analysis is performed to differentiate the causes of intervention: The first layer is environmental correlation analysis, which statistically analyzes the correlation between environmental parameters (such as outdoor temperature, solar irradiance, and time) at the time of each intervention and the triggering conditions of the control policy. When interventions occur multiple times under the same triggering conditions, the intervention is considered related to policy failure. The second layer is user feedback correlation analysis, which analyzes whether user interventions are accompanied by fault alarms, temperature discomfort complaints, or other explicit feedback through building property management systems or sensors. Only when the intervention is related to clear discomfort feedback is it determined to be a policy-induced intervention. Only the proportion of interventions identified as "policy-related interventions" through the above two layers of analysis is counted. When this proportion exceeds 15% within a 30-day sampling period, global policy adjustments are triggered. At the same time, this value is automatically adjusted according to building use, user type, and season (12% for office buildings during working hours and 18% for residential buildings).

[0053] By periodically statistically analyzing and classifying user manual intervention behaviors, the effectiveness of building control strategies can be quantitatively evaluated. When the intervention rate of a certain type of control strategy exceeds a threshold, the system automatically analyzes the deviations between its triggering conditions, parameter settings, and environmental characteristics, identifies unreasonable aspects of the strategy, and performs global optimization at the parameter or logic level. The optimized strategy is first piloted in local buildings, and its improvement effect is verified by comparing user intervention data and energy consumption performance. If the overall performance is better than the original strategy, it is promoted throughout the entire area. A closed-loop system for strategy iteration is constructed, enabling the environmental control strategies of building clusters to continuously evolve, better meet user behavior patterns and energy-saving needs, and improve the system's intelligence and adaptability.

[0054] Example 2: This invention provides an IoT-based intelligent analysis and decision-making method for building data. This method, through cross-building impact identification, intelligent task decomposition, and coordinated device linkage, combined with a user behavior self-learning mechanism, forms a continuously optimized intelligent control system for building clusters. Building upon the basic technical solution, it also includes load forecasting and advance adjustment, and cross-building energy consumption collaborative optimization, further enhancing the method's technological leap from passive response to proactive control, and from single-building optimization to overall building cluster optimization.

[0055] The specific steps for load forecasting and advance adjustment are as follows: Historical environmental data, equipment operation data, and energy consumption data are acquired to establish a time-series dataset. Weather forecast data for future time windows, including temperature, humidity, solar radiation intensity, and cloud cover, is obtained. Meeting reservation records and personnel entry / exit plans are queried from the building management system to extract space occupancy prediction information. Historical data, weather forecast data, and occupancy prediction information are input into a load prediction model to predict the heating and cooling load demand and lighting demand of each building within the future time window. Based on the predicted load curve and equipment response time characteristics, the equipment start-up lead time is calculated. The equipment response time characteristics include equipment start-up delay time, transition time from start-up to stable output, and lag time in response to spatial environmental parameters. The corresponding equipment is started at the lead time before the predicted load arrives, ensuring that the equipment reaches the target operating state when the load arrives.

[0056] The device response time characteristics are obtained through historical operation data statistics. The system records the complete process of each device startup, including the time when the control command is issued, the actual startup time of the device, the time when the output power reaches the target value, and the time when the indoor environmental parameters reach the target value. The statistical distribution of each time interval is calculated, and the median is extracted as the typical response time parameter of the device. The device startup delay time is usually from several seconds to tens of seconds. The transition time varies depending on the type of device. For lighting systems, it is about 1 to 2 minutes, and for air conditioning systems, it is about 10 to 30 minutes. The environmental parameter response lag time is related to the room volume and the device power, and is usually 5 to 60 minutes.

[0057] When calculating equipment start-up lead time, the equipment start-up delay, transition time, and environmental response lag time are added together to obtain the basic lead time. Considering the uncertainty of load forecasting, a safety margin is added to the basic lead time based on historical forecast error statistics. The safety margin is 10% to 20% of the forecast time window length; the longer the forecast time window, the greater the uncertainty, and the higher the safety margin ratio. The final lead time equals the basic lead time plus the safety margin. Equipment is started at the lead time before the predicted load arrives. After start-up, the system continuously monitors changes in actual load and environmental parameters. If the actual load arrives earlier than predicted, the equipment adjustment speed is immediately accelerated or backup equipment is activated. If the actual load arrives later than predicted, the equipment output is appropriately reduced to avoid over-adjustment and energy waste. This dynamic adjustment mechanism compensates for forecast errors, ensuring effective predictive control in uncertain environments.

