LED scene adaptive intelligent control system based on Internet of Things multi-sensor fusion

Abstract

The invention discloses an LED scene adaptive intelligent control system based on Internet of Things multi-sensor fusion, and relates to the technical field of electric illumination light source systems. According to the LED scene adaptive intelligent control system based on Internet of Things multi-sensor fusion, tasks are received in real time for business grading and space partitioning, and the tasks are automatically shunted into different queues according to priorities and space regions; by integrating load operation state indexes, the load state is dynamically judged according to the indexes, and elastic scheduling and resource allocation are automatically driven. And finally, the scheduling sorting and issuing execution module performs comprehensive priority sorting according to grading and partitioning, intelligently allocates resources, and sets a fault-tolerant mode, thereby realizing service continuity of intelligent lighting, and effectively solving the problems that a system in the prior art is difficult to adapt to variable loads, high-concurrency tasks and sudden abnormal scenes, and is insufficient in overall elasticity and service continuity.

Description

Technical Field

[0001] This invention relates to the field of electric lighting source system technology, and in particular to an LED scene adaptive intelligent control system based on IoT multi-sensor fusion. Background Technology

[0002] In lighting systems distributed across multiple locations, including public places, large buildings, industrial parks, and urban roads, various sensors serve as the core of perception. By collecting data on people's status, environmental parameters, and behavioral signals in real time, environmental perception is connected via the Internet of Things (IoT). After multi-sensor fusion and decision-making, LEDs are driven by a standardized lighting bus or wireless network to dynamically adjust the brightness, color temperature, or switch lighting modes, achieving on-demand adaptation while considering energy saving, user comfort, safety, and scenario-based functional requirements. For example, by using a pedestrian flow sensor to detect pedestrian density, the brightness of LED lighting is increased when pedestrian flow is high to ensure safety, and the brightness is reduced when pedestrian flow is sparse to save energy. At the same time, it can be combined with emergency sensors to automatically switch to emergency lighting mode in emergency situations (such as fires). The LED intelligent lighting in each scenario follows a closed-loop logic of sensing, judging, and then controlling and adjusting, which reduces energy waste through dynamic adjustment and accurately matches lighting solutions according to scenario characteristics, forming a comprehensive intelligent application system covering various fields.

[0003] For example, Chinese Invention Patent CN107277968B discloses an intelligent control platform for LED lights in residential buildings, including a data server and user terminals. The data server includes a central control unit and a storage unit. The storage unit is used to authenticate login requests sent by user terminals. The user terminals include a login unit, a GPS positioning unit, and a human-machine interface. The LED light load terminals include a central control unit and unit building control modules. After obtaining the floor number sent by the data server, the central control unit sends an activation signal to the unit building control unit. The unit building control modules are used to turn the LED lights on and off on each floor.

[0004] For example, Chinese Invention Patent CN107734760B discloses an Internet of Things (IoT) lighting control system, which includes a server control terminal and multiple regional lighting networks. The regional lighting networks are connected to the Internet through signal base stations and communicate with the server control terminal. The regional lighting networks include multiple lighting devices, which form a wireless communication network. Each lighting device includes an LED light and a controller. The controller supplies power to the LED light and controls its switching. The multiple controllers are wirelessly connected to each other.

[0005] In existing technologies, intelligent lighting systems control lights through a main controller / cloud platform / regional gateway. Multiple sensors act as the "sensing front end" within the system, collecting real-time information on pedestrian flow, ambient light, temperature, humidity, and smoke. This data is then uploaded to a central control platform or local controller, allowing the system to intelligently determine current scenario requirements. For example, when a pedestrian flow sensor detects an increase in population density, it automatically increases light brightness to ensure lighting safety; when ambient light intensity is high, it automatically reduces LED brightness to save energy; temperature, humidity, and smoke sensors can be used for special scenarios or emergency responses, such as automatically switching to emergency lighting mode upon detecting a fire risk. This complete closed loop—sensor acquisition, intelligent decision-making, lighting control, and status feedback—enables the lighting system to adapt to changing environments, meet scenario-based requirements, save energy, and ensure safety.

[0006] However, existing lighting systems and control node configurations are typically designed based on average pedestrian traffic or peak daily usage, rather than being fully optimized for extreme peak scenarios to control costs and energy consumption. During occasional peak periods, this can lead to a lag where increased pedestrian traffic doesn't immediately adjust the lights. Furthermore, during peak times, the number of instructions that gateways, buses, wireless channels, or main control servers can process simultaneously is limited, causing all data and control signals to queue. With a large number of tasks entering the same entry point, queue lengths explode. For example, each sensor (such as pedestrian flow, environment, temperature, light intensity, access control, etc.) continuously reports its status. Each lighting node may receive commands such as "brighten / dim / switch mode" at any time. Especially in emergency situations, there may be extreme tasks such as "emergency switching of emergency lighting for the entire building / area". In this case, each task needs to queue up and wait for all the tasks in front to be processed before it can be scheduled or issued, causing congestion and delay. Ultimately, critical emergency signals cannot bypass the priority of ordinary traffic to reach the system. Current mainstream intelligent lighting control systems generally have problems such as difficulty in layering business and regional needs, rigid resource allocation and scheduling mechanisms, and weak local autonomy and anomaly recovery capabilities. The system is difficult to adapt to variable loads, high-concurrency tasks and sudden abnormal scenarios, and its overall elasticity and business continuity are insufficient. Summary of the Invention

[0007] To address the technical problems of existing technologies, such as the system's inability to adapt to varying loads, high-concurrency tasks, and sudden anomalies, resulting in insufficient overall resilience and business continuity, this invention provides an LED scene adaptive intelligent control system based on IoT multi-sensor fusion. This system includes: a task hierarchical partitioning module, a load operation elastic scheduling module, and a scheduling sorting and execution module. The task hierarchical partitioning module receives tasks (including events, instructions, planned tasks, telemetry, strategy outputs, etc.) to be scheduled or executed in real time. Based on the task type, source, and affected objects, it performs both business hierarchical classification and spatial partitioning, automatically diverting all tasks into different priority and partition queues (e.g., high-priority queue for critical areas, ordinary queue for ordinary areas, etc.). Business hierarchical classification determines the scheduling priority of each task, and spatial partitioning determines the area of ​​action of each task. The system includes modules for backup, grouping, and resource allocation; the elastic scheduling module for load operation is used to periodically collect key indicators across the entire link and analyze them to obtain a comprehensive load operation status index. The key indicators across the entire link are used to comprehensively and in real time reflect the health status of task queuing, communication resource occupation, and response in the entire scheduling process of the intelligent control system. The comprehensive load operation status index is used to dynamically determine the system load operation status and automatically drive elastic scheduling and resource allocation based on the comprehensive load operation status index. The scheduling sorting and execution module is used to automatically adjust resources based on the comprehensive priority score obtained from the hierarchical partitioning results of tasks. The resource allocator allocates bandwidth, time slots, computing power, and other resources to the tasks with higher priority in sequence to complete the task execution. Tasks in the critical area / high priority queue are quickly executed, while tasks in the ordinary area are automatically down-frequency, batched, frame-combined, or delayed according to the adjustment factor.

