Building space scheduling system based on multi-source perception and decision optimization

Abstract

The invention relates to the technical field of buildings, and particularly discloses a building space scheduling system based on multi-source perception and decision optimization. The system comprises a multi-source sensing module, a data processing module, a decision optimization module and an execution module. The multi-source sensing module collects environment data, personnel data and equipment operation data; the data processing module performs cleaning, feature extraction and fusion on the multi-source data to generate structured data; the decision optimization module comprises a global optimization layer and a local optimization layer, formulates a macroscopic scheduling strategy and a specific scheduling strategy based on structured data prediction, and dynamically optimizes the macroscopic scheduling strategy and the specific scheduling strategy by adopting a machine learning model; and the execution module regulates and controls equipment and resources in the building space according to the macroscopic scheduling strategy and the specific scheduling strategy. According to the method, decision optimization and multi-source data perception are combined, and resource scheduling, energy efficiency management and user comfort of the building space are innovatively optimized.

Description

Technical Field

[0001] The present invention relates to the field of building management, and particularly to a building space scheduling system based on multi-source perception and decision optimization, belonging to the category of cross-application of building intelligence and energy management technologies. Background Art

[0002] With the continuous improvement of building intelligence level, the scheduling system of building space has gradually become an important direction to improve building operation efficiency and personnel experience. Existing building space scheduling systems usually rely on a single data source for equipment control, lacking the comprehensive perception ability of environmental changes, personnel activities and equipment operation status, resulting in problems such as unreasonable resource allocation, high energy consumption and low space utilization efficiency. At the same time, some building space scheduling systems rely on fixed rules in the process of formulating resource scheduling strategies, lacking the ability of adaptive optimization and being difficult to respond to the dynamically changing demands in the building space in real time. Therefore, there is an urgent need for a scheduling system that can integrate multi-source perception data and combine a decision optimization mechanism to realize the efficient, intelligent and adaptive management of building space resources. Summary of the Invention

[0003] The purpose of the present invention is to provide a building space scheduling system based on multi-source perception and decision optimization to solve the problems of single data source, unreasonable resource allocation, insufficient energy efficiency management and lack of dynamic optimization ability existing in the existing building space scheduling system. By integrating environmental data, personnel data and equipment operation data, and combining machine learning models for real-time prediction and scheduling optimization, the dynamic scheduling of equipment and resources in the building space is realized, thereby effectively improving the energy efficiency level, space utilization rate and personnel comfort.

[0004] According to the present invention, a building space scheduling system based on multi-source perception and decision optimization is provided, including: A multi-source perception module for collecting multi-source data in the building space, where the multi-source data includes environmental data, personnel data and equipment operation data, and dynamically adjusting the collection strategies of each monitoring unit through a self-organizing network cooperation mechanism to optimize data integrity; A data processing module connected to the multi-source perception module for preprocessing and feature extraction of the multi-source data to generate structured data; A decision optimization module connected to the data processing module, adopting a hierarchical decision-making architecture, where the hierarchical decision-making architecture includes a global optimization layer and a local optimization layer. The global optimization layer formulates a macro scheduling strategy based on the structured data and combines cross-space resource demand prediction; the local optimization layer formulates a specific scheduling strategy based on the structured data and combines real-time scenarios and context information; the decision optimization module uses a machine learning model to dynamically optimize the macro scheduling strategy and the specific scheduling strategy in combination with execution feedback; An execution module, connected to the decision optimization module, is used to perform real-time scheduling of the equipment and resources in the building space according to the macro scheduling strategy and the specific scheduling strategy, and feed back the execution results to the decision optimization module to verify the effectiveness of the strategy and trigger adaptive adjustment; Among them, the system supports collaborative scheduling of multiple building spaces, and realizes distributed decision-making and resource sharing among modules through a self-organizing network collaboration mechanism.

[0005] Furthermore, the multi-source perception module includes: An environment monitoring unit, used to collect temperature, humidity, light, and air quality data in the building space; A personnel monitoring unit, used to obtain personnel location, density, and behavior data; A device monitoring unit, used to monitor the operating status and energy consumption data of the devices in the building space.

[0006] Among them, the multi-source perception module, through a self-organizing network collaboration mechanism, determines the global scheduling requirements across spaces according to the macro scheduling strategy generated by the global optimization layer, and determines the local scheduling requirements of a single space according to the specific scheduling strategy generated by the local optimization layer, and dynamically negotiates the acquisition tasks and frequencies of the above monitoring units to optimize data integrity and real-time performance.

[0007] Furthermore, the data processing module includes: A data cleaning unit, used to denoise and process outliers in the multi-source data; A feature extraction unit, used to extract scheduling-related feature vectors from the cleaned multi-source data; A data fusion unit, used to fuse feature vectors from different sources to generate unified structured data, and the structured data is represented in the form of a multi-dimensional feature matrix.

[0008] Furthermore, the decision optimization module includes: A prediction sub-module, used to predict the resource requirements of the building space based on historical data and the structured data; An optimization sub-module, used to generate a scheduling strategy through the following hierarchical optimization process according to the resource requirement prediction of the prediction sub-module and in combination with the dynamically adjusted optimization objective: A global optimization layer, based on the structured data, generates a macro scheduling strategy through the following objective function: Among them, is the global optimization target value, B is the cross-space resource allocation efficiency, is the cross-space resource balance degree, λ, μ are dynamically adjusted weight coefficients, satisfying λ + μ = 1; The cross - space resource allocation efficiency B is calculated from the utilization rate of resource sharing, and the cross - space resource balance degree is determined by the variance of the resource allocation deviation of each building space resource.

[0009] The local optimization layer, based on the structured data, fine - tunes the macro - scheduling strategy through the following objective function to generate a specific scheduling strategy: where, is the local optimization target value, E is the energy consumption value based on the equipment operation data, U is the space utilization rate based on the personnel data, C is the personnel comfort score based on the environmental data, and α, β, γ are dynamically adjusted weight coefficients, satisfying α + β + γ = 1; wherein, the optimization objectives include the improvement of the cross - space resource allocation efficiency and the optimization of the resource balance degree of the global optimization layer, as well as the minimization of energy consumption, the maximization of space utilization rate, and the optimization of personnel comfort of the local optimization layer.

[0010] Furthermore, the machine learning model includes a deep neural network model or a reinforcement learning model, and the machine learning model is optimized through a training data set to improve the collaborative performance of the global optimization layer and the local optimization layer.

[0011] Furthermore, the execution module includes: The equipment control unit is used to predictively adjust the operation parameters of lighting, air - conditioning, and ventilation equipment in the building space according to the specific scheduling strategy generated by the local optimization layer; The resource allocation unit is used to dynamically adjust the cross - space resource allocation and usage rights according to the macro - scheduling strategy generated by the global optimization layer in combination with the real - time priority, and adjust the resource allocation in a single building space according to the specific scheduling strategy generated by the local optimization layer; wherein, the execution module adjusts the equipment operation state in advance through a predictive scheduling mechanism to reduce the response delay, supports cross - space execution synchronization according to the macro - scheduling strategy to coordinate the resource use of multiple building spaces, and feeds back the execution result to the decision - making optimization module to optimize the subsequent scheduling.