[0058] The load forecasting model employs a Long Short-Term Memory (LSTM) network architecture, training an independent forecasting model for each building. The model inputs include historical environmental data from the past seven days (sampled hourly), equipment operation data, and energy consumption data, forming a time series of 168 time steps. Each time step includes 12 feature dimensions such as indoor and outdoor temperature, humidity, solar radiation intensity, equipment power, and energy consumption. Weather forecast data for the next 24 hours (one data point per hour) and occupancy forecast information serve as external feature inputs. The model outputs the cooling and heating load demand for the next 24 hours. Lighting demand is predicted hourly; the model structure is a three-layer LSTM with 128 hidden units per layer, followed by two fully connected layers with 64 and 24 neurons respectively. The last layer of 24 neurons corresponds to the predicted value for the next 24 hours; the model uses mean squared error as the loss function and is trained using the Adam optimizer; the training dataset uses at least three months of historical data, divided into training and validation sets in chronological order, with the training set accounting for 80%; data preprocessing includes removing outliers and imputing missing values, and normalizing each feature to zero mean and unit variance; The model is trained offline. It is trained on historical data during the initial deployment and then retrained monthly to adapt to seasonal and usage pattern changes. Predictions are executed at midnight every day to generate the 24-hour load prediction curve for that day. The system monitors the prediction error and triggers the model retraining procedure when the average prediction error for three consecutive days exceeds 20%.

[0059] By integrating multi-source data for load forecasting and adjusting equipment in advance based on its response characteristics, this approach avoids the response lag problem caused by starting equipment only after the load arrives, which is common in traditional methods. This not only improves environmental comfort but also avoids energy waste caused by over-adjustment due to response lag. This solution transforms passive response into active regulation, achieving predictive control.

[0060] The steps for cross-building energy consumption collaborative optimization are as follows: The system acquires real-time energy consumption data and equipment operating status of each building in the building complex; obtains time-of-use electricity pricing information and energy consumption budget constraints for each building; constructs a multi-building energy consumption optimization model, the objective function of which is to minimize the total energy consumption cost of the building complex, with constraints including that the indoor environmental parameters of each building meet comfort requirements, equipment operating parameters are within safe limits, and the energy consumption of each building does not exceed budget constraints; identifies buildings with redundant energy consumption and buildings with insufficient energy consumption in the building complex, the identification of buildings with redundant energy consumption including: solving the problem of each building under comfort constraints. The single-building energy consumption minimization problem yields the theoretically optimal energy consumption. When a building's current energy consumption exceeds a preset percentage of the theoretically optimal energy consumption, it is identified as an energy-redundant building. An energy-redundant building is one whose current energy consumption exceeds the minimum energy consumption required to maintain comfort. An energy-insufficient building is one whose current environmental parameters do not meet comfort requirements. A multi-building energy consumption optimization model is solved to generate equipment control commands for each building. These commands cause energy-redundant buildings to reduce equipment power and energy-insufficient buildings to increase equipment power, thereby minimizing the overall energy consumption cost of the building complex.

[0061] The multi-building energy consumption optimization model adopts a mixed-integer linear programming form. The decision variables are the power setpoints of each building and each device within a future time window. The objective function is the sum of the total energy consumption costs of all buildings within that time window, specifically calculated by multiplying the power of each device by its operating time and then by the electricity price for the corresponding time period, and summing the results over all buildings and all devices. The constraints include: the indoor temperature of each building must be maintained within the set comfortable temperature range, with a deviation not exceeding ±1 degree Celsius; the indoor illuminance must meet the working requirements and be no less than 300 lux; the power of each device must be within its rated power range; and the total energy consumption of each building must not exceed its budget constraint. The method for identifying buildings with redundant energy consumption is as follows: Construct an energy consumption minimization model for each building individually, with the decision variable being the power of each device in the building, the objective function being the total energy consumption of the building, and the constraints being comfort requirements and upper and lower limits of device power; solve the model using a linear programming solver to obtain the theoretical optimal energy consumption; compare the building's current actual energy consumption with the theoretical optimal energy consumption; if the current energy consumption is more than 20% higher than the theoretical optimal energy consumption, it is determined to be an energy-redundant building. The method for identifying buildings with insufficient energy consumption is as follows: check whether the current indoor environmental parameters meet the comfort constraints; if the temperature is more than one degree Celsius below the lower limit of the comfort temperature, or the illuminance is more than 50 lux below the required illuminance, it is determined to be an energy-insufficient building. Commercial linear programming solvers, such as CPLEX or Gurobi, are used to solve the multi-building energy consumption optimization model. To improve efficiency, the future time window is set to one hour, with a time granularity of 15 minutes, thus optimizing the equipment power settings for four time steps. Time-of-use electricity price information is obtained in real time from the power company's system. Depending on whether the current time is peak, normal, or off-peak, the corresponding electricity price coefficient is applied: 1.5 times the base price during peak hours, 1 time during normal hours, and 0.5 times during off-peak hours. The solved equipment power settings are the control commands for each building, which are then sent to the local controllers of each building for execution.