[0008] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: 1. By setting up a full-link multi-sensor fusion and hierarchical partitioning priority decision-making mechanism, the system can collect real-time sensor data of various types, including queue length, communication link occupancy, energy consumption, ambient illuminance, temperature, personnel flow, message round-trip latency, and resource utilization. This multi-source data is normalized and weighted to form a comprehensive load operation status index, which, combined with business classification scores and spatial partitioning weights, is ultimately fused into a globally unique comprehensive priority score. Because the system comprehensively perceives the environment and operational status, various lighting control scheduling tasks have significantly increased, and task types and business requirements have become more diversified, necessitating efficient and intelligent task scheduling and resource allocation solutions. Through multi-sensor full-link fusion, dynamic priority adjustment, and intelligent task scheduling, the system can still ensure rapid response and resource priority for critical areas and high-priority tasks even under conditions of large task volume, complex business, and changing scenarios. This significantly improves scheduling efficiency, lighting intelligence, and overall system operation quality, breaking through the technical bottlenecks of traditional static, single-parameter lighting control systems.

[0009] 2. By establishing an elastic scheduling and resource allocation mechanism driven by a comprehensive load operation status index, the allocation quotas of various resource pools, such as bandwidth, computing power, and scheduling cycle, can be automatically and dynamically adjusted according to real-time changes in the overall system load. When an increase in system load is detected, the task issuance cycle is automatically shortened, resource quotas are increased, and high-priority task scheduling is prioritized. When the system load decreases, some resources are dynamically released to improve resource utilization efficiency and reduce unnecessary energy consumption. This solution achieves real-time adaptive linkage between resource scheduling and system load, significantly reducing congestion and energy consumption caused by static quotas or over-scheduling, and realizing the synergistic effect of peak-hour elastic guarantee and off-peak high-efficiency energy saving in intelligent lighting systems.

[0010] 3. By establishing a task hierarchical partitioning and unified scheduling priority queue fusion mechanism, the system can identify the type, source, and target of each scheduling task in real time. Combining the spatial partition weights of key areas such as the main channel and emergency channel, as well as the hierarchical scores of different types of services, multi-dimensional task attributes are integrated into a unique comprehensive priority score for unified sorting and centralized distribution. This solution not only intelligently compares and sorts tasks such as "main channel emergency lighting tasks" and "normal area timed dimming tasks" within the same scheduling queue, but also automatically prioritizes tasks in key areas and postpones tasks in ordinary areas during emergencies or peak periods. This significantly reduces congestion and resource waste, ensuring layered protection of various lighting needs and the reliability of the overall system scheduling. By setting a resource allocation and priority linkage mechanism between scheduling sorting and execution, the system can automatically allocate resources to the adjusted resource pool based on real-time changes in the comprehensive load index and priority score, achieving dynamic allocation of scheduling resources from high to low. When resources are limited, high-priority tasks are issued first, while ordinary tasks are automatically batch-framed or postponed, effectively avoiding task loss or system blockage due to resource shortages. Meanwhile, the system has a task execution result feedback and closed-loop optimization mechanism, which can dynamically adjust scheduling parameters based on actual execution results, enabling self-learning and continuous optimization.

[0011] 4. By establishing a fusion mechanism between local autonomous offline execution and global unified scheduling, the system can automatically switch to local autonomous mode in extreme situations such as loss of connection between the central platform and partition control nodes, or network disconnection. Each partition independently collects its own multi-source sensor data, and autonomously completes the priority ranking, resource allocation, and scheduling of local tasks based on load index normalization, priority calculation, and resource allocation algorithms consistent with the global system. This ensures the continuous operation of core services such as lighting in critical areas and emergency channels, unaffected by global link issues. When the central platform or global scheduling system regains connectivity, the system automatically merges the accumulated scheduling results and resource allocation parameters of each partition through bidirectional synchronization and normalization mapping between local partition status and global load status. It then performs conflict resolution and parameter fusion based on task priority and actual business status, smoothly restoring to the global unified scheduling mode. Employing unified calculation logic and scheduling algorithms, the priority scores, resource allocation methods, and scheduling ranking rules of each partition remain consistent regardless of whether the system is in global centralized mode or local autonomous offline mode. This not only significantly reduces the risks of state transitions and parameter conflicts during the switching process, but also ensures that the system can achieve seamless continuity, state traceability and high availability of lighting services in any network environment and operating scenario. It greatly improves the overall robustness and security level of the system, which is superior to the problems of state inconsistency and service interruption caused by the separation of local and global and the inability to align parameters in traditional solutions. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic diagram of the structure of an LED scene adaptive intelligent control system based on IoT multi-sensor fusion provided in an embodiment of the present invention; Figure 2 A flowchart of concurrent scheduling and resource allocation provided in an embodiment of the present invention; Figure 3 A flowchart of the local autonomous mode provided in an embodiment of the present invention; Detailed Implementation The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0014] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0015] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0016] like Figure 1 The diagram shown is a structural schematic of an LED scene adaptive intelligent control system based on IoT multi-sensor fusion provided in an embodiment of the present invention. The system includes: a task hierarchical partitioning module, a load operation elastic scheduling module, and a scheduling sorting and execution module.

[0017] Among them are the task hierarchical partitioning module, the load operation elastic scheduling module, and the scheduling sorting and execution module; The hierarchical and partitioned task module is used to receive tasks to be scheduled or executed in real time. A task refers to a functional operation in the scheduling and execution process of the intelligent lighting system that has clear triggering conditions, target objects, execution actions, and feedback mechanisms. Clear triggering conditions mean that each task has its own preconditions for activation or triggering. For example, the triggering condition for a timed control task is a set time point; the triggering condition for a pedestrian sensing task is that a sensor detects people passing by; the triggering condition for an emergency response task is a fire alarm signal; and the triggering condition for a scene switching task is user operation or changes in environmental parameters (such as outdoor darkness). The target object is the specific execution object targeted by the task, which can be a single device, a group of area devices, or a logical scene, for example... The system controls all LED lights in a specific area, including lighting systems in key areas such as main corridors, stairwells, and emergency exits, as well as lighting equipment in general areas and public zones. Execution actions refer to the specific control operations required for the task, such as turning lights on / off, adjusting brightness and color temperature, switching lighting modes, starting / stopping emergency lighting, and implementing grouped lighting linkage. The feedback mechanism requires the system to obtain relevant status or execution results after task completion for closed-loop management and dynamic optimization. For example, it confirms whether lights are turned on or off as instructed and reports the status to the system; reports abnormal situations (such as light malfunction or communication failure); and provides feedback on energy consumption statistics and execution latency for subsequent scheduling and optimization.

[0018] Based on the type, source, and affected objects of each task, business classification and spatial partitioning are carried out respectively to achieve accurate identification of multi-dimensional attributes of tasks. The results of business classification and spatial partitioning are merged, and tasks are distributed to a unified scheduling priority queue based on the fusion results. The elastic scheduling module for load operation is used to periodically collect key indicators across the entire link and analyze them to obtain a comprehensive load operation status index. The key indicators across the entire link are used to comprehensively and in real time reflect the task queuing, communication resource occupation and response health status in the entire scheduling process of the intelligent control system. The comprehensive load operation status index is used to dynamically determine the system load operation status and automatically drive elastic scheduling and resource allocation based on the comprehensive load operation status index. The scheduling, sorting, and execution module is used to automatically adjust the allocation quota of various resource pools based on the comprehensive load operation status index, sort the resources according to the comprehensive priority score obtained from the hierarchical partitioning results of the tasks, and allocate the adjusted resources in order of priority from high to low until the resource pool is exhausted or all tasks are completed.