[0012] Furthermore, the building space scheduling system further includes: The communication module is used to realize data transmission between the multi - source perception module, the data processing module, the decision - making optimization module, and the execution module, as well as the interconnection and intercommunication with external systems; wherein, the communication module supports wired communication and wireless communication protocols.

[0013] Furthermore, the building space scheduling system further includes: A user interaction module, which is used to receive the scheduling requirements input by the user and feedback the scheduling results and the system operation status to the user. Among them, the user interaction module includes a display interface and a voice interaction unit.

[0014] Furthermore, the building space scheduling system further includes: An adaptive learning module, which is used to continuously optimize the performance of the machine learning model according to the operation data of the building space and user feedback; among them, the adaptive learning module supports online learning and offline learning.

[0015] Furthermore, the building space scheduling system further includes: A cloud service platform, which is used to store the multi-source data and scheduling strategies, and provide remote monitoring and data analysis functions; among them, the cloud service platform conducts data interaction with the system through an encrypted communication protocol.

[0016] Through the realization of multi-source data perception and decision optimization, the present invention effectively solves the problems of single data, scheduling lag, unreasonable resource allocation, high energy consumption, etc. in the prior art. Specifically, it can be summarized as the following points: 1. Improve the utilization efficiency of building space resources: By introducing a scheduling system based on multi-source perception and decision optimization, the present invention can schedule the equipment and resources in the building space in real time, ensure the reasonable configuration and utilization of resources, and thus improve the overall resource utilization efficiency. In particular, through the optimization of the cross-space resource allocation efficiency (B) and the cross-space resource balance degree (Eq), the phenomenon of waste or unbalanced utilization of building space resources is avoided; 2. Dynamically optimize the scheduling strategy: Through the hierarchical decision-making architecture of the global optimization layer and the local optimization layer, the system can dynamically adjust the scheduling strategy. The global optimization layer coordinates the resource allocation between different building spaces through the macro scheduling strategy, and the local optimization layer adjusts the resource scheduling strategy of a single building space according to the real-time scenario and specific requirements. The introduction of the machine learning model enables the strategy to be continuously optimized through execution feedback, improving the scheduling efficiency; 3. Enhance the system's adaptability: The system has an adaptive adjustment function, which can automatically optimize the scheduling strategy according to the real-time feedback data, ensuring the maximization of the scheduling effect at different times and scenarios. Through the self-organizing network collaboration mechanism, the system can achieve distributed decision-making and resource sharing among multiple building spaces, improving the resource scheduling collaboration efficiency of the entire building complex; 4. Improve the accuracy of data collection and processing: The introduction of the multi-source perception module enables the system to collect environmental data, personnel data, and equipment operation data in real time, ensuring the accuracy and timeliness of the data. The self-organizing network collaboration mechanism can optimize the negotiation of data collection tasks and frequencies, further improving the data integrity and real-time nature, ensuring that scheduling decisions can be made based on the latest and most comprehensive data information; 5. Optimize energy efficiency and comfort: By comprehensively optimizing multi-dimensional data such as the operating status of equipment, energy consumption, space utilization, and user comfort, the system can maximize energy conservation and improve the comfort of users while ensuring the efficient operation of the building space. Minimizing energy consumption, maximizing space utilization, and optimizing user comfort can effectively reduce energy consumption and enhance the sustainability of the building environment.

[0017] In summary, by combining decision optimization with multi-source data perception, the present invention innovatively optimizes the resource scheduling, energy efficiency management, and user comfort of the building space, having significant technical and application advantages. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the framework of a building space scheduling system based on multi-source perception and decision optimization provided by the present invention; Figure 2 It is a schematic diagram of the structure of the multi-source perception module provided by the present invention; Figure 3 It is a schematic diagram of the structure of the data processing module provided by the present invention; Figure 4 It is a schematic diagram of the structure of the decision optimization module provided by the present invention; Figure 5 It is a schematic diagram of the structure of the execution module provided by the present invention.

[0019] Reference Signs: A building space scheduling system 100 based on multi-source perception and decision optimization, a multi-source perception module 101, a data processing module 102, a decision optimization module 103, an execution module 104, an adaptive learning module 105, a communication module 106, a user interaction module 107, a cloud service platform 108; An environmental monitoring unit 1011, a personnel monitoring unit 1012, an equipment monitoring unit 1013, a data cleaning unit 1021, a feature extraction unit 1022, a data fusion unit 1023, a prediction sub-module 1031, an optimization sub-module 1032, an equipment control unit 1041, a resource allocation unit 1042. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] To make the objectives, technical solutions, and advantages of the present application clearer, the technical solutions in the present application will be clearly and completely described below with reference to the accompanying drawings in the present application. Apparently, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in the present application belong to the scope of protection of the present application.

[0021] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features; in the description of this application, unless otherwise specified, the meaning of "a plurality" is two or more.

[0022] To more clearly illustrate the technical solution of the present invention, the present invention will be described in detail below with reference to specific embodiments, but it should not be construed as a limitation on the protection scope of the present invention.

[0023] In this embodiment, as Figure 1 shown, a building space scheduling system 100 based on multi-source perception and decision optimization is provided, including a multi-source perception module 101, a data processing module 102, a decision optimization module 103, and an execution module 104. Through the close cooperation of each module, the system realizes the intelligent scheduling and optimal control of equipment and resources in the building space.

[0024] Specifically, the multi-source perception module 101 is used to collect multi-source data in the building space in real time. The multi-source data includes, but is not limited to, environmental data, personnel data, and equipment operation data. Environmental data includes parameters such as temperature, humidity, light intensity, and air quality, comprehensively reflecting the physical environment status in the building space; personnel data covers the number, location distribution, and activity trajectories of personnel, used to monitor the usage situation and personnel flow in the building space in real time; equipment operation data includes information such as the on / off status, working mode, and energy consumption of lighting, air conditioning, ventilation and other equipment, used to master the operation situation of building facilities.

[0025] It can be understood that the self-organizing network is used to support the collaborative perception tasks of the multi-source perception module, which is composed of multiple perception nodes with data collection and communication capabilities, and adopts a mesh topology structure to realize multi-hop communication and information sharing between nodes. Each perception node has the ability to adaptively adjust tasks based on local status, can determine the overall perception requirements according to the macro scheduling strategy generated by the global optimization layer, and at the same time dynamically negotiate the collection tasks and frequencies in combination with the specific scheduling strategy generated by the local optimization layer. During the operation of the network, the nodes can dynamically optimize the routing path and task allocation method based on factors such as historical communication quality, remaining energy, and task priority, improving the coverage rate and energy efficiency ratio of data collection. Without the premise of centralized control, this self-organizing network realizes the distributed collaboration of perception tasks and the rapid response of scheduling strategies, improving the real-time performance and flexibility of the system.

