Intelligent park management system based on cloud computing and Internet of Things

Abstract

The invention discloses an intelligent park management system based on cloud computing and Internet of Things, and relates to the technical field of park management systems, and the system comprises a dynamic digital twinborn body module which comprises a heterogeneous data fusion layer which integrates a BIM model, LiDAR point cloud, an unmanned aerial vehicle inspection image and distributed loT sensor data, and eliminating timestamp deviation of multi-source data by adopting a space-time alignment algorithm, and generating a millimeter-level precision three-dimensional park twin model. Through the construction of the high-precision digital twinborn model, the panorama and equipment operation state of the park can be accurately presented in real time, the timestamp deviation and spatial registration error of multi-source data are eliminated, the real-time performance and accuracy of the data are improved, a solid data basis is provided for subsequent intelligent simulation and early warning decision, and the response and monitoring capabilities of the system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of park management systems, and in particular to an intelligent park management system based on cloud computing and the Internet of Things. Background Art

[0002] With the rapid development of smart cities and intelligent manufacturing, modern industrial park management faces multiple challenges, including aging equipment, high energy consumption, low resource utilization, and security risks. Traditional management models often rely on experience and a single data source, making it difficult to achieve real-time monitoring and precise control of complex industrial park environments.

[0003] Patent CN118538050B discloses an intelligent management system for terminal equipment in an IoT park. The above patent realizes intelligent control of parking lots, ensuring the efficiency and accuracy of vehicle movement.

[0004] The above patent has the characteristics of optimizing vehicle movement decisions and strong practicality, but there are still multi-source data timestamp deviations and spatial registration errors.

[0005] To this end, this application proposes an intelligent park management system based on cloud computing and the Internet of Things that can accurately present the park's panoramic view and equipment operating status in real time. Summary of the Invention

[0006] The purpose of the present invention is to provide an intelligent park management system based on cloud computing and the Internet of Things to solve the technical problems raised in the above background technology that rely on experience and a single data source, making it difficult to achieve real-time monitoring and precise control of complex park environments.

[0007] To achieve the above objectives, the present invention provides the following technical solutions: an intelligent park management system based on cloud computing and the Internet of Things, including a dynamic digital twin module, wherein the dynamic digital twin module includes: Heterogeneous data fusion layer: Integrates BIM models, LiDAR point clouds, drone inspection images, and distributed IoT sensor data. It uses a spatiotemporal alignment algorithm to eliminate timestamp deviations in multi-source data and generate a 3D campus twin model with millimeter-level accuracy. Predictive simulation engine: Based on a deep reinforcement learning framework, random perturbation parameters are injected into the twin model to generate multimodal emergency response plans. Evacuation routes are optimized through Monte Carlo tree search, and the simulation results form a two-way closed-loop feedback loop with the actual physical system. Low-latency synchronization mechanism: A lightweight digital twin agent at the edge layer is used to run local simulations on 5G MEC nodes, uploading only key decision parameters to the cloud to ensure that the delay between the twin and reality is ≤200ms.

[0008] Preferably, a quantum-enhanced self-healing network is deployed between the IoT sensor and the actuator: Embed a quantum entangled particle pair generator in key equipment. When a device fails, the device status information is instantly transmitted to the backup device through the quantum stealth state. The backup device reconstructs the operating parameters of the faulty device based on the received quantum information and generates an unalterable device switching record on the blockchain. Combined with the federated learning model of the equipment's vibration spectrum, it predicts the remaining lifespan and triggers the maintenance robot to accurately locate the fault point.

[0009] Preferably, the management system includes a security module, which includes: Biometric spatiotemporal correlation engine: Millimeter-wave radar arrays and thermal imaging cameras deployed in corridors collect real-time data on gait cycles, body temperature distribution, and facial expressions, building a spatiotemporal graph convolutional network model to identify abnormal behavior. Dynamic containment strategy: When an anomaly is detected, a drone swarm encirclement path is generated based on a reinforcement learning algorithm. The drones are equipped with directional acoustic transmitters to form a virtual barrier, while simultaneously sending the suspect's biometric augmented reality tag to the security personnel's AR glasses. Privacy protection mechanism: Biometric data is processed using homomorphic encryption technology, feature matching is completed only at the edge node, and the original data is retained on the local device.

[0010] Preferably, the management system includes a service robot module, and the service robot module includes: Multimodal large language model: Integrates visual Transformer and speech recognition models to parse fuzzy instructions and generate intent vectors by searching the spatial semantic database in the digital twin; Dynamic environmental adaptation system: The robot is equipped with a solid-state lidar and event camera to build an instantaneous obstacle probability map. When a temporary obstacle is detected, it uses an inverse reinforcement learning algorithm to replan the path; Knowledge distillation update mechanism: The robot uploads operation logs to the cloud every day, extracts effective experience through comparative learning, generates a lightweight model update package, and reversely deploys it to the edge.

[0011] Preferably, the management system includes a communication network, and the communication network includes: Programmable metasurface antenna arrays: Deployed on the facades of campus buildings, they dynamically form beamforming by applying voltage to alter the surface's electromagnetic properties, addressing multipath interference from traditional base stations. Intelligent channel switching algorithm: Based on the Deep Q network, it evaluates the channel quality of LoRaWAN, NB-IoT, and WiFi 6E in real time and allocates the optimal frequency band for devices of different priorities; Electromagnetic radiation optimization module: Dynamically adjusts transmission power based on personnel location data, reducing radiation intensity to 30% of the international standard value in densely populated office areas.