[0019] In this embodiment, while receiving tasks, a task hierarchy and partitioning system is introduced. Automatic two-dimensional traffic splitting based on business priority and spatial attributes enables real-time separation and scheduling of high-priority and low-priority queues. This effectively isolates emergency tasks from general tasks. In large public spaces such as airports and subway stations during peak passenger flow and emergency evacuations, priority is given to emergency lights and main passageway indicator lights, while traffic lights in ordinary areas automatically yield, greatly improving safety and management efficiency. Multiple indicators are integrated to derive a comprehensive load operation status index, serving as a dynamic decision-making basis for resource allocation, scheduling rhythm, and fault-tolerant switching. An adjustment factor is automatically output through a linear function to precisely control the resource quotas and scheduling frequency of each queue. In smart buildings or urban complexes, during holidays and periods of surge in passenger flow, the system can... Proactively upgrade resources in critical areas to effectively prevent business congestion and system downtime caused by sudden traffic surges. Through hierarchical partitioning and scheduling, high-priority queue tasks are automatically prioritized and allocated bandwidth, time slots, and computing power to enable the delivery of critical business tasks. The built-in real-time fault-tolerant triggering mechanism can automatically enter local autonomy and fault-tolerant resource mode based on events such as continuous over-thresholding of the comprehensive load index and communication loss. It automatically allocates most resources to critical businesses, while ordinary areas implement strong rate limiting, hierarchical caching, and delayed recovery to ensure that critical areas of the system remain operational and emergency services are uninterrupted. In extreme scenarios, such as fires, power outages, and emergency evacuations, critical areas and high-priority queue services can receive the fastest response and resources, while ordinary areas automatically degrade and elastically give way, improving scenario security and business continuity.

[0020] Furthermore, based on the type, source, and affected objects of each task, business classification and spatial partitioning are performed to achieve accurate identification of multi-dimensional task attributes. The results of business classification and spatial partitioning are then merged, and tasks are distributed to a unified scheduling priority queue based on the merged results, including the following steps: The system receives tasks that need to be scheduled or executed, parses and extracts key information fields for each task to obtain the task type, source, and affected objects. The type can be timed control, emergency response, maintenance and repair, daily linkage, etc., and supports multiple predefined types and flexible expansion. The source can be local sensors, user commands, central platform, third-party interfaces, etc., and is compatible with multi-source input and cascaded sources. The affected objects refer to the specific physical area, equipment group, or logical partition where the task is applied (such as main channel lights, all floors, specific areas, or combinations of multiple partitions), and support single-zone or multi-zone affiliation.

[0021] Based on the task type and source, and according to the preset type and priority rule table, the system automatically matches the business level and obtains the corresponding business level score. The priority rule table is dynamically maintained by the system administrator or intelligent algorithm, supporting online adjustment and adaptive optimization, which facilitates rapid adaptation to new business scenarios and changes in management strategies. If a task type / source is not defined in the rule table, the system can use the default priority or automatically assign priority based on the similarity with known types to ensure that all tasks have available business level scores. The rule table supports joint determination of type and source. For example, "fire alarm + local sensor" can take precedence over "fire alarm + remote command".

[0022] Based on the affected object field, the spatial partition mapping table is queried to determine the physical space or logical area where the task operates. The space is then classified and each partition is assigned a spatial weight score. Each spatial partition (such as critical area, ordinary area, special area, etc., the number of partitions here can be determined according to the actual situation; more partitions are needed for complex scenarios, and more partitions can be set for fine control) is assigned a weight score based on factors such as security importance and business sensitivity. Critical areas (such as main channels and emergency exits) usually have a higher weight than ordinary areas. Multi-level partitioning and dynamic weight adjustment are supported. For tasks involving multiple areas, priority can be given to primary and secondary areas, weighted average, or maximum weight can be used to ensure that spatial attributes reasonably reflect the importance of the task. The weight configuration can be dynamically adjusted according to actual business needs.

[0023] The business level score and the spatial partition weight score are weighted and merged to generate a comprehensive priority score for the task; The fusion formula is: Q = ωp × P + ωa × A, where Q is the overall priority score, P is the business level score, A is the spatial weight score, and ωp and ωa are configurable weight coefficients. These weight coefficients can be flexibly set or dynamically optimized based on business scenarios and historical scheduling results, enabling fine-tuning of priority sensitivity under different scenarios. Based on actual operational performance (such as SLA compliance rate, average response latency in critical areas, energy consumption / task ratio, etc.), the weights can be periodically adjusted through online optimization algorithms. A constrained adaptive learning rate is used, and a weight adjustment period and jitter suppression threshold are set to avoid frequent fluctuations in the priority strategy. The weight adaptation process can employ constrained subgradient descent or Bayesian optimization and supports manual intervention and rollback.

[0024] Based on the overall priority score of each task, tasks are automatically sorted and distributed to the corresponding priority queues from highest to lowest priority. If multiple tasks have the same overall priority score, they are further sorted according to their arrival time, type, source, or preset rules to ensure fair and orderly queue scheduling. The system dynamically monitors queue length and resource usage, and can limit, delay, or batch process low-priority tasks when necessary to ensure the scheduling timeliness and responsiveness of high-priority tasks. The "area / partition" and other attributes of each task are retained for subsequent partition statistics, resource allocation, or scheduling strategy optimization. A unified priority order (single logical queue) is maintained at the logical layer, while multiple physical execution queues are maintained at the execution layer according to area / partition; the scheduling order is based on logical priority. The system determines the priority of tasks based on their level and assigns them to the corresponding physical execution queue. Furthermore, it supports a closed-loop, dynamic, and adaptive adjustment of the entire process of task classification, partitioning, and queue sorting. When changes occur in the scene, or when task status or execution results are fed back to the system, the system can automatically reassess the overall priority of tasks and adjust the queue order in real time. It supports dynamic migration and priority optimization of task queues, ensuring efficient, flexible, and intelligent scheduling and resource allocation. The system continuously collects and analyzes the operational performance of the scheduling queues, providing data support for subsequent weight coefficient adaptation and priority rule optimization, thus achieving continuous evolution and efficient closed-loop operation in intelligent lighting scenarios.

[0025] In this embodiment, by setting up a multi-dimensional key information automatic parsing mechanism for task type, source, and affected objects, compared to the existing technology that only relies on a single sensor signal or simple timing logic for scheduling, it can distinguish different types of tasks such as timed switching, emergency response, energy-saving strategies, and pedestrian detection in real time in multi-point scenarios such as large parks and urban roads. This ensures that every sensor event and manual command can be accurately classified and processed by the system, significantly improving the intelligent perception and control capabilities in multi-source heterogeneous scenarios. A priority rule table for joint mapping of type and source is set up, and online maintenance and self-learning optimization are supported. Compared to the existing technology that uses static priority or manual parameter tuning, in LED lighting applications of large buildings or urban roads, emergency tasks, such as those triggered locally by fire sensors, can be automatically prioritized over ordinary timed lighting tasks, ensuring real-time response to critical safety events under multi-service concurrency, while significantly reducing the manpower burden of system maintenance and upgrades. Based on the spatial partitioning weight allocation mechanism for affected objects, compared to the existing technology that can only roughly distinguish areas or tilt resources according to a single rule, in actual park or building management, key areas such as main passages, entrances and exits, and emergency passages can be automatically assigned higher priority. Resource weighting prioritizes lighting tasks in areas experiencing peak traffic during holidays or emergencies, ensuring these areas receive priority access to system resources such as bandwidth and computing power. This significantly improves the precision of scene security and energy consumption zoning management. A weighted fusion algorithm combining business tier scores and spatial zoning weight scores generates a unique comprehensive priority score. Compared to existing hierarchical scheduling or single-priority sorting methods, this approach enables intelligent comparison and dynamic sorting of tasks such as "main channel emergency lighting tasks" and "regular area timed dimming tasks" within the same global sorting queue in real-world IoT lighting scenarios. This allows the system to comprehensively consider tasks at any given time. Prioritizing tasks based on urgency and regional importance, the system dynamically ensures that tasks in high-risk scenarios and critical lighting areas are prioritized. A unified scheduling queue management mechanism based on comprehensive priority scores, compared to existing technologies that use distributed multi-queue or simple cyclic scheduling, enables unified sorting and centralized distribution of all scheduling tasks in large-scale lighting control systems such as smart cities and smart parks. This ensures that high-priority tasks such as those on main roads and emergency exits always receive priority response and resources in high-concurrency, high-traffic scenarios, while tasks in ordinary areas are automatically delayed or batch-combined. This significantly reduces scheduling congestion and resource waste, improving the efficiency and reliability of global intelligent lighting control.

