Intelligent property multifunctional data fusion and prediction optimization control method

Abstract

The invention discloses an intelligent property multifunctional data fusion and prediction optimization control method, and belongs to the technical field of property internet of things encryption, and the method comprises the following steps: S1, collecting property multi-source data through a preset sensor network, and generating a fusion data set; s2, constructing a rule base based on the fusion data set, generating a standard control instruction, and executing a preset key management strategy on the optimization instruction; s4, issuing the standard control instruction to the terminal equipment; the optimization instruction encrypted through the key management strategy is decrypted in real time and transmitted to preset terminal equipment; and S5, collecting execution results of the standard control instruction and the optimization instruction and feedback data of the terminal equipment, and updating a compensation rule and a key management strategy. According to the method, the optimization instruction is subjected to hierarchical encryption processing through the key management strategy, and the core contradictions that the execution of a high-timeliness instruction is delayed due to excessive encryption and the risk of leakage of a high-sensitivity instruction due to insufficient protection under a fixed encryption strategy are solved.

Description

Technical Field

[0001] This invention relates to the field of encryption technology for Internet of Things (IoT) in property management, and in particular to a method for multifunctional data fusion and predictive optimization control of smart properties. Background Technology

[0002] Property management is a comprehensive management system that maintains, services, and operates buildings, facilities, and their supporting environment. Its core objectives are to ensure safety, improve efficiency, and optimize user experience. With the development of IoT, cloud computing, and AI technologies, traditional property management is gradually upgrading to smart property management—that is, using modern information technology to achieve a data-driven intelligent operation and maintenance model, covering multiple dimensions such as security monitoring, equipment maintenance, energy management, environmental regulation, and pedestrian flow scheduling. The core characteristics of smart property management include multi-system collaborative management, real-time data interaction, and automated decision support. For example, sensor networks collect environmental parameters and equipment status in real time, providing a data foundation for dynamic control.

[0003] The working principle of existing smart property management technologies mainly includes data collection, transmission, processing and analysis, and decision control. First, various sensors and devices deployed in the property environment collect multi-dimensional data in real time, such as temperature, humidity, air quality, pedestrian flow, and equipment operating status. This data is transmitted to a central control system or cloud platform via wireless or wired networks. In the processing and analysis stage, the system uses preset algorithms to calculate the data and generate corresponding control commands or operational suggestions. Finally, the control commands are executed through IoT devices or management terminals to achieve automated control of the property environment and equipment. Additionally, some systems provide a user interface, allowing managers and users to view real-time data and historical records, and to manually intervene as needed.

[0004] Existing technologies generally adopt fixed encryption strategies, which cannot dynamically adjust the encryption strength according to the risk level of the instruction content. For example, time-sensitive instructions (such as emergency equipment stop) require lightweight encryption to ensure real-time performance, while highly sensitive instructions (such as owner privacy data) require enhanced encryption protection. However, existing systems lack such dynamic classification mechanisms, which leads to delays for time-sensitive instructions due to excessive encryption, or exposure of risks for highly sensitive instructions due to insufficient protection.

[0005] Therefore, there is an urgent need to provide a multi-functional data fusion and predictive optimization control method for smart property management to solve the above problems. Summary of the Invention

[0006] The technical problem to be solved by this invention is to overcome the fact that the existing technologies generally adopt fixed encryption strategies, which cannot dynamically adjust the encryption strength according to the risk level of the instruction content. For example, high time-sensitive instructions (such as equipment emergency stop) require lightweight encryption to ensure real-time performance, while high-sensitivity instructions (such as owner privacy data) require enhanced encryption protection. However, the existing systems lack such dynamic hierarchical mechanisms, which leads to the problem that high time-sensitive instructions are delayed due to excessive encryption, or high-sensitivity instructions are exposed to risks due to insufficient protection. This invention provides a smart property multi-functional data fusion and predictive optimization control method.

[0007] To solve the above-mentioned technical problems, one technical solution adopted by the present invention is: to provide a smart property multi-functional data fusion and predictive optimization control method, comprising the following steps:

[0008] S1. Collect multi-source data of the property through a preset sensor network and generate a fused dataset;

[0009] S2. Construct a rule base based on the fused dataset: Generate basic rules based on the fused dataset, configure preset compensation rules for the basic rules, and calculate the trigger threshold of the compensation rules;

[0010] S3. Analyze the fused dataset based on the preset prediction model to generate a predicted change trend. If the predicted change trend conforms to the basic rules in the rule base, a standard control instruction is generated. If the predicted change trend deviates from the basic rules in the rule base and exceeds the trigger threshold of the compensation rule, an optimization instruction is generated according to the compensation rule, and a preset key management strategy is executed on the optimization instruction.