[0012] Preferably, the management system includes an energy module, and the energy module includes: Bio-photovoltaic-thermoelectric composite skin: The building's exterior walls are coated with genetically edited cyanobacteria, which generates bioelectricity when exposed to sunlight. The temperature difference between the inside and outside of the wall drives thermoelectric modules to generate electricity. Metabolic state sensing network: A microfluidic chip monitors the photosynthetic efficiency of cyanobacterial biofilms. When a drop in efficiency is detected, nanorobots are automatically activated to clean contaminants from the membrane surface. Energy Routing Protocol: Builds an energy internet based on the improved RPL protocol, prioritizes high-value load functions, and records energy transaction flows on the blockchain.

[0013] Preferably, the management system includes an equipment maintenance module, which includes: Topological insulator sensor: A topological insulator film is deposited on the surface of a motor bearing. When the device undergoes submillimeter deformation, the edge-state electrons in the film undergo directed migration, generating a detectable quantized Hall voltage. Fault knowledge graph: This method associates historical equipment maintenance records, physical simulation data, and topological insulator signals to build a graph neural network model that outputs fault probability and remaining available time. Self-organizing maintenance cluster: Multiple maintenance robots allocate tasks through swarm intelligence algorithms, use magnetic adsorption wheels to achieve vertical wall movement, and calibrate equipment installation accuracy using laser interferometers after maintenance is completed.

[0014] Preferably, the management system further includes an environment control module, which includes: Pulse neural network thermostat: This mimics the hypothalamic temperature regulation mechanism, receives artificial infrared thermal imaging data and external meteorological data, and generates pulse sequences to control the frequency conversion parameters of the air conditioning compressor; Olfactory digital twin: Using gas sensor arrays and mass spectrometer data, a volatile organic compound (VOC) diffusion model is recreated in a virtual space, linking with the fresh air system to implement source-level air purification. Optogenetic lighting system: uses genetically modified luminous microbial lamps to regulate luminous intensity through blue light irradiation and simultaneously release negative oxygen ions to improve indoor air quality.

[0015] Preferably, the management system further includes a resource management module, which includes: Four-dimensional NFT minting protocol: Encodes the use rights of conference room and parking space resources into NFTs containing spatial coordinates (x, y, z) and time windows (t), supporting second-level ownership confirmation on the consortium chain; Option-type reservation contracts: allow users to reserve priority use rights for a certain period of time in the future at a lower cost. If the resources are not used, a penalty will be automatically charged through the smart contract. Fragmented trading market: Idle resources are split into 15-minute granularity for auction, and the bidding data is written into the blockchain after verification by zero-knowledge proof.

[0016] Preferably, the management system further includes a decision-making module, which includes: Causal Rebate Model: Using the gradient boosting causal forest algorithm, we extract the potential causal relationship between equipment failure rate and energy consumption from campus operation data. Counterfactual decision optimization: For a specific problem, simulate different intervention measures in the digital twin and evaluate the cost-effectiveness of each option; Group decision-making circuit breaker mechanism: When the DAO voting results seriously conflict with the causal model recommendations, the expert committee's manual review process is triggered to ensure that the system evolution complies with physical constraints.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention uses a high-precision digital twin model to accurately present a panoramic view of the park and the operating status of equipment in real time, eliminating timestamp deviations and spatial registration errors in multi-source data, improving data real-time performance and accuracy. This provides a solid data foundation for subsequent intelligent simulation and early warning decision-making, and enhances system response and monitoring capabilities. 2. This invention uses predictive simulation and an intelligent decision-making engine to perform predictive simulation and optimize decision-making for equipment failures, energy consumption changes, and resource scheduling. This provides early warning of equipment failures and abnormal phenomena, optimizes operational strategies, reduces downtime and energy waste, significantly improves park safety and operational efficiency, and reduces maintenance costs and energy consumption. 3. This invention uses intelligent resource management and blockchain-based rights confirmation and transactions to achieve instant confirmation, flexible reservations, and efficient transactions for resources such as conference rooms and parking spaces in the park. This solves problems in resource allocation, such as delayed ownership confirmation, difficulty in making reservations, and low transaction trust. It improves resource utilization and scheduling efficiency, and enhances the transparency and fairness of park services. 4. The present invention is designed with intelligent equipment maintenance and self-organizing decision-making mechanisms to monitor equipment deformation in real time, predict failure risks and automatically organize maintenance operations, provide early warning of equipment hidden dangers, reduce maintenance response time, improve repair accuracy and operation safety, extend equipment service life, reduce maintenance costs, and ensure the continuous and stable operation of the park. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a schematic diagram of the management system structure framework of the present invention; Figure 2It is a schematic diagram of the management process of the present invention. DETAILED DESCRIPTION