[0026] Further, the multi-source perception module 101 transmits the collected multi-source data to the data processing module 102 in real time via wired or wireless means. After receiving the raw data, the data processing module 102 first preprocesses the data. The preprocessing process includes processing steps such as noise removal, missing data completion, and data format unification to improve the quality and reliability of the data. The preprocessed data then enters the feature extraction stage to extract key features of the operation state of the building space, such as the temperature fluctuation range, personnel density change, equipment energy consumption curve, etc., to ensure the effectiveness of subsequent analysis.

[0027] The decision-making and optimization module 103 is connected to the data processing module 102, receives the structured data and conducts in-depth analysis. The decision-making and optimization module 103 generates and optimizes the building space scheduling strategy by introducing a global optimization layer and a local optimization layer. Specifically, the global optimization layer generates a macro scheduling strategy based on the overall demand prediction of all resources in the building space and conducts cross-space resource scheduling optimization. This layer focuses on the global configuration of resources and strives to maximize energy efficiency, optimize space use, and balance personnel comfort. The local optimization layer generates specific scheduling strategies based on real-time scenario data and context information and conducts micro-level scheduling adjustments, mainly focusing on the resource allocation of specific spaces, such as adjusting environmental parameters such as temperature and humidity in a single area to ensure the efficient utilization of equipment and resources in that space. Among them, the context information refers to auxiliary information that has a dynamic impact on resource prediction and strategy formulation during the building space scheduling process, mainly including but not limited to: current time (such as weekdays or holidays, time periods, etc.), space use type (such as office, meeting, rest, etc.), equipment operation status (such as whether the air conditioner is turned on, lighting mode, etc.), user behavior patterns (such as activity frequency, path distribution, historical behavior trajectories), and environmental mutation information (such as external weather changes or emergencies), etc. The context information is mainly processed and utilized by the local optimization layer in the decision-making and optimization module to improve the adaptability and accuracy of specific scheduling strategies. When formulating specific scheduling strategies, the local optimization layer can dynamically correct and optimize the target and strategy parameters based on the structured data combined with the context information in the real-time scenario, enabling the system to more accurately respond to the actual operation state and user needs of the current space, thereby achieving more flexible and refined scheduling control.

[0028] Further, at least one machine learning model is deployed inside the decision-making and optimization module 103. The machine learning model can combine historical data and real-time data and, through continuous training, predict the dynamic changes in the demand for building space resources. Based on the prediction results, the decision-making and optimization module 103 will dynamically generate the scheduling strategy for the building space, including the operation plan of equipment, space use arrangement, priority configuration of resources, etc., to ensure the comprehensive optimization of the building space in terms of energy efficiency utilization, comfort maintenance, and resource coordination.

[0029] Specifically, the execution module 104 is connected to the decision optimization module 103 and is used to perform real-time scheduling operations on the equipment and resources in the building space according to the macro scheduling policy and the specific scheduling policy. The equipment and resources include lighting equipment, air conditioning systems, fresh air systems, as well as adjustable space areas and energy units. Further, the execution module 104 determines the overall scheduling goal according to the macro scheduling policy, for example, giving priority to ensuring the energy supply in high-usage areas. Subsequently, corresponding control instructions are generated according to the specific scheduling policy and are sent to the control interfaces of relevant equipment in real time to trigger actions, such as adjusting the air conditioning temperature, adjusting the lighting brightness, or switching the resource allocation path. It can be understood that after the equipment completes the response operation, the execution module 104 collects the equipment operation status, space resource allocation data, and energy efficiency indicators during the corresponding time period as the execution result, and feeds back this execution result to the decision optimization module 103 for evaluating the adaptability and effectiveness of the current policy. Further, if the decision optimization module 103 determines that the policy effect is not ideal based on the feedback data, it automatically triggers the adaptive adjustment process of the scheduling policy to form a new optimal policy, thereby continuously optimizing the operation status of the building space.

[0030] Further, the building space scheduling system supports the collaborative scheduling of multiple building spaces. Scheduling subsystems with independent sensing, decision-making, and execution capabilities are respectively deployed in different building spaces. Each subsystem communicates and coordinates through the self-organizing network collaboration mechanism, thereby realizing cross-space dynamic resource scheduling and optimal control. Further, the self-organizing network collaboration mechanism is based on a peer-to-peer communication structure, allowing each subsystem to independently establish a communication connection without a central control node and negotiate resource usage plans according to local and global states. For example, when there is a load peak in a certain building space, it can request neighboring spaces to share redundant energy or reconfigure equipment tasks through the network negotiation mechanism to relieve local pressure. It can be understood that each subsystem shares sensing data, scheduling intentions, and execution feedback through this collaboration mechanism, thereby jointly participating in the distributed decision-making process, gradually approaching the globally optimal resource allocation plan, while retaining their respective control priorities for local resources, and realizing the system-level elastic and adaptive scheduling capabilities.

[0031] By realizing multi-source data perception and decision optimization, the present invention effectively solves the problems of single data, lagging scheduling, unreasonable resource allocation, high energy consumption, etc. in the prior art. This building space scheduling system collects multi-source data such as environment, personnel, and equipment operation through the multi-source perception module 101 and uses the self-organizing network collaboration mechanism to dynamically adjust the collection strategy to optimize data integrity. It generates structured data through preprocessing and feature extraction of the data processing module 101 to provide high-quality input for decision-making. Through the global optimization layer in the hierarchical decision-making architecture, combined with cross-space resource demand prediction, a macro scheduling strategy is formulated to achieve balanced allocation of resources in multiple building spaces. Through the local optimization layer, combined with real-time scenarios and context information prediction, specific scheduling strategies are formulated to improve the energy consumption efficiency, space utilization rate, and comfort of a single space. Through the machine learning model, combined with execution feedback, the scheduling strategy is dynamically optimized to achieve adaptive adjustment. Through the execution module 104, the equipment and resources are scheduled in real time according to the macro and specific scheduling strategies, and cross-space execution synchronization is supported to improve the collaborative efficiency of multiple building spaces, thereby significantly improving the resource utilization rate, energy efficiency, comfort, and multi-space collaborative scheduling ability of the building space.

[0032] In some embodiments, as Figure 2 shown, the multi-source perception module 101 further includes an environment monitoring unit 1011, a personnel monitoring unit 1012, and an equipment monitoring unit 1013. The environment monitoring unit 1011 is configured at multiple positions within the building space for real-time collection of environmental parameter data inside the space. Specifically, the environment monitoring unit 1011 can collect indicators related to temperature, humidity, light intensity, and air quality, such as carbon dioxide concentration, PM2.5 concentration, formaldehyde content, etc. Through multi-point sampling and comprehensive analysis, a complete perception of the environmental state of the building space is formed.

[0033] Furthermore, the personnel monitoring unit 1012 is deployed in the main channels, entrances and exits, and important areas of the building space for real-time detection of the position, distribution density, and behavior characteristics of personnel. It can be understood that the personnel monitoring unit 1012 can use infrared sensors, cameras combined with image recognition algorithms, radio frequency identification (RFID) technology, or Bluetooth beacons, etc. to achieve personnel detection. Through the personnel monitoring unit 1012, not only can the immediate position and quantity change of personnel in the space be accurately obtained, but also the personnel activity path, stay time, and behavior pattern can be inferred, thereby providing accurate human flow dynamic data support for subsequent intelligent scheduling.