[0026] Furthermore, the analysis yields a comprehensive load operation status index, which is obtained by automatically collecting key indicators of the entire process of scheduling within a set time, normalizing the key indicators of the entire process so that all indicators fall under the same statistical standard, assigning weights to each indicator, and summing the normalized indicators according to their weights to obtain the comprehensive load operation status index. The comprehensive load operating state index S is obtained by the following formula: in, It is the i-th normalization index. It is its weight, N is the total number of all participating indicators, and the comprehensive load operation status index S is usually in the range of 0 to 1. The larger the comprehensive load operation status index, the higher the system load.

[0027] Key end-to-end metrics include queue lengths at all levels (reflecting task backlog), communication link occupancy (e.g., wireless channel, bus utilization), round-trip time (RTT), resource utilization (CPU, memory, bandwidth, etc.), packet loss rate, retry rate, and successful delivery rate. These metrics provide the foundational data support for operational monitoring, elastic scheduling, adaptive adjustment, and fault tolerance. The number of key metrics can be set according to actual needs, flexibly tailored to the system's scenario, complexity, business priorities, and implementation requirements. Initially, primary metrics (e.g., queue length, link utilization, response latency) can be selected. If new bottlenecks or anomalies are identified after system deployment, more key metrics can be added, such as energy consumption, number of abnormal alarms, and lighting status in specific areas. Special metrics can also be temporarily introduced for high-priority assessment during emergency responses.

[0028] In addition, all collected indicator data are normalized to eliminate the influence of different units and value ranges. Here, the minimum-maximum normalization method or Z-score normalization is used. For negative indicators that are "worse the larger the value", the original value is directly used for aggregation after normalization, so that high values ​​directly increase the total load index. For positive indicators that are "better the larger the value", the complementary value (1-normalized value) is taken. In this way, high values ​​actually lower the total load index. That is, positive indicators that "perform well" have a "minus" effect on the overall pressure. This makes the direction of influence of all indicators on the comprehensive index consistent. After this processing, regardless of whether the original parameters are "high and good" or "high and bad", the monotonicity of all indicators on the comprehensive index is consistent, which is convenient for weighted summation and subsequent threshold discrimination.

[0029] In this embodiment, the essence of the comprehensive load operation status index is to weight and integrate a set of core indicators that reflect the current scheduling / operation pressure of the system. Based on the three indicators of queue length, link occupancy rate, and RTT, other indicators and quantities of the comprehensive load operation status index can be flexibly configured, added, deleted, or combined according to actual business needs, system architecture, and operating characteristics. By setting a weighted fusion mechanism of multiple key indicators across the entire link, such as queue length, link occupancy rate, and RTT, compared with the existing technology that relies on single or fixed indicators for scheduling, it achieves comprehensive, real-time, and dynamic perception of system health status and load pressure, avoiding blind spots and misjudgment problems, and significantly improving the system's operational transparency and the scientific nature of scheduling decisions. Furthermore, by utilizing the fluctuation characteristics of the comprehensive load index, it can automatically and accurately adjust the allocation of core resources such as bandwidth, computing power, and time slots for high-priority and low-priority tasks according to the actual load situation. This achieves controllable pressure and adequate emergency support during peak hours, and energy efficiency optimization and rational resource utilization during off-peak hours, greatly improving the global resilience and efficient operation capability of the intelligent lighting system in multiple scenarios.

[0030] Furthermore, the allocation quotas of various resource pools are automatically adjusted based on the comprehensive load operation status index, including the following steps: Calculate the difference between the comprehensive load operating status index and the set threshold in real time; For each type of resource, a global adjustment factor is automatically output through a preset configurable mathematical function. Based on the adjustment factor of each type of resource, the quota allocated to each type of resource from the total system resource pool is automatically and dynamically adjusted. This ensures that when the load pressure increases, the corresponding resource pool quota is increased to meet the processing needs of high-concurrency tasks and critical businesses. When the load pressure decreases, the resource quota is automatically reduced. The resources include, but are not limited to, bandwidth, time slots, and computing power. The entire adjustment process is automated, continuous, smooth, and requires no manual intervention, thereby improving the system's resilience and the ability to ensure optimal resource availability during peak periods.

[0031] In this embodiment, besides bandwidth, time slots, and computing power, which are the three types of resources that can be scheduled in this way, storage resources, energy quotas, wireless channels / spectrum, security policy quotas, and AI inference or data circulation quotas can also be scheduled using the same method according to the actual situation. In the embodiment based on bandwidth, time slots, and computing power scheduling, storage resources, energy quotas, wireless channels / spectrum, security policy quotas, and AI inference or data circulation quotas are all objects to be scheduled. That is, when the overall load operation of the task scheduling subsystem is not good, it will borrow resources from other subsystems or upper-level systems to ensure normal task scheduling. When the system detects that the current comprehensive load operation index exceeds the set threshold, indicating increased system resource pressure and potential anomalies such as local bottlenecks or resource shortages, the system will automatically trigger a multi-resource collaborative scheduling mechanism. Based on the real-time adjustment factor of each type of resource, it not only dynamically increases the allocation quota of the local resource pool, but also requests / borrows resources across subsystems, modules, and even from higher-level systems or other physical nodes. This ensures the continuity of task scheduling and the overall elasticity of the system. The dynamic adjustment and borrowing strategies for various resources can be flexibly configured according to actual scenarios, business priorities, and policy rules, as well as setting a list of borrowable resources and upper / lower limits. For example, in critical moments, priority can be given to ensuring AI inference capabilities and data flow bandwidth, or wireless spectrum and energy quotas can be temporarily expanded during peak periods to achieve on-demand elastic expansion of resources. This solution not only achieves adaptive adjustment of local multi-resource pools, but also supports dynamic collaborative scheduling of resources across subsystems / higher-level systems. When local resources are strained and scheduling pressure increases, the system can automatically borrow external resources and temporarily increase the quota of a resource pool, achieving "elastic expansion of the resource pool." This effectively prevents business congestion, delays, or interruptions caused by a single bottleneck. The collaborative dynamic scheduling mechanism for multiple types of resources is particularly suitable for IoT scenarios such as large-scale smart lighting combined with distributed AI inference. Whether it's responding to sudden surges in traffic or batch events, or strengthening security strategies and scheduling energy consumption in extreme scenarios, the system can accurately prioritize critical resources for main business and emergency links when system load fluctuates, ensuring the continuous and secure operation of core system functions.

[0032] Furthermore, for various resources such as bandwidth, time slots, and computing power, a global adjustment factor is automatically output through a preset configurable mathematical function. Based on this global adjustment factor, various resources are dynamically allocated and adjusted. The specific process is as follows: the difference between the comprehensive load operating status index and a set threshold is used as input to the preset configurable mathematical function; the function automatically calculates the global adjustment factor for each type of resource based on its different resource types and sensitivity parameters; the baseline quota is adjusted according to the latest global adjustment factor to obtain the allocation value for each type of resource within the actual scheduling cycle; and a preset linear function is used to generate adjustment factors for various resources such as bandwidth, time slots, and computing power. The configurable mathematical function drives the allocation direction and magnitude of resources. The adjustment function form for each resource can be configured independently; this function can be linear, piecewise, exponential, or other forms to match its optimal response characteristics to load changes.