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0020] In the description of the present invention, it should be noted that the terms "upper," "lower," "inner," "outer," "front end," "rear end," "both ends," "one end," "the other end," and the like, indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limiting the present invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0021] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "installed," "provided with," "connected," etc., should be understood in a broad sense. For example, "connected" may refer to a fixed connection, a detachable connection, or an integral connection; it may refer to a mechanical connection or an electrical connection; it may refer to a direct connection or an indirect connection through an intermediate medium; it may refer to internal communication between two components. Those skilled in the art will be able to understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0022] See also Figure 1 and Figure 2 The present invention provides an embodiment of an intelligent park management system based on cloud computing and the Internet of Things, including a dynamic digital twin module, wherein the dynamic digital twin module includes: Heterogeneous data fusion layer: Integrates BIM models, LiDAR point clouds, drone inspection images, and distributed IoT sensor data. It uses a spatiotemporal alignment algorithm to eliminate timestamp deviations in multi-source data and generate a 3D campus twin model with millimeter-level accuracy. Predictive simulation engine: Based on a deep reinforcement learning framework, random perturbation parameters are injected into the twin model to generate multimodal emergency response plans. Evacuation routes are optimized through Monte Carlo tree search, and the simulation results form a two-way closed-loop feedback loop with the actual physical system. Low-latency synchronization mechanism: Using a lightweight digital twin agent at the edge layer, local simulation is run on the 5G MEC node, and only key decision parameters are uploaded to the cloud, ensuring that the delay between the twin and reality is ≤ 200ms; A quantum-enhanced self-healing network is deployed between the IoT sensors and actuators: Embed a quantum entangled particle pair generator in key equipment. When a device fails, the device status information is instantly transmitted to the backup device through the quantum stealth state. The backup device reconstructs the operating parameters of the faulty device based on the received quantum information and generates an unalterable device switching record on the blockchain. Combined with a federated learning model of the equipment's vibration spectrum, it predicts the remaining lifespan and triggers a maintenance robot to precisely locate the fault point. Furthermore, the heterogeneous data fusion layer includes the following hardware deployment: BIM model: IFC-formatted building information model is input using Revit software; LiDAR point cloud: Velodyne VLP-16 laser radar is deployed, with a point cloud accuracy of ≤2cm and a sampling frequency of 10Hz; drone imaging: DJI Matric300RTK equipped with a 500,000-pixel camera, inspection height of 50m, and an overlap rate of 80%; IoT sensors: temperature and humidity (DHT22, accuracy of ±0.5°C / ±2%), vibration (ADXL335, resolution of 3mg), gas (MQ-135), etc. Spatiotemporal alignment algorithm: Time alignment: The extended Kalman filter is used to correct the timestamps of each data stream. The error covariance P and observation noise R are added to the state equation. After fusion, the time synchronization error is ≤5ms. Spatial registration: ICP coarse registration is first performed on the LiDAR point cloud and the BIM model, and then fine registration based on SIFT features is used. The final registration error is ≤1mm. 3D twin model generation: The aligned point cloud and BIM geometry are converted to a unified coordinate system, stored in an octree structure, and rendered in real time using the Unity3D engine, with an update frequency of ≥1Hz and an error of ≤0.5cm. Predictive Simulation Engine: Deep Reinforcement Learning Framework: Algorithm: Using PPO to achieve safe convergence; State Space: Contains the character's position (x, y, z), ambient temperature, and obstacle field distribution; Action Space: The unit vector of the character's movement direction; Reward Function: ,in =1.0, =5.0; Random disturbance injection: Random parameters: fire source location (random in the interval [-10, 10] m), wind speed (0-5 m / s), obstacle addition or deletion probability 0-0.2; disturbance frequency: once per simulation iteration, for a total of 1000 iterations; Monte Carlo tree search optimization: tree depth: maximum 10 layers; number of simulations: 100 per node; selection strategy: UCB1, parameter c = 1.4; output of the optimal evacuation path and real-time comparison with the on-site GPS, error ≤ 0.5m; Closed-loop feedback: The actual evacuation path collected by the sensor (UWB positioning accuracy of 15cm) is compared with the simulated path, and the simulation model parameters are updated every 30 seconds to ensure that the error between the simulation and the actual situation is ≤1m; Quantum Enhanced Self-Healing Network: Quantum Entangled Particle Pair Generator: Structure: Internal: Two 5mm diameter NV-diamond cavities, coupled fiber length 20cm; Doping: Nitrogen-Vacancy Center Concentration 10 17 / cm 3 Temperature control: coupled with a micro-cooler to maintain the cavity temperature at 4K ± 0.1K; shielding: u-metal shell, attenuating electromagnetic noise > 60dB; Entanglement preparation: Method: Photoexcitation (532nm laser), microwave pulse (2.87GHz) drive; Fidelity: Bell state preparation fidelity ≥ 0.92; Rate: 10kHz pairs / second; Stealth transmission and reconstruction: Fault detection trigger: Sensor: Bearing micro-vibration sensor (bandwidth 0-2kHz), deformation threshold 0.2um; Trigger: Microwave writing is triggered when the instantaneous increase in the vibration power band is greater than 20dB; Information replication: Encoding: Encode 10-bit voltage, temperature, and vibration spectrum characteristics into entangled spin states; Transmission delay: ≤1ms; Decoding: FPGA analysis at the receiving end, bit error rate ≤10 -6 ; Backup equipment switching: Data link: PCle GEN3×4, bandwidth 16GB / s; Switching time: ≤5ms; Self-test: Compared with the target speed, voltage deviation is ≤0.5%; Federated learning prediction: Model: FedAvg aggregated CNN, input vibration spectrum 512×512; training round: once every 10 minutes, accuracy ≥92% after convergence; trigger: notify the maintenance robot when the predicted remaining life is less than 100 hours.