[0034] Furthermore, the device monitoring unit 1013 is used to monitor the status of key operating devices in the building space in real time. It can be understood that the device monitoring unit 1013 can collect the start-stop status, operating mode, working duration, and energy consumption data of various devices. For example, for air conditioning equipment, the device monitoring unit 1013 can record its current temperature setting, wind speed gear, and power consumption; for the lighting system, it can collect the switch status and illuminance output; for the ventilation system, it can monitor parameters such as fresh air volume and fan speed. Through the device monitoring unit 1013, the operating conditions and energy consumption of various facilities and equipment inside the building space can be comprehensively grasped, providing accurate and detailed basic data support for subsequent data processing and optimization decisions.

[0035] Furthermore, during the operation of the system, the multi-source perception module 101 determines the global scheduling requirements across regions in the building space according to the macro scheduling strategy generated by the decision optimization module 103. Specifically, the macro scheduling strategy can be embodied as resource linkage goals across floors and functional areas, such as increasing the ventilation intensity in crowded areas and coordinating the adjustment of lighting brightness during peak energy consumption periods. The multi-source perception module 101 coordinates the monitoring units in multiple regions through a self-organization mechanism, and adjusts the acquisition range, frequency, and data upload method as needed to ensure the coverage breadth and timeliness of the global perception data, supporting the scheduling execution of the global optimization layer.

[0036] Furthermore, the multi-source perception module 101 simultaneously identifies the local scheduling requirements of a single space according to the scheduling strategy of the specific space. For example, before a certain meeting room is reserved for use, according to the predicted personnel density and activity time period, the local optimization layer will generate corresponding comfort guarantee strategies, involving the timing scheduling of the operation of devices such as lighting and air conditioning. The multi-source perception module 101 dynamically adjusts the sampling intervals of the personnel monitoring unit 1012 and the environmental monitoring unit 1011 accordingly, and preferentially allocates processing resources for high-frequency acquisition of key parameters, thereby improving the timeliness and resolution of local data.

[0037] It can be understood that while processing the global and local scheduling requirements, the multi-source perception module 101 realizes the dynamic negotiation of tasks between monitoring units through a self-organization mechanism. For example, when the performance of a sensing node in a certain area degrades due to power decline or channel interference, the system can automatically adjust the task coverage range of adjacent nodes to achieve temporary compensation for the sensing blind area. Through the task negotiation and frequency adjustment mechanism, it is ensured that the perception data maintains a high level in terms of spatial coverage and time continuity, effectively optimizing the integrity and real-time nature of the overall data, and providing high-quality perception support for the scheduling strategy of the system.

[0038] In some embodiments, such as Figure 3As shown, the data processing module 102 further includes a data cleaning unit 1021 for denoising and processing outliers of the multi-source data; and a feature extraction unit 1022 for extracting feature vectors related to scheduling from the cleaned multi-source data; and further includes a data fusion unit 1023 for fusing feature vectors from different sources to generate unified structured data, wherein the structured data is represented in the form of a multi-dimensional matrix or vector.

[0039] Specifically, the data processing module 102 works in coordination through multiple sub-units. First, the data cleaning unit 1021 performs denoising and outlier processing on the raw data obtained from the multi-source perception module 101. For example, the temperature sensor may generate noise due to external interference, and the data cleaning unit 1021 will remove these abnormal data to ensure that subsequent processing is based on accurate data. It can be understood that the data cleaning unit 1021 can use mean deviation detection, standard deviation analysis, or machine learning-based anomaly detection methods to identify and filter outliers in sensor data such as temperature and personnel density, thereby improving the accuracy and robustness of subsequent scheduling strategies.

[0040] Furthermore, the cleaned data is transmitted to the feature extraction unit 1022, which extracts key feature vectors related to the building space scheduling from the data. For example, environmental data such as temperature, humidity and air quality will be converted into feature vectors that affect comfort, personnel location and behavior data will be used to determine the frequency and density of space use, and equipment operation data can extract energy consumption levels and equipment failure characteristics. Then, the feature extraction unit 1022 converts these data into structured feature vectors according to certain rules to ensure that each data point can be effectively used.

[0041] It can be understood that the feature extraction unit 1022 can identify which data is most critical to the scheduling optimization and which data has less impact on the decision-making process.

[0042] Finally, the extracted feature vectors are transmitted to the data fusion unit 1023, where data from different sources are fused into unified structured data through an incremental fusion algorithm. Specifically, the data fusion unit 1023 combines feature vectors from multiple aspects such as environment, personnel, and equipment to form a multi-dimensional feature matrix. The goal of this process is to ensure that each input data is used reasonably and organized in a form that can support subsequent decision analysis, so as to ultimately form a clear data structure that can guide the scheduling of building space resources.

[0043] It is understandable that the incremental fusion algorithm is used to support the dynamic integration of real-time perception data in multiple building spaces, and has good scalability and timeliness. When the data is updated each time, the algorithm only performs feature extraction and vector fusion operations on the newly added data, without repeating the processing of the existing data, thus effectively reducing the computational redundancy and improving the system response efficiency. Specifically, after receiving the perception data from different spaces, the data fusion unit first performs standardization processing on the environmental parameters, equipment status, and space usage information of each space respectively, and extracts multi-dimensional feature vectors. Subsequently, an incremental update strategy is adopted to perform weighted integration of the newly added feature vectors at the current moment and the historical fusion results to generate unified structured data with a consistent structure.

[0044] This structured data is represented in the form of a multi-dimensional feature matrix, where the row vectors represent different data dimensions, such as temperature, humidity, illumination intensity, personnel density, etc., and the column vectors represent time evolution or spatial distribution, thus providing high-dimensional input data support for the subsequent prediction and optimization modules.

[0045] In some embodiments, as Figure 4 shown, the decision optimization module 103 further includes a prediction sub-module 1031 for predicting the resource requirements of the building space based on historical data and the structured data; an optimization sub-module 1032 for generating an optimal scheduling strategy according to the resource requirements predicted by the prediction sub-module 1031 in combination with the preset optimization objectives.

[0046] Specifically, in this embodiment, the prediction sub-module 1031 predicts the resource requirements of the building space based on historical data and structured data. First, the prediction sub-module 1031 receives data from the multi-source perception module, including temperature, humidity, light, personnel behavior data, and the operating status and energy consumption data of the equipment. By analyzing this structured data, the prediction sub-module 1031 can identify the resource requirement patterns in the building space, and then predict the resource requirement situation in the future period of time. Specifically, the prediction sub-module 1031 uses methods such as time series analysis and regression analysis to process the historical data, and combines the real-time input data for dynamic prediction. By establishing a prediction model, the system can take into account factors such as personnel activities, environmental changes, and equipment loads in different time periods, and accurately predict the energy consumption requirements, equipment operation requirements, and space utilization conditions of the building space, etc.