[0033] This embodiment adaptively adjusts the three main factors—bandwidth, time slot, and computing power—all using linear models (taking the factor's change with the load exponential as an example): ; Where R represents any type of resource (such as bandwidth BW, time slot TS, computing power CPUT, etc.). This represents the global adjustment factor for resource R (dimensionless, determining the adjustment range of resource quota relative to the baseline within the current cycle). The sensitivity coefficient of resource R is represented (positive or negative depending on the resource attribute, such as a negative number for time slots), and △S is the difference between the comprehensive load operation status index of this round and the set threshold.

[0034] The specific resource quota is as follows: ; in, This represents the absolute value of resources actually allocated within the current scheduling period (this value has specific units, such as Mbps, ms, number of cores, etc.). This indicates the baseline quota / baseline allocation value for the resource.

[0035] Furthermore, the resource adjustment factor is dynamically adjusted using a formula, and different resources can be independently configured with sensitivity according to actual business needs. Bandwidth, time slot, and computing power adjustments do not interfere with each other, ensuring that each type of resource can cope with the current load pressure. Global bandwidth adjustment factor: ; in The bandwidth normalization global bandwidth adjustment factor is a dimensionless proportional coefficient used to dynamically adjust bandwidth allocation quotas. The bandwidth sensitivity coefficient, a positive value, represents the system's response strength to load changes. To reflect the degree of change in current load pressure by comparing the overall load operating status index with a preset threshold, when the system detects a significant increase in key indicators such as scheduling queue length, communication occupancy, and task processing latency (i.e., When the value is greater than 0, the system is in a high-load state. Increased bandwidth allocation, when At that time, the allocated bandwidth decreases.

[0036] The specific corresponding bandwidth allocation is as follows: ; in, This represents the actual bandwidth resources allocated to the system during the current scheduling period. This is the bandwidth baseline quota, which is the basic bandwidth resource allocated each cycle under normal circumstances.

[0037] Using the above adjustment factor formula, the system can adjust the bandwidth allocation in real time according to the current overall load status. Under high load or high concurrency task pressure, Positive value indicates bandwidth adjustment factor. Automatically increasing bandwidth, the system instantly upgrades the bandwidth quota for the main channel, critical queues, or emergency services, effectively reducing command congestion, communication delays, and data packet loss risks, ensuring real-time response capabilities for critical tasks, and providing support during off-peak hours or when system pressure is low. Negative, bandwidth adjustment factor The system automatically reduces and reclaims redundant bandwidth allocations, allocating resources to other modules or optimizing power consumption, thereby further improving overall energy efficiency and network stability and avoiding unnecessary resource waste.

[0038] Global time slot adjustment factor: ; in, This represents the normalized global time slot adjustment factor, which is dimensionless and dynamically adjusts the ratio of the scheduling cycle to the baseline. A value less than 1 indicates a shorter cycle and faster scheduling. This represents the time slot sensitivity coefficient, which is negative. The larger the absolute value of the adjustment factor, the more sensitive the adjustment is to load changes. This is the difference between the comprehensive load operating status index and the threshold. The specific corresponding time slot allocation is as follows: ; in, This indicates the actual time slot after adjustment, i.e., the actual scheduling cycle length used in this cycle. Indicates the baseline time slot, the system's default scheduling period; By introducing a global time-slot adjustment factor, dynamic adaptive adjustment of scheduling system parameters is achieved. This factor automatically adjusts the scheduling cycle, the number of polling cycles within each cycle, and the concurrent capacity of each scheduling based on the system's real-time load status, task queue length, and the response requirements of critical services. Specifically, in scenarios with high load, high concurrency, or requiring emergency response, the adjustment factor shortens the scheduling cycle, increases the scheduling frequency, and appropriately increases the number of tasks that can be processed in parallel per scheduling, thereby significantly improving the overall throughput and task processing speed of the system and ensuring that critical tasks can obtain scheduling resources first. When the system load is low, the adjustment factor moderately extends the scheduling cycle, reduces unnecessary scheduling frequency and concurrency, reduces the ineffective consumption of system resources, and thus improves overall energy efficiency. The entire adjustment process adopts step-by-step adjustment, hysteresis threshold, and change rate limit to ensure the stability of system operation and avoid frequent parameter fluctuations. In addition, the solution also sets up priority guarantee and minimum service quota mechanisms to ensure that low-priority tasks can also obtain basic resources under high load and will not be "starved" due to long-term adjustment. Through this global, parameterized adjustment method, the system can flexibly adapt to real-time business needs, achieving rapid response during peak periods and energy consumption optimization during idle periods, thus balancing business flexibility and resource utilization.

[0039] Global computing power adjustment factor: ; in, Represents the normalized global computing power adjustment factor, which is the adjustment ratio (dimensionless) of the total computing power resource pool within the current scheduling cycle. This represents the computing power sensitivity coefficient; a positive value indicates the sensitivity of computing power to changes in system load. This is the difference between the comprehensive load operating status index and the threshold.

[0040] The specific corresponding computing power allocation is as follows: ; in, This represents the total adjusted computing power resource pool allocated to task scheduling within the current scheduling cycle. The unit can be the number of allocated CPU cores, threads, or memory capacity, etc. The baseline computing power quota refers to the basic computing power resource pool capacity that can be allocated when the system is in normal or under normal load.

[0041] Through a global computing power adjustment factor, the system can dynamically adjust the total allocation based on real-time load pressure. When a surge in high concurrency, high-priority tasks, or emergency events is detected, the computing power pool automatically expands, and the scheduling system immediately calls on more CPU cores / threads to participate in scheduling and processing, significantly reducing processing queuing and congestion. This ensures the system's processing capacity and real-time performance during peak business periods and emergency response scenarios. The expanded computing power resource pool is allocated preferentially within the queue according to the priority or weight of each task. High-priority / critical tasks receive more computing power for concurrent execution and faster response times, while ordinary tasks are appropriately delayed, fully reflecting resource priority and efficient utilization. In scenarios such as smart lighting, emergency security, and park automation, this ensures that core tasks are always prioritized and processed quickly. During off-peak or low-load conditions, the computing power adjustment factor decreases, and the system automatically shrinks the total allocation, reducing CPU core and thread usage, lowering energy consumption, equipment wear and tear, and maintenance costs, achieving green and energy-saving operation. At the same time, it provides greater flexibility for sharing the computing power resource pool across multiple business scenarios during off-peak periods, maximizing the overall system's energy efficiency and resource utilization.

[0042] like Figure 2 The diagram shows a concurrent scheduling and resource allocation flowchart provided in this embodiment of the invention. In this flowchart, various resource pools are adjusted and allocated according to the real-time load status. All tasks to be scheduled are concurrently scheduled. During the scheduling process, if a high-priority, emergency, or sudden task is detected, the system will prioritize the allocation of resources and schedule it. For queued tasks that have not yet been scheduled, the system will reserve a certain amount of polling resources for them and monitor the status changes of all tasks in the queue in real time through a periodic queue scanning mechanism. When it is found that the status of a queued task changes to a high-priority, emergency, or sudden task, its priority score is immediately dynamically increased and it is scheduled to the high-priority queue. Otherwise, the original queue is maintained.