[0023] See also Figure 1 and Figure 2 The present invention provides an embodiment of a smart park management system based on cloud computing and the Internet of Things, wherein the management system includes a security module, and the security module includes: Biometric spatiotemporal correlation engine: Millimeter-wave radar arrays and thermal imaging cameras deployed in corridors collect real-time data on gait cycles, body temperature distribution, and facial expressions, building a spatiotemporal graph convolutional network model to identify abnormal behavior. Dynamic containment strategy: When an anomaly is detected, a drone swarm encirclement path is generated based on a reinforcement learning algorithm. The drones are equipped with directional acoustic transmitters to form a virtual barrier, while simultaneously sending the suspect's biometric augmented reality tag to the security personnel's AR glasses. Privacy protection mechanism: Biometric data is processed using homomorphic encryption technology, feature matching is completed only at the edge node, and the original data is retained on the local device; The management system includes a service robot module, which includes: Multimodal large language model: Integrates visual Transformer and speech recognition models to parse fuzzy instructions and generate intent vectors by searching the spatial semantic database in the digital twin; Dynamic environmental adaptation system: The robot is equipped with a solid-state lidar and event camera to build an instantaneous obstacle probability map. When a temporary obstacle is detected, it uses an inverse reinforcement learning algorithm to replan the path; Knowledge distillation update mechanism: The robot uploads its operation logs to the cloud daily, extracts effective experience through comparative learning, generates a lightweight model update package, and reversely deploys it to the edge. Furthermore, the security module has the following features: biometric spatiotemporal correlation engine: hardware configuration: millimeter wave radar array: using TI76-81GHz FMCW radar, each sensor unit antenna array size is 50×50mm, can detect 0.1mm level motion changes, the array is evenly distributed along both sides of the corridor, with a unit spacing of 0.5m, forming a monitoring zone with a coverage width of 10m; thermal imaging camera: using FLIR A Boson 320×256 infrared module with a 30Hz frame rate and 0.1°C temperature resolution was used to collect facial and body temperature distribution. Data preprocessing: The radar micro-Doppler signal was denoised using bandpass filtering, and the gait cycle feature vector was extracted. The infrared image was denoised using the Nguyen-Hoang algorithm and bilaterally filtered to segment the head and shoulders, and expression keyframes were pre-extracted using HOG+SVM. Spatiotemporal graph convolutional network: Network architecture: Input nodes include gait vectors, temperature distribution grids, and expression features, forming a total of 512 spatiotemporal graph nodes. Convolutional layer: Temporal convolution uses a graph attention mechanism combined with 1D temporal convolution, with 3 layers and 256 hidden units per layer. Training details: A dataset of 5,000 labeled abnormal behaviors was used, with a cross-entropy loss, the Adam optimizer, and 200 epochs, resulting in an anomaly recognition accuracy of 98.2%. Dynamic containment strategy: Abnormal trigger: When the ST-GCN output abnormal probability is greater than 0.9 and lasts for more than 5 seconds, the drone containment submodule is triggered; Reinforcement learning path planning: Environmental state: including drone position, target personnel position, obstacle coordinates; Algorithm: Deep Q network, state dimension 128, action space 4 directions of steering + lifting, ε-greedy strategy ε initial 0.9→0.1; Reward: R=-d drone,target -5×5d drone，drone+10 (d distance) to ensure tight encirclement and collision avoidance; Training: After 5,000 iterations of the simulation environment, the average decision time converged to ≤50ms per decision; Execution and Feedback: Directional Acoustic Transmitter: Each drone is equipped with a 1kHz-4kHz directional ultrasonic array, with a sound pressure level of 100dB at a distance of 1m, which can achieve a deterrent effect; AR Glasses Push: Through the ROS2 message bus, the suspect ID, question, gait, and facial expression features are encapsulated into JSON and sent to the security personnel wearing MagicLeap2 via the 5G network; Service Robot Module: Multimodal Large Language Model: Model Architecture: Visual Transformer: Uses ViT-Base to encode 360° fisheye camera images; Speech Recognition: Uses Consformer-XS to achieve real-time speech-to-text conversion; Fusion Strategy: Speech-text embedding is concatenated with visual features and mapped to intent vectors via a two-layer fully connected network; Spatial Semantic Database: Stores spatial entities such as building rooms and corridors in the digital twin in the form of a graph database. Node attributes include (x, y, z) coordinates, function labels, and adjacency relationships. Intent vectors are matched with node features using cosine similarity, and the top-3 positions are returned to the robot navigation unit. Dynamic environment adaptation system: Sensor configuration: Solid-state lidar: Ouster OS0-128, range 120m, 128 laser lines; Event camera: Prophesee Gen4, 1us time resolution, used to capture dynamic obstacles; Instantaneous obstacle probability map: LiDAR point cloud is projected onto a 2D grid, combined with the event camera event stream, and the probability of each obstacle is calculated using the Bayesian update method; Map update frequency 20Hz; Inverse reinforcement learning replanning: Algorithm: Based on MaxEnt IRL, first learn the reward function from expert demonstrations; Replanning: Run A* search on the current local map to initialize the path, and then use the learned reward function to fine-tune the path weights to generate a smooth, obstacle-avoiding new path.