[0062] By optimizing energy consumption across buildings, the information silos of each building operating independently are broken down, achieving optimal overall energy consumption for the building complex while ensuring the comfort of each building. In particular, under the time-of-use pricing mechanism, energy consumption of non-critical buildings can be reduced as a whole during periods of high electricity prices, and energy consumption can be rationally allocated during periods of low electricity prices, significantly reducing operating costs. This solution achieves an improvement from single-building optimization to overall building complex optimization.

[0063] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A building data intelligent analysis and decision-making method based on the Internet of Things, characterized in that, include: Acquire spatial parameters, glass material parameters, and real-time environmental data collected by IoT sensors for each building; By combining the sun's position and spatial geometry, a building association map is constructed to identify the impact transmission chain; A control task list is generated based on the building association map. Each control task is decomposed into a source-side occlusion sub-task and a receiver-side compensation sub-task. Priorities are assigned to the sub-tasks and constraints are set. Based on the sub-tasks and constraints, query the equipment capacity list and equipment status association matrix of this building, determine the equipment control parameters, and use sequential execution control for strongly coupled equipment and parallel execution control for weakly coupled equipment. Collect equipment execution feedback data and environmental change data. When the deviation from the expected value exceeds the threshold, adjust the equipment control parameters and / or regenerate the control scheme. Record manual intervention behaviors and environmental parameters at the same time, and adjust the optimization parameters of the corresponding room; statistically analyze group intervention patterns, and trigger global strategy adjustment when the intervention ratio caused by the same optimization strategy exceeds the preset value.

2. The building data intelligent analysis and decision-making method based on the Internet of Things according to claim 1, characterized in that: The spatial parameters include floor height, building spacing, and three-dimensional coordinates; the building association map includes multiple floor nodes and directed influence edges between nodes, and the weight of the directed influence edges represents the influence intensity and influence time period of the source floor on the target floor; The steps to construct a building association map include: The spatial area of ​​each building is divided into grid cells, and nodes are created for each floor of each building. The node attributes include the floor center coordinates, glass area, and orientation angle. The solar altitude angle and azimuth angle are calculated at multiple time points within a preset time window. For each time point, the floor nodes with glass curtain walls are traversed as influence source nodes. The propagation direction of reflected light is calculated based on the solar incident angle and glass reflectivity. The target floor node reached by the reflected light is determined by ray tracing. When the reflected light energy reaching the target node exceeds a preset threshold, a directed edge is established between the influence source node and the target node. The weight of the directed edge is the calculated influence intensity. A set of directed graphs with time series is generated to form a building association map. The identification of the influence transmission chain includes: analyzing the multi-level transmission paths existing in the directed graph, and when the transmission relationship of the first node influencing the second node and the second node influencing the third node is identified, the second node is marked as a key transit node.

3. The intelligent analysis and decision-making method for building data based on the Internet of Things according to claim 1, characterized in that: The steps for generating a control task list based on a building association map include: Query the set of directed edges in the building association map where the influence intensity is greater than a preset threshold within a preset future time window; extract the source node, target node, influence intensity, and influence time period corresponding to each directed edge; obtain the number of affected rooms and historical user feedback records; calculate the task priority score based on the influence intensity, the number of affected rooms, and the historical user feedback records; sort the influence relationships according to the priority score to generate a task list containing the source node, target node, influence intensity, time window, and priority.