[0043] Furthermore, when tasks are scheduled concurrently based on the adjusted resources in descending order of priority, the method also includes inputting the difference between the comprehensive load operation status index and the set threshold into the resource tilt factor function to obtain the resource tilt factor. The priority of concurrently scheduled tasks can be adjusted by a resource tilt factor so that critical tasks in the scheduling process can be given priority. The specific priority is adjusted through a resource tilt factor. Specifically, the system weights the original priority score of each task in the concurrent scheduling queue with this resource tilt factor; that is, it multiplies the original priority of each task by the resource tilt factor to generate an adjusted priority weight. This adjusted priority weight is used for actual resource allocation decisions, ensuring that high-priority, critical tasks receive a larger proportion of available resources under high load or emergency conditions, while ordinary tasks receive less resource allocation. This achieves dynamic tilting of resource allocation within the queue, improving the scheduling priority of critical tasks and the overall emergency response capability of the system.

[0044] The specific formula is as follows: ; in, The priority queue factor changes dynamically with the system load. This represents the sensitivity coefficient of the resource skew factor. It is a positive value and can be set to distinguish different queue weights. The higher the load, the larger the resource skew factor, and the more pronounced the skew effect. This is the difference between the comprehensive load operating status index and the threshold. The adjusted priority is: ; in, This represents the adjusted priority score, used for the actual resource allocation weight within the concurrent queue. This is the original priority score of the i-th task, which is also the priority score of each resource calculated in the previous embodiment.

[0045] In the case of original concurrent tasks, the system directly allocates the total resources in the resource pool proportionally based on the original priority score of each task. The resource share received by each task is only related to the proportion of its original priority score to the sum of the priority scores of all tasks. Under this allocation method, the proportion of resources received by high-priority tasks and low-priority tasks depends entirely on their respective priority scores. When the priority scores of all tasks differ significantly, high-priority tasks receive more resources, and low-priority tasks receive fewer resources; if the priority scores are similar, the resource allocation tends to be more even. This method is suitable for scenarios where the load does not change much or the importance of tasks is relatively stable, but it lacks the ability to adapt to dynamic sudden events and changes in system pressure.

[0046] The adjusted resource allocation for different priorities is as follows: ; This represents the resources allocated to the i-th task. This represents the total amount of resources that can be allocated within the current scheduling period. The resources here are the same as those in the previous embodiments, and can be bandwidth resources, time slot resources, computing power resources, storage resources, etc., where i=1...n, and n is the total number of concurrently scheduled tasks.

[0047] After correction, using the resource skew factor, the system first dynamically calculates the resource skew factor based on the difference between the current comprehensive load operating status index and the threshold, and then uses this factor to weight and adjust the original priority score of each task. The weighted priority score determines the proportion of resources actually allocated to the task.

[0048] When the system load is high or a sudden emergency task occurs, the resource skewing factor significantly amplifies the priority weight of critical tasks, further increasing their allocation ratio in the resource pool. This ensures that high-priority or emergency tasks can obtain more system resources under high concurrency or emergency conditions. This approach improves the system's responsiveness to sudden loads and dynamic scenarios, achieves adaptive optimization of critical business resource allocation, and effectively avoids resource waste or untimely response of critical tasks caused by average allocation.

[0049] For queued tasks that have not been scheduled, reserved polling resources are allocated. Through a periodic queue scanning mechanism, periodic event listening and important status change detection are achieved for all queued tasks. This ensures that once a high-priority, emergency, or sudden event is triggered, its priority score is updated and increased, thus preventing critical tasks from being missed or delayed in the queue.

[0050] The system reserves a portion of resources from the total resource pool for each scheduling cycle; it can also directly use the total resource pool available for scheduling in the current cycle, with the remainder serving as the total resources for parallel processing.

[0051] ; This indicates a polling resource pool reserved for queuing task scanning and task listening. This represents the total amount of resources that the system can allocate during this period. The polling resource ratio coefficient is obtained by real-time monitoring of indicators such as queue task load, overdue rate, and sudden task pressure, combined with business experience and adaptive algorithms. 0 < α < 1, which can be configured more according to actual conditions.

[0052] All queued tasks are periodically polled. When a queued task is found to have changed its status to high priority, emergency, or sudden during the inspection process, the system immediately and dynamically increases its priority score and schedules it to a high priority queue. This enables timely detection and priority processing of critical tasks, effectively preventing response delays caused by queuing omissions.

[0053] like Figure 3The diagram shows a flowchart of the local autonomous mode provided in an embodiment of the present invention. In this flowchart, the communication status between the monitoring space partition and the central control platform is monitored. If the communication is normal, it continues to be in the global scheduling mode; if the communication is abnormal, it automatically switches to the local autonomous mode, continuously collects local load and resource status, and dynamically adapts task scheduling and resource allocation. When all historical tasks have been processed, the adjustment factor and task order are dynamically adjusted. If there are unfinished tasks, the system will continuously monitor and process the task queue, and periodically collect local task, load parameters and resource status.

[0054] Furthermore, it also includes fault-tolerant mode, also known as local autonomous mode. The specific process is as follows: continuously monitor the communication status between each spatial partition and the central control platform. When it is detected that effective communication with the central platform has been lost, it immediately and automatically switches to local autonomous mode, disconnects all external control services and data channels, and achieves logical isolation from the central platform.

[0055] Once in autonomous mode, each local partition or regional controller operates independently, merging all tasks in the queue before the disconnection into a local autonomous scheduling pool, and removing expired, duplicate, or invalid tasks, retaining only the set of tasks that still need to be scheduled.

[0056] For all tasks to be scheduled in the current scheduling pool, collect their original priority scores and the initial values ​​of adjustment factors for various resources before disconnection.

[0057] The original priority scores and historical adjustment factors of all tasks to be scheduled are normalized.

[0058] The normalized priority score, adjustment factor, and local remaining resources are used as initial parameters to serve as the allocation baseline for the autonomous period.

[0059] ; in, This represents the normalized priority score (ranging from 0 to 1). This represents the original priority score for the i-th task. This represents the maximum priority score of all tasks in the current scheduling pool. This represents the minimum priority score of all tasks in the current scheduling pool.

[0060] Task scheduling is dynamically and adaptively adjusted based on normalized results and local real-time data collection to reflect the comprehensive load status.

[0061] During local autonomous operation, local task, load parameters, and resource status are periodically re-collected. After historical tasks are completed, adjustment factors and task sorting are dynamically adjusted based on real-time load and task changes.