[0024] See also Figure 1 and Figure 2 The present invention provides an embodiment of a smart park management system based on cloud computing and the Internet of Things, wherein the management system includes a communication network, and the communication network includes: Programmable metasurface antenna arrays: Deployed on the facades of campus buildings, they dynamically form beamforming by applying voltage to alter the surface's electromagnetic properties, addressing multipath interference from traditional base stations. Intelligent channel switching algorithm: Based on the Deep Q network, it evaluates the channel quality of LoRaWAN, NB-IoT, and WiFi 6E in real time and allocates the optimal frequency band for devices of different priorities; Electromagnetic radiation optimization module: Dynamically adjusts the transmission power based on personnel location data, reducing the radiation intensity in dense office areas to 30% of the international standard value; Furthermore, on the facades of the campus buildings, antenna arrays cover the main entrances and exits of the campus and high-density work areas. The antenna array is composed of multiple metamaterial units, each of which integrates microelectronic components such as varactor diodes. The electromagnetic response of the unit can be controlled by applying different voltages. The controller issues instructions based on the real-time network status and environmental parameters to change the phase and amplitude of each unit, dynamically forming a directional beam. Through beamforming technology, the interference caused by traditional multipath propagation is effectively offset, and the stability and coverage efficiency of signal transmission are improved. The system first collects channel interference information through the environmental monitoring module. When it detects that the multipath interference is more significant, it sends a control voltage to each unit in the metasurface array, so that the entire antenna array can be focused on a specific area in real time, so that the wireless signal is transmitted along the desired path, reducing the delay and attenuation caused by signal reflection. During the transmission process, based on the channel quality feedback information collected from the terminal device, the communication controller further adjusts the voltage control parameters to form a dynamic closed-loop control to achieve precise beamforming. The campus simultaneously deploys three communication protocols: LoRaWAN, NB-IoT, and WiFi 6E. These protocols differ in transmission rate, coverage, and interference resistance. To meet the communication needs of various devices, such as low-power sensors and high-bandwidth data devices, the system must be able to dynamically evaluate the quality of each channel and allocate the most appropriate frequency band. The system collects real-time channel metrics such as signal-to-noise ratio (SNR), interference level, latency, and packet loss rate. These parameters are used as state inputs for the Deep Q Network (DQN). The DQN algorithm then allocates channels to devices of different priorities based on their current state. Critical monitoring or emergency communication data is prioritized to ensure stable, low-interference communication links for important devices. The system assigns positive or negative rewards to selected strategies based on actual transmission success rate, latency optimization, and interference suppression effectiveness. This iterative optimization model ensures more accurate and effective channel switching decisions. When a WiFi 6E channel signal drops due to dense crowds or building obstruction in a particular area, the Deep QN immediately identifies the current channel state and selects the low-power, high-penetration NB-IoT channel to allocate the optimal transmission channel for emergency monitoring devices in that area, ensuring smooth communication. Utilizing a network of positioning sensors and cameras deployed within the campus, the system collects real-time information on the distribution of people in offices and public areas, calculating regional population density. Based on this collected positioning data, the system combines environmental noise and signal quality feedback to dynamically adjust the transmit power of each base station and access device using pre-set algorithms, such as adaptive adjustment. In densely populated areas, the system reduces power levels; in sparsely populated or outdoor areas, it increases transmit power appropriately to ensure coverage. During this adjustment process, the electromagnetic radiation optimization module continuously monitors actual radiation levels, using sensor feedback to form a closed-loop control loop to ensure that the target radiation intensity is consistently ≤30% of the international standard.

[0025] See also Figure 1 and Figure 2 The present invention provides an embodiment of a smart park management system based on cloud computing and the Internet of Things, wherein the management system includes an energy module, and the energy module includes: Bio-photovoltaic-thermoelectric composite skin: The building's exterior walls are coated with genetically edited cyanobacteria, which generates bioelectricity when exposed to sunlight. The temperature difference between the inside and outside of the wall drives thermoelectric modules to generate electricity. Metabolic state sensing network: A microfluidic chip monitors the photosynthetic efficiency of cyanobacterial biofilms. When a drop in efficiency is detected, nanorobots are automatically activated to clean contaminants from the membrane surface. Energy Routing Protocol: Builds an energy internet based on the improved RPL protocol, prioritizes high-value load functions, and records energy transaction flows on the blockchain; Furthermore, energy module: bio-photovoltaic-thermoelectric composite epidermis: gene-edited cyanobacteria coating structure: cyanobacteria strain: Synechocystis sp. PCC6803 was selected, and the efficient electron carrier gene was inserted into the psbA gene by CRISPR / Cas9: coating collection: polyvinyl alcohol-co-acrylic acid film, 200um thick, and the surface was plasma treated to improve hydrophilicity; coating process: using spray deposition technology, the coating uniformity was ±5um, and after drying, it was incubated in a 37℃ incubator with 95% humidity for 24 hours to form a stable biofilm; bio-photovoltaic current collection: electrode structure: a layer of ITO transparent conductive film was deposited on the back of the coating surface, and the contact area between ITO and cyanobacteria cells was 10cm 2 ;Power output: Under 1000lux white light irradiation, open circuit voltage OCV≈0.6V, short circuit current ISC≈120uA / cm 2 , boosted to 5V output by DC-DC; Thermoelectric module integration: Module structure: Each 15×15mmBi2Te3 / n-type-Bi2Te3 thermoelectric pair unit, Seebeck coefficient ≈200uV / K; Nanorobot cleaning mechanism: Robot configuration: 5mm diameter sphere, built-in micro stepper motor and segmented robotic arm; cleaning tools: microfiber brush head and α-trypsin micro-injection system, mechanical brushing + enzymatic decontamination; execution process: the chip signal sends the task to the edge controller, the nanorobot is positioned to the target area through optical marking, and after completing the brushing and enzyme spraying, it returns to the charging base in reverse order. After 30 minutes, it will be retested and executed again or the task is completed.