[0047] The implementation of a prediction model is usually based on the trend analysis of historical data, identifying patterns such as seasonal fluctuations and personnel activity patterns, and using machine learning techniques for further optimization. Common prediction algorithms include Support Vector Machine (SVM), Random Forest Regression, and Long Short-Term Memory (LSTM) networks. The prediction sub-module 1031 learns the changing trends of historical data through these algorithms and can perform personalized resource demand predictions according to the characteristics of different building spaces. For example, the system can predict the lighting and air-conditioning loads of a certain floor or area during a specific period, or predict the impact of changes in personnel density on the demand for the air-conditioning system.

[0048] Based on the resource demand prediction results provided by the prediction sub-module 1031 and combined with the dynamically adjusted optimization objectives, the optimization sub-module 1032 generates a scheduling strategy through a hierarchical optimization process. The task of the optimization sub-module is to generate a reasonable resource scheduling plan according to the changes in resource demands and different optimization objectives. In the optimization sub-module, first, according to the prediction results, the demand quantities of different resources in the building space are determined, and appropriate priorities are assigned to each resource. The optimization process is carried out through a two-level structure. The global optimization layer focuses on the allocation of cross-space resources, aiming to improve the sharing efficiency of resources between different building spaces and optimize the utilization rate and balance of cross-space resources. The local optimization layer focuses on the optimal allocation of resources within a single building space, especially the balance among equipment energy consumption, space utilization rate, and personnel comfort. The global optimization layer and the local optimization layer each guide the generation of the scheduling strategy through an optimization objective function.

[0049] Furthermore, in the global optimization objective function, represents the global optimization objective value, and λ and μ are dynamically adjusted weight coefficients in the optimization process. The form of the objective function is:

[0050] where B represents the cross-space resource allocation efficiency, which measures the configuration efficiency of resources in each building space by calculating the utilization rate of resource sharing. Specifically, based on the structured multi-dimensional feature matrix provided by the data processing module 102, the decision-making optimization module 103 calculates the ratio of the resource call volume in each space to the total available resources, thereby calculating the resource allocation efficiency. The improvement of resource allocation efficiency can ensure the reasonable sharing and full utilization of resources in each space, thus maximizing the overall resource utilization efficiency. It represents the cross - space resource balance degree, which measures whether the resource allocation is balanced by calculating the variance of the resource allocation deviation among each building space. The decision - making optimization module 103 calculates this variance based on the difference distribution between the utilization rate of each space and the overall average utilization rate. A smaller variance of resource deviation indicates that the resources are more evenly distributed among different spaces, avoiding the situation where some space resources are over - utilized while others are idle. The λ and μ in the objective function are dynamically adjusted weight coefficients, and the sum of these two coefficients is always equal to 1, that is, λ + μ = 1. By adjusting the values of these two coefficients, the relationship between resource allocation efficiency and resource balance degree can be balanced. In practical applications, the values of λ and μ can be dynamically adjusted according to the requirements of different scenarios. For example, when it is necessary to improve resource utilization rate, the value of λ can be increased; while in the scenario where resource balance degree needs to be optimized, the value of μ can be increased.

[0051] In the global optimization layer, by minimizing to generate a global scheduling strategy, which guides the overall resource scheduling within the building space to ensure the efficient use of cross - space resources.

[0052] Then enter the local optimization layer. In the local optimization layer, the optimization goal is to fine - tune the macro - scheduling strategy generated by the global optimization layer, so as to generate a specific scheduling strategy to meet the specific needs of each building space. In this layer, the form of the optimization objective function is:

[0053] where is the local optimization objective value, E is the energy consumption value based on the device operation data, U is the space utilization rate based on the personnel data, C is the personnel comfort score based on the environmental data, and the personnel comfort in the space is mainly calculated based on the environmental data (such as temperature, humidity, etc.). α, β, and γ are dynamically adjusted weight coefficients, and satisfy α + β + γ = 1. By adjusting these weight coefficients, the system can find the optimal balance among energy consumption, space utilization rate, and personnel comfort.

[0054] Understandably, the acquisition of the three parameters E, U, and C all depends on the raw data collected by the multi-source perception module 101 and the structured data output by the data processing module, and is calculated by the decision optimization module 103 in the global or local optimization layer based on this data. Specifically: E is the energy consumption value based on the device operation data, which is collected in real time by the device monitoring unit 1013 of the multi-source perception module 101, including information such as the power, operation duration, and start / stop times of each device. After the data processing module 102 cleans and extracts features from this data, it provides the energy consumption statistical result to the decision optimization module 103 in numerical form, and the decision optimization module 103 directly reads this result as the energy consumption value E; U is the space utilization rate based on the personnel data, which is collected by the personnel monitoring unit 1012 in the multi-source perception module 101, including information such as personnel density, entry / exit frequency, and stay duration. After the data processing module 102 fuses and normalizes this personnel behavior data, it calculates the ratio of the actual occupied area to the available area in each space and passes this ratio to the decision optimization module to determine the space utilization rate U; C is the personnel comfort score based on the environmental data, which is collected by the environmental monitoring unit 1011 in the multi-source perception module 101, including indicators such as temperature, humidity, light intensity, and air quality. After the data processing module 102 compares the environmental parameters with thresholds and performs weighted calculations, it outputs the comfort score, and the decision optimization module 103 directly calls this score as the personnel comfort score C. The calculation of the above three parameters is completed in the decision optimization module, ensuring that the optimization strategy can be dynamically adjusted and precisely controlled based on real perception data.

[0055] Specifically, the minimization of E helps to reduce energy consumption in the building space and improve energy use efficiency; the maximization of U helps to increase the utilization rate of the space, especially in areas with high personnel density, ensuring that resources can be utilized efficiently; and the maximization of C can improve the work or living comfort of personnel, especially in aspects such as air conditioning and lighting adjustment, which can be optimized according to actual needs.

[0056] In the local optimization layer, by minimizing and combining with the macro scheduling strategy generated by the global optimization layer to adjust the specific scheduling strategy. This process ensures that the resource allocation of each building space can not only meet the requirements of global optimization but also satisfy its respective local needs.

[0057] In addition, in the decision optimization module 103, it is also crucial to implement a feedback mechanism. The system continuously monitors and feeds back the device operating status, resource allocation effect, and energy efficiency data, and feeds this information back to the decision optimization module 103 to verify the effectiveness of the scheduling strategy and trigger adaptive adjustments. With the support of a machine learning model, this process enables the global and local optimization strategies to be continuously optimized according to the actual execution situation, improving the system's response speed and flexibility.

[0058] Such a system based on a hierarchical decision-making architecture can fully consider the various requirements of the building space and provide a dynamically adjustable and real-time responsive resource scheduling scheme through optimization algorithms. This not only ensures the efficient utilization of resources in each building space but also guarantees the sustainability and comfort of the system.

[0059] In some embodiments, the machine learning model adopted in the decision optimization module 103 may include a deep neural network model or a reinforcement learning model. Specifically, the deep neural network model is based on a multi-layer perceptron structure. The input layer receives the multi-dimensional structured feature matrix output by the data processing module 102. The intermediate hidden layer performs high-dimensional abstract extraction of the features through a non-linear activation function, and the output layer outputs the prediction results of resource requirements, such as the energy consumption load, personnel density distribution, or device usage trend in different regions in the future for a period of time. It can be understood that this deep neural network model is trained with a large amount of historical building space operation data, enabling the model to accurately learn the laws of resource requirement changes from complex and variable input data.