[0062] In one specific embodiment, a large smart park is suddenly hit by extreme weather (such as typhoons or rainstorms) late at night, causing the communication links between the central control platform and several zones (such as the main road and the eastern area of ​​the park) to be interrupted, making it unable to receive centralized instructions from the global scheduling platform. At this time, the main road is experiencing nighttime patrols and peak employee departure times, placing high demands on the real-time performance and safety of road and safety passage lighting. Each autonomous zone independently and continuously collects various local sensor data, including illuminance, energy consumption, current, temperature, zone queue length, and equipment status. Following globally unified normalization and weighting standards, the multi-source data is fused into a zone-wide comprehensive load status index, accurately reflecting the local real-time lighting pressure and task demands. Various lighting tasks within the zone (such as main road patrol lights, roadside safety guidance lights, and decorative lights around the park) are automatically fused, scored, and sorted according to parameters such as business classification (critical, important, general), spatial zone weight, and task queue, based on a globally consistent comprehensive priority scoring system. This forms a local task queue under autonomous status. The zone automatically adjusts local bandwidth, computing power, time slot period, and other resource pool parameters based on the current load index. Critical tasks (such as emergency lighting and main passage patrol light control) receive priority in resource allocation and command issuance, while ordinary tasks (such as beautification lighting, etc.) receive priority. Low-priority timed lighting control automatically degrades, delays, or merges batch processing, achieving flexible resource allocation and intelligent scheduling optimization. Real-time closed-loop feedback of zone-based autonomous scheduling results ensures continuous emergency lighting online. If a lighting fixture exhibits abnormal response or energy consumption, the system automatically fine-tunes scheduling parameters and priorities. Once communication with the central platform is restored, the zone-based autonomous status and global scheduling status are automatically normalized and synchronized. Scheduling history and parameters (priority, resource allocation, task feedback, etc.) for each zone are batch-transmitted back. The system integrates local and global parameters using a globally unified algorithm, prioritizing critical business and security events in case of conflicts, ensuring data consistency and smooth scheduling transitions without interruption or "jumps." Because both autonomous and global systems use the same calculation logic and parameter system, parameters can be smoothly migrated and states seamlessly integrated, avoiding scheduling conflicts and lighting anomalies caused by the separation of traditional zone and global algorithms. The autonomous zone module can independently collect data, fuse multiple factors, sort locally, and schedule flexibly in extreme scenarios such as network outages and disconnections, ensuring uninterrupted lighting in critical areas and achieving high reliability of emergency lighting at the park level. It can also smoothly integrate with global recovery, ensuring consistent status and continuous tasks, greatly improving the robustness, business continuity, and engineering adaptability of the intelligent lighting system, surpassing the traditional autonomous mode that relies solely on simple local backups or static parameters.

[0063] During local autonomous operation, the system continuously collects task queues, resource availability, and operational load parameters of the current local partition at set periodic intervals. After historical tasks are completed, the system dynamically adjusts various adjustment factors and task priority ranking based on the latest collected local task distribution, resource pool status, and partition load index. The processes of adjustment factor calculation, priority normalization, and weighted resource allocation under local autonomy all use the same calculation logic and allocation formulas as the global scheduling phase. The only difference is that the data source scope is switched from the entire network or central system to collection and analysis within the local partition. The objects scheduled and allocated are limited to the tasks and resource pools of the current local autonomous partition.

[0064] This embodiment significantly improves the adaptive scheduling capability and resource utilization efficiency of distributed autonomous partitions by adopting the same adjustment factor calculation, priority normalization, and dynamic resource allocation algorithms as global scheduling in the local autonomous mode. Even in extreme scenarios such as network outages and isolated operation, local partitions can still achieve precise task queuing by priority and dynamic resource allocation based on real-time load, maximizing the response timeliness of critical business and emergency tasks. Compared with traditional autonomous modes that rely solely on fixed rules or simple polling, this system's local autonomous scheduling mechanism achieves seamless integration and algorithm consistency between global and local scheduling logic, effectively avoiding problems such as algorithm fragmentation and resource allocation imbalance. In multi-partition autonomy, each partition can automatically adjust its resource allocation scheme according to its actual load and task priority, greatly improving the system's business continuity, resilience, and security capabilities in large-scale city and park scenarios.

[0065] Based on real-time load and task changes, the adjustment factors and task sorting are dynamically adjusted. The specific process is as follows: Key indicators of the partitions are collected at set time intervals, and the partition load operation status index is obtained through analysis.

[0066] The partition priority score is obtained by analyzing the partition task.

[0067] Collect information on the availability of resources in the data collection area.

[0068] Based on the difference between the partition's comprehensive load operation status index and the threshold, the adjustment factors for various types of resources are calculated using preset configurable mathematical functions, and the actual quota of the partition resource pool is dynamically adjusted.

[0069] Based on the priority score of each task, a resource allocation weight is determined from the sum of the priorities of all tasks. The dynamically adjusted resource quota is then allocated accordingly, and the task queue is sorted.

[0070] After obtaining the normalized priority score, the priority score for each task will be... Substituting into the above formula: the adjusted resource allocation for different priorities is as follows: This allows us to obtain the allocation of local resources with different priorities as follows: ; in, This represents the amount of resources actually allocated to the i-th task in the local autonomous system. It can be any resource type that needs to be dynamically allocated, such as bandwidth, computing power, time slots, energy consumption, and storage. The specific allocation method is basically the same as the global resource allocation rules. This represents the normalized priority score. This represents the total amount of resources that the local autonomous (or primary / secondary scheduling unit) can actually control within the current period. In special scenarios such as network outages, local autonomy, offline mode, or primary system rate limiting, it is less than the global resource pool, representing the currently available portion. n represents the total number of tasks (or services, jobs) to be allocated in the current scheduling pool.

[0071] This embodiment normalizes the priority scores of all scheduled tasks (such as events, instructions, and periodic jobs), stretching or compressing the priority score of each task to a uniform standard range (e.g., [0,1]). Then, combining historical adjustment factors and current resource allocation factors, the final priority of each task is dynamically adjusted. During final allocation, the actual resource quota received by each task is proportional to its normalized and weighted priority, ensuring that the overall resource pool allocation process is adaptive, flexible, and fair.

[0072] By normalizing priorities, the resource allocation imbalance caused by excessive differences in original priorities between tasks or distortions in historical weights is effectively eliminated. This ensures that high-priority tasks can still obtain quotas commensurate with their importance, even under extreme load or resource-scarce scenarios, thereby improving the response speed and stability of critical business and emergency tasks.

[0073] By adopting a proportional normalization allocation mechanism, dynamic coupling between the resource pool and task load is achieved. Compared with the traditional "static weight" or "round-robin average" allocation method, this system can flexibly adjust the allocation scheme according to real-time task distribution, resource pool dynamics, and historical adjustment factors. In complex environments such as multiple IoT scenarios, edge autonomy, and network outage emergencies, this significantly reduces the risk of resource waste and business congestion, ensuring the overall elasticity and availability of the system.

[0074] Furthermore, upon detecting a loss of effective communication with the central control platform (main server, cloud, or master control node), the system immediately enters offline autonomous mode. In this mode, the division of autonomous regions can be flexibly determined based on the actual physical network structure, edge node deployment, and interconnectable physical links. For example, spatial units with interconnectivity capabilities such as local bus, LAN, and short-range wireless (e.g., WiFi, Zigbee, LoRa) can be automatically grouped into an autonomous partition, enabling interconnected autonomy and collaborative scheduling within the partition. For more granular autonomous requirements, the system supports edge sensors or smart terminals to run autonomous control logic independently in short-term isolated states based on their built-in computing and storage capabilities. They can still collect local data, determine task priorities, and adaptively allocate resources according to simplified rules. This allows even if some edge nodes are completely disconnected, core tasks can still be scheduled and responded to independently in a short time. During autonomous mode operation, each autonomous region needs to periodically attempt to communicate with the main server. Once network recovery and main server availability are detected, the system immediately synchronizes and merges key data accumulated during local autonomy, such as task status, resource adjustment factors, and task priority scores, based on normalization formulas and global parameters. Specifically, the system first maps locally normalized scheduling parameters (such as priority scores and resource allocation factors) to the global standard through normalization / denormalization operations, enabling seamless switching and dynamic access between autonomous regions and the central platform. After synchronization, all tasks and resource management are reintegrated into the global scheduling system, ensuring the integrity and business continuity of the entire IoT system. This mechanism significantly enhances the system's resilience and security in extreme scenarios (such as backbone network outages, regional disconnections, and isolated edge environments). Furthermore, it enables a smooth transition and automatic fusion of global and local scheduling parameters after network recovery, minimizing risks such as task loss, resource fragmentation, or business interruption, and comprehensively improving the system's self-healing capabilities and long-term operational reliability.