[0026] See also Figure 1 and Figure 2 The present invention provides an embodiment of a smart park management system based on cloud computing and the Internet of Things, wherein the management system includes a device maintenance module, and the device maintenance module includes: Topological insulator sensor: A topological insulator film is deposited on the surface of a motor bearing. When the device undergoes submillimeter deformation, the edge-state electrons in the film undergo directed migration, generating a detectable quantized Hall voltage. Fault knowledge graph: This method associates historical equipment maintenance records, physical simulation data, and topological insulator signals to build a graph neural network model that outputs fault probability and remaining available time. Self-organizing maintenance cluster: Multiple maintenance robots use swarm intelligence algorithms to allocate tasks, use magnetic wheels to achieve vertical wall movement, and use laser interferometers to calibrate equipment installation accuracy after maintenance is completed. Furthermore, a topological insulator film is deposited on the surface of the motor bearing in advance, and the sensor monitors its quantized Hall voltage in real time. During continuous monitoring, the voltage changes caused by tiny deformations are recorded and transmitted to the fault prediction system. The fault knowledge graph correlates and analyzes the real-time collected Hall voltage data, historical maintenance records and physical simulation data, and uses the graph neural network to output the current failure probability and remaining useful life of the equipment. When the failure probability reaches the warning threshold, the central dispatching system starts the maintenance process; multiple maintenance robots are automatically assigned to target maintenance tasks according to the swarm intelligence algorithm, and move to the target area along the vertical surface of the equipment through magnetic adsorption wheels. The robots complete the inspection and fixation of key components on site, and then use a laser interferometer to perform precision calibration on the repair area. The calibration data is fed back to the central system to form a closed-loop maintenance and inspection record and update the equipment status database.

[0027] See also Figure 1 and Figure 2 The present invention provides an embodiment of a smart park management system based on cloud computing and the Internet of Things, wherein the management system further includes an environmental control module, which includes: Pulse neural network thermostat: This mimics the hypothalamic temperature regulation mechanism, receives artificial infrared thermal imaging data and external meteorological data, and generates pulse sequences to control the frequency conversion parameters of the air conditioning compressor; Olfactory digital twin: Using gas sensor arrays and mass spectrometer data, a volatile organic compound (VOC) diffusion model is recreated in a virtual space, linking with the fresh air system to implement source-level air purification. Optogenetic lighting system: uses genetically modified luminescent microorganisms to regulate luminous intensity through blue light irradiation and simultaneously release negative oxygen ions to improve indoor air quality; Furthermore, the pulse neural network temperature controller: Hardware and data acquisition: Infrared thermal imaging: Equipment: FLIR Boson 320×256LWIR module, pixel pitch 12um, frame rate 30Hz, operating temperature range -40℃-80℃, installation height: 2.5m from the ground, field of view covers 4×3m office area; pulse neural network architecture: neuron model: Leaky Integrate-and-Fire, membrane time constant τm = 20ms, threshold Vm = -50mV, reset voltage Vreset = -65mV; Network topology: Input layer: 16×16 nodes, corresponding to a downsampled 16×16 heat map; Hidden layer: 2 layers with 256 LIF neurons, sparsely connected; Output layer: 4 LIF neurons, corresponding to four compressor frequency settings; Decoding method: The heat map temperature (18-30°C) is linearly mapped to a pulse frequency (0-100Hz), and the outdoor temperature is mapped to four additional Poisson pulse channels with a frequency mapping range of (0-50Hz); Control process and online learning: Cycle: The network has a 50ms processing cycle, including acquisition and encoding, SNN simulation, and pulse train output; Compressor control: Outputs four pulse counts, selects the gear with the highest count, and sends it to the variable frequency drive via the CAN bus; Feedback and adaptation: Collects indoor temperature sensor and energy consumption data, and fine-tunes synaptic weights based on STDP to ensure long-term stable and energy-efficient operation. Olfactory digital twin: Gas sensor array: Sensors: Figaro TGS2600, AIphasens H2S, NO2 / CO, etc., 4 of each, deployed indoors at a height of 3m, with a sampling period of 10s; Mass spectrometer: Equipment: Pico-Tag portable mass spectrometer, sampling period of 1min, used to calibrate sensor readings and obtain VOC composition spectra; CFD-based olfactory digital twin construction: Computational model: Utilize O-spray FOAM v9 to implement 3D incompressible Navier-Stokes equations and convection-diffusion equation coupling: , where C is the VOC concentration, u is driven by the indoor air conditioning and the fresh air speed boundary condition, D=1.2×10 -5 m 2 / s is the diffusion coefficient, S is the source strength measured at each sensing point; Grid and solution: Indoor grid: 0.1m cubic unit, the total number of units is about 1.2×10 6 Time step: 0.5s. When the simulated VOC concentration in any area exceeds 500ppb, the digital twin identifies the pollution source, adjusts the corresponding air supply volume to 150% of the original maximum value, closes the return air outlet of the polluted area, and opens the bypass purification device. Optogenetic lighting system: Genetically modified luminescent microbial lamp: Microorganism: LumiEcoli, with a strong promoter inserted into the LuxCDABE cluster via CRISPR / Cas9; Package base: A transparent hydrogel chip with dimensions of 3×3×1cm, containing a capillary channel for circulating culture medium; LED triggering: Built-in 475nm blue light LED array; Luminescence and negative oxygen ion release: Blue light regulation: PWM frequency 1Hz, brightness cycle [0.1-10Hz], indoor illuminance sensor feedback, real-time adjustment of LED duty cycle to ensure target intensity ±5lux; Negative oxygen ions: LumiEcoli triggers the NOX enzyme system to release oxygen ions, which diffuse into the air through the porous package and are monitored in real time by an ion meter, with a target concentration of 500±50ions / cm 3 , automatically reducing the blue light intensity when it exceeds the threshold.