4. The intelligent analysis and decision-making method for building data based on the Internet of Things according to claim 1, characterized in that: The steps of decomposing each control task into source-side shading sub-tasks and receiver-side compensation sub-tasks include: for each task in the control task list, generating a source-side shading sub-task based on the type of the influencing source node, wherein the source-side shading sub-task targets the shading devices and / or dimming glass of the influencing source building; generating a receiver-side compensation sub-task based on the type of the target node, wherein the receiver-side compensation sub-task targets the lighting system and air conditioning system of the affected building; setting a first constraint for the source-side shading sub-task, wherein the first constraint includes that the indoor illuminance of the influencing source building is not lower than the lower limit of comfortable illuminance; setting a second constraint for the receiver-side compensation sub-task, wherein the second constraint includes that the increase in lighting power does not exceed a preset proportion of the base power and a limit on the increase in air conditioning load; and assigning execution priorities to each sub-task based on the impact intensity and affected range of the task.

5. The intelligent analysis and decision-making method for building data based on the Internet of Things according to claim 1, characterized in that: The steps for determining equipment control parameters include: reading the equipment state correlation matrix, which records the influence strength and delay time of the first equipment action on the state of the second equipment; extracting the equipment combinations involved based on the equipment control instructions to be executed, and calculating the correlation strength between the equipment; when the correlation strength between the equipment exceeds the strong coupling threshold, it is determined to be a strongly coupled equipment, and the control instructions of the strongly coupled equipment are arranged into sequential execution control, starting the master equipment first and monitoring its state changes, and starting the slave equipment after the master equipment state stabilizes; when the correlation strength between the equipment is lower than the weak coupling threshold, it is determined to be a weakly coupled equipment, and the control instructions of the weakly coupled equipment are arranged into parallel execution control, and control instructions are issued simultaneously; when the correlation strength between the equipment is between the strong coupling threshold and the weak coupling threshold, staggered execution control is adopted, and the execution time interval is set according to the influence delay time.

6. The intelligent analysis and decision-making method for building data based on the Internet of Things according to claim 1, characterized in that: The steps for adjusting equipment control parameters and / or regenerating control schemes include: collecting position sensor data from the shading device, illuminance sensor data from the lighting system, and temperature sensor data from the air conditioning system as equipment execution feedback data; collecting environmental sensor data from the affected area as environmental change data, including indoor illuminance, temperature, and user activity status; calculating the deviation between the equipment execution feedback data and the expected target value, and calculating the deviation between the environmental change data and the desired environmental parameters; determining the deviation type when any deviation exceeds a preset threshold; if the deviation is caused by inadequate equipment execution, adjusting the control parameters of the corresponding equipment to increase and / or decrease the equipment's action amplitude; if the deviation is caused by environmental interference and / or the mutual influence of multiple devices, recalculating the influence intensity, updating the building association map, and regenerating the task decomposition scheme and equipment control parameters based on the updated map.

7. The intelligent analysis and decision-making method for building data based on the Internet of Things according to claim 1, characterized in that: The steps for triggering global policy control include: counting the number of manual interventions by users in each room unit within a preset time period; classifying user manual intervention behaviors according to control policy type, and calculating the number of interventions and intervention ratios triggered by each control policy; when the intervention ratio of a certain control policy exceeds a preset intervention ratio threshold, extracting the triggering conditions, control parameters, and execution result data corresponding to that control policy; analyzing the environmental parameter characteristics and time distribution characteristics that trigger interventions, and adjusting the global trigger threshold, control intensity, and / or execution logic of the control policy; conducting trial runs of the adjusted control policy in some buildings, and collecting user intervention data and energy consumption data during the trial run; when the performance is better than the original policy, extending the adjusted control policy to all buildings.

8. A building data intelligent analysis and decision-making system based on the Internet of Things, characterized in that: The building data intelligent analysis and decision-making method based on the Internet of Things (IoT) for executing any one of claims 1-7 includes: The data acquisition module is used to acquire spatial parameters, glass material parameters, and real-time environmental data collected by IoT sensors for each building. The influence correlation modeling module is used to combine the sun's position and spatial geometric relationships to construct a building correlation map and identify influence transmission chains; The task decomposition module queries the building's equipment capacity list and equipment status association matrix based on sub-tasks and constraints to determine equipment control parameters. The scheme adjustment module is used to collect equipment execution feedback data and environmental change data. When the deviation from the expected value exceeds the threshold, the equipment control parameters are adjusted and / or the control scheme is regenerated. The behavior learning module is used to record user manual intervention behaviors and group intervention patterns, and dynamically adjust control parameters and global strategies.

Images