[0075] The above embodiments can be implemented in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0076] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An adaptive intelligent control system for LED scenes based on IoT multi-sensor fusion, characterized in that, include: The module includes a task hierarchical partitioning module, a load-bearing elastic scheduling module, and a scheduling sorting and execution module. The task hierarchical partitioning module is used to receive tasks that need to be scheduled or executed in real time. The task refers to a functional operation with clear triggering conditions, target objects, execution actions and feedback mechanisms during the scheduling and execution process of the intelligent lighting system. Based on the type, source and affected objects of each task, business hierarchical and spatial partitioning are performed respectively to achieve accurate identification of multi-dimensional attributes of tasks. The results of business hierarchical and spatial partitioning are merged, and the tasks are diverted to a unified scheduling priority queue based on the fusion results. The elastic scheduling module for load operation is used to periodically collect key indicators across the entire link and analyze them to obtain a comprehensive load operation status index. The key indicators across the entire link are used to comprehensively and in real time reflect the task queuing, communication resource occupation and response health status in the entire scheduling process of the intelligent control system. The comprehensive load operation status index is used to dynamically determine the system load operation status and automatically drive elastic scheduling and resource allocation based on the comprehensive load operation status index. The scheduling, sorting, and execution module is used to automatically adjust the allocation quota of various resource pools based on the comprehensive load operation status index, sort the tasks according to the comprehensive priority score obtained from the hierarchical partitioning results, and schedule tasks in order of priority from high to low based on the adjusted resources until the resource pool is exhausted or all tasks are completed.

2. The LED scene adaptive intelligent control system based on IoT multi-sensor fusion as described in claim 1, characterized in that, The process involves classifying and spatially partitioning tasks based on their type, source, and affected objects to accurately identify multi-dimensional task attributes. The results of the classification and spatial partitioning are then fused, and tasks are routed to a unified scheduling priority queue based on this fusion. This includes the following steps: The system will receive tasks that need to be scheduled or executed, parse and extract the key information fields of each task, and obtain the task type, source and affected objects. Based on the type and source of the task, and according to the preset type and priority rule table, the business level is automatically matched and the corresponding business level score is obtained. Based on the affected object field, query the spatial partition mapping table to determine the physical space or logical region where the task operates, and classify the space and obtain the spatial weight score assigned to each partition. The business level score and the spatial partition weight score are weighted and merged to generate a comprehensive priority score for the task; Based on the overall priority score of each task, tasks are automatically sorted and distributed to the corresponding priority queues from highest to lowest priority.

3. The LED scene adaptive intelligent control system based on IoT multi-sensor fusion as described in claim 1, characterized in that, The analysis yields a comprehensive load operating status index, specifically: The system automatically collects key metrics across the entire scheduling process at set intervals, including but not limited to: queue length, number of critical business tasks, and resource utilization. The key indicators across the entire chain are normalized, and a weight is assigned to each indicator. The normalized indicators are then summed according to their weights to obtain the comprehensive load operation status index. These key indicators across the entire chain are used as the basic data support for operation monitoring, elastic scheduling, adaptive adjustment, and fault tolerance.

4. The LED scene adaptive intelligent control system based on IoT multi-sensor fusion as described in claim 1, characterized in that, The automatic adjustment of allocation quotas for various resource pools based on the comprehensive load operation status index includes the following steps: Calculate the difference between the comprehensive load operating status index and the set threshold in real time; For each type of resource, a global adjustment factor is automatically output through a preset configurable mathematical function. Based on the global adjustment factor of each type of resource, the quota allocated to each type of resource from the total system resource pool is automatically and dynamically adjusted so that when the load pressure increases, the corresponding resource pool quota is increased to meet the processing needs of high-concurrency tasks and key businesses. When the load pressure decreases, the resource quota is automatically reduced. The resources include, but are not limited to, bandwidth, time slots and computing power.

5. The LED scene adaptive intelligent control system based on IoT multi-sensor fusion as described in claim 1, characterized in that, A global adjustment factor is automatically output through a preset configurable mathematical function. Based on this global adjustment factor, various resources are dynamically allocated and adjusted. The specific process is as follows: The difference between the comprehensive load operating status index and the set threshold is used as input to a preset configurable mathematical function. The function automatically calculates the global adjustment factor for each type of resource based on its different resource types and sensitivity parameters. The baseline quota is adjusted based on the latest global adjustment factor to obtain the allocation value of various resources within the actual scheduling cycle.

6. The LED scene adaptive intelligent control system based on IoT multi-sensor fusion as described in claim 1, characterized in that, When the task scheduling based on the adjusted resources in descending order of priority is concurrent scheduling, it also includes inputting the difference between the comprehensive load operation status index and the set threshold into the resource tilt factor function to obtain the resource tilt factor. The priority of concurrently scheduled tasks can be adjusted using a resource tilt factor to ensure that critical tasks in the scheduling process are given priority.

7. The LED scene adaptive intelligent control system based on IoT multi-sensor fusion as described in claim 6, characterized in that, For queued tasks that have not been scheduled, reserved polling resources are allocated. Through a periodic queue scanning mechanism, periodic task monitoring and important status change detection are achieved for all queued tasks. This ensures that once a high-priority, emergency, or sudden task is triggered, its priority score is updated and increased, thus preventing critical tasks from being missed or delayed in the queue.

8. The LED scene adaptive intelligent control system based on IoT multi-sensor fusion as described in claim 1, characterized in that, Also includes: Continuously monitor the communication status between each spatial partition and the central control platform. When it is detected that effective communication with the central platform has been lost, it will immediately and automatically switch to local autonomous mode, disconnect all external control services and data channels, and achieve logical isolation from the central platform. After entering autonomous mode, each local partition or regional controller runs independently, merging all tasks in the queue before the disconnection into a local autonomous scheduling pool, and removing expired, duplicate, or invalid tasks, retaining only the set of tasks that still need to be scheduled. For all tasks to be scheduled in the current scheduling pool, collect their original priority scores and the initial values ​​of adjustment factors for various resources before disconnection. The original priority scores and historical adjustment factors of all tasks to be scheduled are normalized. Using the normalized priority score, adjustment factor, and local remaining resources as initial parameters, the allocation baseline during the autonomous period is used. Based on the normalization results and the comprehensive load status collected locally in real time, the task scheduling is dynamically and adaptively adjusted. During local autonomous operation, local task, load parameters, and resource status are periodically re-collected. After historical tasks are completed, adjustment factors, task sorting, and scheduling are dynamically adjusted based on real-time load and task changes.

9. The LED scene adaptive intelligent control system based on IoT multi-sensor fusion as described in claim 8, characterized in that, Based on real-time load and task changes, the adjustment factor, task sorting, and scheduling are dynamically adjusted. The specific process is as follows: Collect key indicators of the partition at set time intervals and analyze them to obtain the partition load operation status index; The partition priority score is obtained by analyzing the partitioned tasks; Collect information on the availability of resources in the data zone; Based on the difference between the partition's comprehensive load operation status index and the threshold, the adjustment factors for various types of resources are calculated using preset configurable mathematical functions, and the actual quota of the partition resource pool is dynamically adjusted. Based on the priority score of each task, the resource allocation weight is determined from the sum of the priorities of all tasks; accordingly, the dynamically adjusted resource quota is allocated, and the task queue is sorted and scheduled.

10. The LED scene adaptive intelligent control system based on IoT multi-sensor fusion as described in claim 9, characterized in that, In the local autonomous mode, mathematical functions, adjustment factor calculations, resource weight allocation, and queue scheduling can be configured. The processing methods of the global scheduling phase are used, but the scope of action is limited to tasks and resource pools within the current autonomous partition.

Images