[0028] See also Figure 1 and Figure 2 The present invention provides an embodiment of a smart park management system based on cloud computing and the Internet of Things, wherein the management system further includes a resource management module, which includes: Four-dimensional NFT minting protocol: Encodes the use rights of conference room and parking space resources into NFTs containing spatial coordinates (x, y, z) and time windows (t), supporting second-level ownership confirmation on the consortium chain; Option-type reservation contracts: allow users to reserve priority use rights for a certain period of time in the future at a lower cost. If the resources are not used, a penalty will be automatically charged through the smart contract. Fragmented trading market: Idle resources are split into 15-minute granularity for auction, and the bidding data is written into the blockchain after verification by zero-knowledge proof; The management system further includes a decision-making module, which includes: Causal Rebate Model: Using the gradient boosting causal forest algorithm, we extract the potential causal relationship between equipment failure rate and energy consumption from campus operation data. Counterfactual decision optimization: For a specific problem, simulate different intervention measures in the digital twin and evaluate the cost-effectiveness of each option; Group decision-making circuit breaker mechanism: When the DAO voting results seriously conflict with the causal model recommendations, the expert committee will be triggered to manually review the process to ensure that the system evolution complies with physical constraints; Furthermore, for resources such as conference rooms and parking spaces, their spatial coordinate information (x, y, z) is extracted through on-site measurements and digital twin data. This is then combined with a predetermined time window (t) to form a four-dimensional data description. Using a specific smart contract encoding algorithm, this four-dimensional resource information is embedded in a non-fungible token (NFT). Each NFT uniquely identifies the right to use a resource and also carries metadata such as resource status, usage rules, and expiration date. Minting NFTs on a consortium chain utilizes a lightweight consensus mechanism to ensure that resource ownership information is confirmed and broadcasted within seconds, avoiding resource disputes caused by multi-party competition and achieving real-time ownership confirmation within seconds. For resource usage in a certain period of time in the future, the system allows users to reserve usage rights at a lower cost, forming a reservation contract with the nature of an option. The contract stipulates that users pay a certain deposit when making a reservation and enjoy priority use rights in the future. The contract has built-in smart contract rules. When the scheduled period arrives, the system automatically confirms the user's reservation. If the resource is not used in the end, the contract terms automatically trigger the liquidated damages deduction logic, converting the deposit or prepayment into liquidated damages to compensate the resource owner. This solution not only reduces the user's reservation cost, but also introduces an economic incentive mechanism to ensure the rational flow of resources and provide a market-based solution for idle resources. To improve resource utilization, the system breaks down idle resource periods, such as unused conference rooms or parking spaces, into 15-minute units, allowing users to bid on these periods in this fragmented market. All bidding data is verified using zero-knowledge proof technology to ensure participant privacy and data authenticity, and is then recorded on the blockchain to generate transaction records. The blockchain ledger ensures that all transactions are open, transparent, and tamper-proof, creating a fair competitive environment for resource transfer and leasing across the entire park while also improving resource utilization. The system comprehensively collects equipment operation data, maintenance records, environmental monitoring data, and energy consumption indicators within the park to construct a multidimensional data set between equipment failure rate and energy consumption. It uses a causal forest model based on gradient boosting, integrating decision trees with causal reasoning ideas, to extract the potential causal relationship between equipment failure rate and energy consumption from a large amount of heterogeneous data, and calculate the marginal effect of each intervention measure. The model outputs quantifiable causal impact values, providing data support for equipment updates, preventive maintenance, and energy consumption optimization, and based on this, returns part of the cost savings to form an incentive feedback mechanism.

[0029] How it works: The management system collects all kinds of data from the park in real time through a variety of sensors deployed by the Internet of Things. Through spatiotemporal alignment and multimodal fusion of these data sources, a high-precision, real-time updated digital twin is constructed. This digital twin maps the park's physical space with millimeter-level accuracy, forming a complete virtual environment. Data is centrally stored and deeply processed remotely, while local real-time simulation is achieved with the help of edge computing nodes. Using algorithms such as deep reinforcement learning and Monte Carlo tree search, the system performs predictive simulations of equipment status, human behavior, and environmental changes, and outputs emergency response plans and control decisions. With comprehensive data support, the system integrates modules such as fault knowledge graph, causal rebate model, and counterfactual decision optimization to make intelligent decisions on operational issues such as equipment maintenance, energy consumption control, and resource reservation. Finally, through technologies such as blockchain and smart contracts, real-time confirmation, transaction, and full-process supervision of resource use rights are achieved, thereby promoting the intelligent upgrade of the overall operating efficiency and service experience of the park.

[0030] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. An intelligent park management system based on cloud computing and the Internet of Things, including a dynamic digital twin module, characterized by: The dynamic digital twin module includes: Heterogeneous data fusion layer: Integrates BIM models, LiDAR point clouds, drone inspection images, and distributed IoT sensor data. It uses a spatiotemporal alignment algorithm to eliminate timestamp deviations in multi-source data and generate a 3D campus twin model with millimeter-level accuracy. Predictive simulation engine: Based on a deep reinforcement learning framework, random perturbation parameters are injected into the twin model to generate multimodal emergency response plans. Evacuation routes are optimized through Monte Carlo tree search, and the simulation results form a two-way closed-loop feedback loop with the actual physical system. Low-latency synchronization mechanism: A lightweight digital twin agent at the edge layer is used to run local simulations on 5G MEC nodes, uploading only key decision parameters to the cloud to ensure that the delay between the twin and reality is ≤200ms.

2. The intelligent park management system based on cloud computing and the Internet of Things according to claim 1, characterized in that: Deploy a quantum-enhanced self-healing network between the IoT sensors and actuators: Embed a quantum entangled particle pair generator in key equipment. When a device fails, the device status information is instantly transmitted to the backup device through the quantum stealth state. The backup device reconstructs the operating parameters of the faulty device based on the received quantum information and generates an unalterable device switching record on the blockchain. Combined with the federated learning model of the equipment's vibration spectrum, it predicts the remaining lifespan and triggers the maintenance robot to accurately locate the fault point.