[0060] Furthermore, the reinforcement learning model takes the building space as the environment, the scheduling strategy as the action, and the state of the building space (such as the real-time energy consumption level, personnel distribution, and environmental comfort) as the state input. The system selects a scheduling action according to the current state and obtains a reward value based on the feedback after the action (such as energy consumption reduction, space utilization improvement, or personnel comfort improvement), and learns the optimal scheduling strategy through continuous trial and error and the reward mechanism. It can be understood that the reinforcement learning model can achieve the optimization of resource allocation through continuous adaptive adjustment in a dynamically changing building space environment.

[0061] Furthermore, to improve the generalization ability of the model and meet the requirements of different building space scenarios, the machine learning model is trained through a specially constructed training dataset. This training dataset covers a variety of different types of building space usage scenarios, such as office buildings, shopping malls, schools, and hospitals, etc., ensuring that the model can output reasonable prediction and decision results when facing different building characteristics.

[0062] Understandably, in order to achieve the synergistic effect of the global optimization layer and the local optimization layer, the machine learning model needs to share and coordinate information between the global and local levels. Specifically, the machine learning model exchanges information in the system to improve the overall scheduling efficiency. For example, a deep neural network model can provide predictions for the global resource allocation within the building space, while a reinforcement learning model dynamically adjusts the local resource allocation based on these predictions to achieve coordinated scheduling of cross-space resources.

[0063] In this way, the machine learning model can not only improve the global scheduling ability of the building space scheduling system, but also optimize the resource utilization of the local space, enabling the system to achieve efficient and precise scheduling in a complex environment with multiple building spaces and multiple resources.

[0064] In some embodiments, as Figure 5 shown, the execution module 104 further includes a device control unit 1041 and a resource allocation unit 1042.

[0065] Specifically, in this embodiment, the execution module 104 is used to perform real-time scheduling of the devices and resources within the building space according to the macro scheduling strategy and the specific scheduling strategy output by the decision optimization module, and feedback the execution result to the decision optimization module to verify the effectiveness of the strategy and trigger adaptive adjustment. The execution module 104 mainly includes a device control unit 1041 and a resource allocation unit 1042, and their functions complement each other to jointly achieve the optimal scheduling of resources and devices within the building space.

[0066] Specifically, the device control unit 1041 is responsible for performing predictive adjustment of devices such as lighting, air conditioning, and ventilation within the building space according to the specific scheduling strategy generated by the local optimization layer. The device control unit 1041 adjusts the operating parameters of the devices in advance according to the device operation requirements in the scheduling strategy to ensure that the operating state of the devices within the building space is consistent with the real-time scheduling requirements. The core of predictive adjustment lies in that the system can predict device requirements according to the specific scheduling strategy in the local optimization layer and make a response in advance, thereby reducing the response delay in the actual execution process.

[0067] For example, if the specific scheduling strategy generated by the local optimization layer indicates that the air conditioning demand in certain areas is high during a certain period, the device control unit 1041 will adjust the working state of the air conditioning device in advance, including adjusting the temperature setting, wind speed, and other operating parameters, so as to ensure that the required temperature environment can be provided quickly when needed. Similarly, lighting and ventilation devices will also be adjusted accordingly according to the scheduling strategy to achieve the optimization of energy efficiency and the maximization of personnel comfort.

[0068] The resource allocation unit 1042 is mainly responsible for cross-space resource allocation and resource usage adjustment within a single building space. The resource allocation unit 1042 dynamically adjusts the allocation and usage rights of cross-space resources based on the macro scheduling strategy generated by the global optimization layer and the real-time priority. In the case of multi-building space scheduling, the resource allocation unit 1042 coordinates resource sharing between building spaces to ensure the efficiency and fairness of overall resource allocation. The resource allocation unit 1042 also adjusts the allocation of resources within a single building space based on the specific scheduling strategy generated by the local optimization layer to achieve the optimal configuration of equipment and resources within the space.

[0069] For example, assuming that in a multi-building space environment, the equipment in a certain building space is in good operating condition, while the resource demand in another building space is high, the resource allocation unit 1042 will adjust the cross-space allocation of resources according to the global optimization strategy to ensure that the resource utilization of the entire system achieves the best effect. At the same time, the specific scheduling strategy of the local optimization layer will also be adjusted in real time within a single building space, for example, adjusting the energy consumption of lighting or air conditioning in a certain space to reduce the total energy consumption and improve the efficiency of space use.

[0070] During the execution process, the execution module 104 will also feed back the execution results such as the equipment operation status, resource allocation effect and energy efficiency data to the decision optimization module. This feedback mechanism is crucial to verifying the effectiveness of the scheduling strategy. If it is found during the execution process that some equipment fails to adjust according to the predetermined scheduling strategy, or the resource allocation of some building spaces is unreasonable, the execution module 104 will feed back this information to the decision optimization module, thereby triggering the system adaptive adjustment. The adaptive adjustment mechanism enables the entire system to dynamically optimize the scheduling strategy according to the actual execution results, and gradually improve the accuracy and effect of the scheduling.

[0071] In general, this embodiment enables the building space scheduling system to perform precise equipment control and resource allocation according to the macro and local scheduling strategies provided by the decision optimization module through the coordinated work of the equipment control unit 1041 and the resource allocation unit 1042. Through predictive regulation and dynamic resource adjustment, the execution module 104 realizes efficient scheduling of equipment and resources in the building space, and continuously optimizes the scheduling strategy through the feedback mechanism, thereby improving the system's adaptive ability and scheduling effect.

[0072] In some embodiments, Figure 1 As shown, the system also includes a communication module 106 for realizing data transmission among the multi-source perception module 101, the data processing module 102, the decision optimization module 103 and the execution module 104, and supporting interconnection with external systems.

[0073] Furthermore, the communication module 106 can adopt local area network communication, cellular mobile communication, Wi-Fi communication, Bluetooth communication, ZigBee communication, or other mainstream wired or wireless communication protocols to ensure stable, low-latency data transmission among various modules within the system and have a certain fault tolerance. It can be understood that the communication module 106 plays the role of an information center in the system architecture, not only carrying the upload of data collected by sensors inside the building space but also responsible for real-time sending of processing results and scheduling instructions to each execution unit, thus forming an end-to-end closed-loop scheduling link.

[0074] Furthermore, the communication module 106 also has a protocol conversion function, which is used to coordinate the communication protocol differences between devices of different manufacturers and achieve interconnection and interoperability of heterogeneous devices. For example, when accessing an external building automation system (BAS), energy management system (EMS), or video management system (VMS), the communication module 106 can achieve data interoperability with the external system through standard industrial communication protocol interfaces such as OPC UA, BACnet, Modbus, etc., improving the compatibility and scalability of the system. It can be understood that the communication module 106 schedules and controls the traffic of various communication tasks through the built-in communication management unit to ensure the stable operation of the system when multiple modules transmit data simultaneously and avoid data congestion or packet loss.