3. The intelligent park management system based on cloud computing and the Internet of Things according to claim 1, characterized in that: The management system includes a security module, which includes: Biometric spatiotemporal correlation engine: Millimeter-wave radar arrays and thermal imaging cameras deployed in corridors collect real-time data on gait cycles, body temperature distribution, and facial expressions, building a spatiotemporal graph convolutional network model to identify abnormal behavior. Dynamic containment strategy: When an anomaly is detected, a drone swarm encirclement path is generated based on a reinforcement learning algorithm. The drones are equipped with directional acoustic transmitters to form a virtual barrier, while simultaneously sending the suspect's biometric augmented reality tag to the security personnel's AR glasses. Privacy protection mechanism: Biometric data is processed using homomorphic encryption technology, feature matching is completed only at the edge node, and the original data is retained on the local device.

4. The intelligent park management system based on cloud computing and the Internet of Things according to claim 1, characterized in that: The management system includes a service robot module, which includes: Multimodal large language model: Integrates visual Transformer and speech recognition models to parse fuzzy instructions and generate intent vectors by searching the spatial semantic database in the digital twin; Dynamic environmental adaptation system: The robot is equipped with a solid-state lidar and event camera to build an instantaneous obstacle probability map. When a temporary obstacle is detected, it uses an inverse reinforcement learning algorithm to replan the path; Knowledge distillation update mechanism: The robot uploads operation logs to the cloud every day, extracts effective experience through comparative learning, generates a lightweight model update package, and reversely deploys it to the edge.

5. The intelligent park management system based on cloud computing and the Internet of Things according to claim 1, characterized in that: The management system includes a communication network, which includes: Programmable metasurface antenna arrays: Deployed on the facades of campus buildings, they dynamically form beamforming by applying voltage to alter the surface's electromagnetic properties, addressing multipath interference from traditional base stations. Intelligent channel switching algorithm: Based on the Deep Q network, it evaluates the channel quality of LoRaWAN, NB-IoT, and WiFi 6E in real time and allocates the optimal frequency band for devices of different priorities; Electromagnetic radiation optimization module: Dynamically adjusts transmission power based on personnel location data, reducing radiation intensity to 30% of the international standard value in densely populated office areas.

6. The intelligent park management system based on cloud computing and the Internet of Things according to claim 1, characterized in that: The management system includes an energy module, which includes: Bio-photovoltaic-thermoelectric composite skin: The building's exterior walls are coated with genetically edited cyanobacteria, which generates bioelectric current when exposed to light. The temperature difference between the inside and outside of the wall drives thermoelectric modules to generate electricity. Metabolic state sensing network: A microfluidic chip monitors the photosynthetic efficiency of cyanobacteria biofilms. When a drop in efficiency is detected, nanorobots are automatically activated to clean pollutants from the membrane surface. Energy Routing Protocol: Builds an energy internet based on the improved RPL protocol, prioritizes high-value load functions, and records energy transaction flows on the blockchain.

7. The intelligent park management system based on cloud computing and the Internet of Things according to claim 1, characterized in that: The management system includes an equipment maintenance module, which includes: Topological insulator sensor: A topological insulator film is deposited on the surface of a motor bearing. When the device undergoes submillimeter deformation, the edge-state electrons in the film undergo directed migration, generating a detectable quantized Hall voltage. Fault knowledge graph: This method associates historical equipment maintenance records, physical simulation data, and topological insulator signals to build a graph neural network model that outputs fault probability and remaining available time. Self-organizing maintenance cluster: Multiple maintenance robots allocate tasks through swarm intelligence algorithms, use magnetic adsorption wheels to achieve vertical wall movement, and calibrate equipment installation accuracy using laser interferometers after maintenance is completed.

8. The intelligent park management system based on cloud computing and the Internet of Things according to claim 1, characterized in that: The management system further includes an environment control module, which includes: Pulse neural network thermostat: This mimics the hypothalamic temperature regulation mechanism, receives artificial infrared thermal imaging data and external meteorological data, and generates pulse sequences to control the frequency conversion parameters of the air conditioning compressor; Olfactory digital twin: Using gas sensor arrays and mass spectrometer data, a volatile organic compound (VOC) diffusion model is reconstructed in a virtual space, linking with the fresh air system to implement source-level air purification. Optogenetic lighting system: uses genetically modified luminescent microbial lamps to regulate luminous intensity through blue light irradiation and simultaneously release negative oxygen ions to improve indoor air quality.

9. The intelligent park management system based on cloud computing and the Internet of Things according to claim 1, characterized in that: The management system further includes a resource management module, which includes: Four-dimensional NFT minting protocol: Encodes the use rights of conference room and parking space resources into NFTs containing spatial coordinates (x, y, z) and time windows (t), supporting second-level ownership confirmation on the consortium chain; Option-type reservation contracts: allow users to reserve priority use rights for a certain period of time in the future at a lower cost. If the resources are not used, a penalty will be automatically charged through the smart contract. Fragmented trading market: Idle resources are split into 15-minute granularity for auction, and the bidding data is written into the blockchain after verification by zero-knowledge proof.

10. The intelligent park management system based on cloud computing and the Internet of Things according to claim 1, characterized in that: The management system further includes a decision-making module, which includes: Causal Rebate Model: Using the gradient boosting causal forest algorithm, we extract the potential causal relationship between equipment failure rate and energy consumption from campus operation data. Counterfactual decision optimization: For a specific problem, simulate different intervention measures in the digital twin and evaluate the cost-effectiveness of each option; Group decision-making circuit breaker mechanism: When the DAO voting results seriously conflict with the causal model recommendations, the expert committee's manual review process is triggered to ensure that the system evolution complies with physical constraints.

Images