[0075] Furthermore, in terms of ensuring data security, the communication module 106 uses encryption communication protocols such as HTTPS, SSL / TLS, AES, etc. to encrypt the data interaction process within the system and between the system and external platforms, preventing data from being illegally intercepted, tampered with, or forged during transmission. The communication module 106 can also cooperate with authentication mechanisms, device authentication mechanisms, and access control mechanisms to ensure that internal system modules and externally connected systems are all trusted nodes, effectively preventing illegal access and control and enhancing the overall security protection ability of the system.

[0076] It can be understood that by introducing the communication module 106, efficient and stable data interaction can be achieved among the system's sub-modules, while also having good expansion capabilities and security guarantees, providing a solid foundation for the stable operation and function expansion of the building space scheduling system.

[0077] In some embodiments, as Figure 1 shown, the system further includes a user interaction module 107, which is used to receive the scheduling requirements input by the user and feedback the scheduling results and system operating status to the user.

[0078] Specifically, the user interaction module 107 includes a display interface and a voice interaction unit, which are used to provide a multimodal human-computer interaction method, enabling users to communicate with the system in the form of graphics, text, or voice. Understandably, the display interface can adopt the form of a touch screen, a mobile application interface, or a web-based visualization interface, intuitively displaying the real-time status of each area within the building space, including temperature, humidity, personnel distribution, equipment operation status, and the current execution scheduling strategy, etc., facilitating users to comprehensively grasp the system operation status.

[0079] Furthermore, users can customize scheduling preferences or input specific management strategy instructions through this display interface. For example, they can set a certain area to maintain a certain comfort level during a specific time period, or manually prioritize the allocation of certain resources to a specific area. The voice interaction unit, through natural language processing technology, realizes voice control and query of the system by users. For example, users can start system scheduling, query energy consumption reports, adjust resource allocation strategies, etc. through voice commands, improving the operation convenience and intelligence level. Understandably, the voice interaction unit can integrate a microphone array and a voice recognition processing chip, and combine a pre-trained semantic recognition model for command parsing and intention understanding, thereby achieving high-accuracy voice recognition and response.

[0080] Furthermore, to enhance the user experience, the user interaction module 107 also supports personalized settings and interaction history recording functions. The system can dynamically adjust the display content and interaction process according to the user's usage habits, feedback preferences, and historical input data, providing customized operation suggestions and scheduling recommendations. For example, for frequent users, the system can preferentially display their frequently used function modules in the interaction interface, and automatically complete and prompt for confirmation of repeated instructions in specific scenarios. Understandably, this module interacts with the decision optimization module 103 through a data interface, and real-time feeds back user input information to the system core decision logic, ensuring that user requirements can be quickly and accurately recognized and executed by the system.

[0081] Understandably, by setting up the display interface and the voice interaction unit, the user interaction module 107 effectively improves the operability and user-friendliness of the system, enabling users to conveniently and efficiently participate in the scheduling process and real-time control the operation status of the building space, providing crucial human-machine collaboration support for the intelligent management of the building space.

[0082] In some embodiments, such as Figure 1As shown, the system further includes an adaptive learning module 105, which is used to continuously optimize the performance of the machine learning model according to the operation data of the building space and user feedback. Specifically, the adaptive learning module 105 supports two modes: online learning and offline learning. Online learning means that during the actual operation of the building space, the system collects newly generated data in real time and dynamically updates the model parameters to adapt to environmental changes and changes in user behavior patterns. Offline learning means that during the non-peak operation period, the system centrally processes and deeply trains the accumulated large-scale historical data to comprehensively improve the performance of the model.

[0083] It can be understood that the adaptive learning module 105 synchronizes and integrates multi-source data collected in real time (including environmental change data, personnel dynamic data, equipment operation status data, etc.) with user interaction data (such as user satisfaction feedback on scheduling results, manual adjustment records, etc.) by establishing a data caching mechanism to form a new training sample set. Further, in the online learning mode, the module uses an incremental learning algorithm to fine-tune the existing machine learning model instead of completely retraining it, so as to ensure that the system maintains response speed and inference efficiency during continuous operation and avoid system downtime or performance degradation caused by frequent retraining.

[0084] Further, in the offline learning mode, the adaptive learning module 105 adopts a batch training method based on a complete data set and applies methods including but not limited to transfer learning, reinforcement learning, or deep neural network retraining to explore better feature expressions and decision-making logics in the building space resource scheduling strategy. It can be understood that to ensure data security and privacy protection during the offline training process, the adaptive learning module 105 also combines data desensitization processing, encrypted storage, and access control strategies to ensure that all operation data and user feedback meet data security standards during the training process.

[0085] Further, after the model is updated, the adaptive learning module 105 has a model evaluation and verification mechanism. By setting a set of key performance indicators (such as the proportion of energy consumption reduction, the increase in space utilization rate, the change in user comfort score, etc.), the updated model is tested and compared in multiple rounds. Only after ensuring that the new model is better than the old model in actual application can it be replaced into the decision optimization module 103 for formal application. It can be understood that the system also supports the model version management function, allowing rolling back to a previous model version with better performance when necessary to ensure system stability and continuous optimization of the scheduling effect.

[0086] It can be understood that setting the adaptive learning module 105 enables the system to dynamically evolve according to changes in the building space environment and user behavior, continuously improving the intelligent level and operation efficiency of building space resource scheduling, thus significantly improving the problems of slow response and poor adaptability of traditional fixed model systems.

[0087] In some embodiments, as Figure 1 shown, the system includes a cloud service platform 108 for storing the multi-source data and scheduling policies and providing remote monitoring and data analysis functions.

[0088] Specifically, the cloud service platform 108 interacts with the system through an encrypted communication protocol to ensure the security and integrity during data transmission and storage. It can be understood that the cloud service platform 108 has a highly available and highly scalable distributed architecture, supporting centralized management and unified analysis of data from different building space systems and providing collaborative optimization support for multiple buildings.

[0089] Furthermore, after the multi-source data is collected, it is preliminarily processed by edge nodes and then uploaded to the cloud service platform 108. The platform classifies and stores the data and labels it, constructing a time series data warehouse for subsequent big data analysis and model training. The scheduling policy is also uploaded to the cloud for scheduling historical traceability, optimization policy comparison, and cross-regional policy migration applications. It can be understood that the platform also provides a visual data analysis interface, and users or managers can access the platform through the web or mobile terminal to view the operating indicators, equipment status, and optimization effects of the building space.

[0090] Furthermore, the remote monitoring function supports the presentation and alarm functions of real-time data streams. When equipment anomalies, environmental parameter exceedances, or energy consumption surges occur, the system sends alarm messages to preset users through the cloud platform and automatically generates emergency scheduling suggestions. At the same time, the platform supports manual remote intervention in scheduling control to improve the flexibility and security of system response. It can be understood that the cloud service platform 108 also integrates an AI-based data mining engine, which can perform pattern recognition and trend prediction on long-term operation data to assist building managers in formulating energy consumption optimization strategies and operation and maintenance plans.

[0091] Furthermore, the encrypted communication protocol includes a combination of symmetric encryption and asymmetric encryption for the transmitted data. At the same time, through identity authentication mechanisms, access control, and data integrity verification means, it ensures that all data interactions during system operation comply with enterprise-level security standards, preventing data leakage, tampering, and illegal access. It can be understood that the cloud service platform 108 also supports deploying the trained model to edge devices to implement a cloud-edge collaborative intelligent scheduling policy deployment mechanism, thus taking into account the real-time nature of the system and the intelligence of the model.

[0092] Understandably, by setting up the above-mentioned cloud service platform 108, not only the data processing ability and remote management efficiency of the system are improved, but also the building space scheduling system is enabled to have the unified scheduling ability across scenarios and regions, meet the intelligent operation requirements of large-scale building clusters, effectively reduce the risk of single-point failures, and enhance the stability and scalability of the system.

[0093] The above are only exemplary embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural transformation made under the technical concept of the present invention by using the content of the specification and drawings of the present invention, or any direct / indirect application in other related technical fields is included in the patent protection scope of the present invention.

Claims

1. A building space scheduling system based on multi-source perception and decision optimization, characterized in that Including: A multi-source perception module for collecting multi-source data in a building space. The multi-source data includes environmental data, personnel data, and equipment operation data, and dynamically adjusts the collection strategies of each monitoring unit through a self-organizing network collaboration mechanism to optimize data integrity; A data processing module connected to the multi-source perception module for preprocessing and feature extraction of the multi-source data to generate structured data; A decision-making optimization module connected to the data processing module, adopting a hierarchical decision-making architecture. The hierarchical decision-making architecture includes a global optimization layer and a local optimization layer. The global optimization layer formulates a macro scheduling strategy based on the structured data and in combination with cross-space resource demand prediction; the local optimization layer formulates a specific scheduling strategy based on the structured data and in combination with real-time scenarios and context information; the decision-making optimization module uses a machine learning model to dynamically optimize the macro scheduling strategy and the specific scheduling strategy in combination with execution feedback; An execution module connected to the decision-making optimization module for real-time scheduling of equipment and resources in the building space according to the macro scheduling strategy and the specific scheduling strategy, and feeding back the execution result to the decision-making optimization module to verify the effectiveness of the strategy and trigger adaptive adjustment; Among them, the building space scheduling system supports collaborative scheduling of multiple building spaces, and realizes distributed decision-making and resource sharing among modules through a self-organizing network collaboration mechanism.

2. The building space scheduling system according to claim 1, wherein The multi-source perception module includes: An environmental monitoring unit for collecting temperature, humidity, light, and air quality data in a building space; A personnel monitoring unit for obtaining personnel location, density, and behavior data; An equipment monitoring unit for monitoring the operation status and energy consumption data of equipment in a building space. Among them, the multi-source perception module determines the global scheduling requirements across spaces according to the macro scheduling strategy generated by the global optimization layer through a self-organizing network collaboration mechanism, and determines the local scheduling requirements of a single space according to the specific scheduling strategy generated by the local optimization layer, and dynamically negotiates the collection tasks and frequencies of the above-mentioned monitoring units to optimize data integrity and real-time performance.

3. The building space scheduling system according to claim 1, characterized in that The data processing module includes: A data cleaning unit for denoising and outlier processing of the multi-source data; A feature extraction unit for dynamically selecting feature vectors related to the scheduling objectives of the global optimization layer and the local optimization layer from the cleaned multi-source data through hierarchical feature extraction; A data fusion unit for integrating feature vectors of different building spaces into unified structured data through an incremental fusion algorithm, and the structured data is represented in the form of a multi-dimensional feature matrix.

4. The building space scheduling system according to claim 1, wherein The decision-making optimization module includes: A prediction sub-module for predicting the resource requirements of a building space based on historical data and the structured data; An optimization sub-module for generating a scheduling strategy through the following hierarchical optimization process according to the resource requirement prediction of the prediction sub-module and in combination with dynamically adjusted optimization objectives: The global optimization layer generates a macro scheduling strategy based on the structured data through the following objective function: Among them, is the global optimization target value, B is the cross-space resource allocation efficiency, is the cross-space resource balance degree, λ and μ are dynamically adjusted weight coefficients, satisfying λ + μ = 1; the cross-space resource allocation efficiency B is calculated from the utilization rate of resource sharing, and the cross-space resource balance degree is determined by the variance of the resource allocation deviation of each building space. The local optimization layer fine-tunes the macro-scheduling policy based on the structured data through the following objective function to generate a specific scheduling policy: Among them, is the local optimization target value, E is the energy consumption value based on the device operation data, U is the space utilization rate based on the personnel data, C is the personnel comfort score based on the environmental data, and α, β, and γ are dynamically adjusted weight coefficients, satisfying α + β + γ = 1; Among them, the optimization objectives include improving the cross-space resource allocation efficiency and optimizing the resource balance degree of the global optimization layer, as well as minimizing energy consumption, maximizing space utilization rate, and optimizing personnel comfort of the local optimization layer.

5. The building space scheduling system according to claim 4, wherein The machine learning model includes a deep neural network model or a reinforcement learning model, and the machine learning model is optimized through a training data set to improve the collaborative performance of the global optimization layer and the local optimization layer.

6. The building space scheduling system according to claim 1, characterized in that The execution module includes: The device control unit is used to predictively adjust the operating parameters of lighting, air conditioning, and ventilation equipment in the building space according to the specific scheduling policy generated by the local optimization layer; The resource allocation unit is used to dynamically adjust cross-space resource allocation and usage rights according to the real-time priority in combination with the macro-scheduling policy generated by the global optimization layer, and adjust the resource allocation within a single building space according to the specific scheduling policy generated by the local optimization layer; Among them, the execution module adjusts the device operating state in advance through a predictive scheduling mechanism to reduce response latency, supports cross-space execution synchronization according to the macro-scheduling policy to coordinate the resource usage of multiple building spaces, and feeds back the execution result to the decision optimization module to optimize subsequent scheduling.

7. The building space scheduling system according to claim 1, wherein The system further includes: The communication module is used to realize data transmission between the multi-source perception module, the data processing module, the decision optimization module, and the execution module, as well as the interconnection and interoperability with external systems; among them, the communication module supports wired communication and wireless communication protocols.

8. The building space scheduling system according to claim 1, characterized in that The system further includes: The user interaction module is used to receive the scheduling requirements input by the user and feedback the scheduling result and the system operation state to the user. Among them, the user interaction module includes a display interface and a voice interaction unit.

9. The building space scheduling system according to claim 1, wherein The system further includes: The adaptive learning module is used to continuously optimize the performance of the machine learning model according to the operation data of the building space and user feedback; among them, the adaptive learning module supports online learning and offline learning.

10. The building space scheduling system according to any one of claims 1 to 9, characterized in that, The system further includes: The cloud service platform is used to store the multi-source data and scheduling policies and provide remote monitoring and data analysis functions; Among them, the cloud service platform conducts data interaction with the system through an encrypted communication protocol.

Images