[0011] S4. Send the standard control command to the terminal device; decrypt the optimized command encrypted by the key management strategy in real time and transmit it to the preset terminal device;

[0012] S5. Collect the execution results of the standard control instructions and the optimization instructions, as well as the feedback data from the terminal device, and update the compensation rules and the key management strategy.

[0013] The present invention is further configured such that the multi-source property data in step S1 includes security monitoring data, equipment operation status data, environmental sensor data, energy pipeline network data, and personnel activity data.

[0014] The present invention is further configured such that the step of generating the fused dataset in step S1 is as follows:

[0015] S11. Based on the sensor network, security monitoring data, equipment operation status data, environmental sensing data, energy pipeline data and personnel activity data are acquired in real time, aligned and mapped to a preset storage space through timestamps, and the security monitoring data, equipment operation status data, environmental sensing data, energy pipeline data and personnel activity data are stored in each storage unit of the storage space, and the storage units are stacked to construct the original data cube.

[0016] S12. Preprocess the original data cube and extract data from each storage unit: For security monitoring data, extract the frequency of personnel activity and the speed of object movement to generate first feature data; for equipment operating status data, extract the operating time of the equipment and the rate of change of equipment power to generate second feature data; for environmental sensing data, extract the temperature and humidity change gradient and the PM2.5 concentration fluctuation value to generate third feature data; for energy pipeline network data, extract the frequency peak value and abnormal fluctuation period of water current pulse to generate fourth feature data; for personnel activity data, extract the dwell hotspot area of ​​Bluetooth beacon trajectory and the time-varying features of gate passage density to generate fifth feature data; fuse the first feature data, the second feature data, the third feature data, the fourth feature data and the fifth feature data to generate a fused dataset.

[0017] The present invention is further configured such that the construction of the rule base in step S2 includes:

[0018] S21. Combine the frequency of personnel activity and the speed of object movement of the first feature data into a security behavior rule; associate the device running time and the device power change rate of the second feature data into a device health rule; map the temperature and humidity change gradient and PM2.5 concentration fluctuation value of the third feature data into an environmental anomaly rule; convert the frequency peak value and abnormal fluctuation period of the water current pulse of the fourth feature data into an energy failure rule; integrate the dwell hotspot area of ​​the Bluetooth beacon trajectory and the time-varying feature of the gate passage density of the fifth feature data into a human flow density rule; and use the security behavior rule, the device health rule, the environmental anomaly rule, the energy failure rule, and the human flow density rule as the basic rules.

[0019] S22. Configure preset compensation rules for each type of basic rule: Configure a first dynamic sub-rule, a second dynamic sub-rule, a third dynamic sub-rule, a fourth dynamic sub-rule and a fifth dynamic sub-rule for the security behavior rule, the equipment health rule, the environmental anomaly rule, the energy failure rule and the crowd density rule respectively.

[0020] S23. Calculate the trigger threshold for each compensation rule according to the various basic rules, and complete the construction of the rule base by using the various basic rules, the compensation rules configured by the various basic rules, and the trigger thresholds of each compensation rule.

[0021] The present invention is further configured such that the method for generating the prediction model in step S3 is as follows:

[0022] S31. Obtain the historical fusion dataset, extract the initial, intermediate, and late data of the first feature data, second feature data, third feature data, fourth feature data, and fifth feature data in the same first sampling period from the historical fusion dataset, and extract the initial, intermediate, and late data collected in the second sampling period adjacent to the same sampling period as the training set, validation set, and test set, respectively.

[0023] S32. Map the initial data, intermediate data, and late data in the previous sampling period to a preset neural network model, and divide the initial data, intermediate data, and late data into grid regions in the neural network model respectively. Through the neural network model, adaptively expand the boundaries of the initial data, intermediate data, and late data to obtain expanded training subsets, expanded verification subsets, and expanded test subsets. Map the expanded training subsets, expanded verification subsets, and expanded test subsets to each of the grid regions to generate a first expanded region, a second expanded region, and a third expanded region connected to each of the grid regions.

[0024] S33. Map the training set, validation set, and test set within the second sampling period to each of the grid regions in the neural network model, generate a first virtual extended region, a second virtual region, and a third virtual extended region connected to each of the grid regions, calculate whether the difference between the first extended region, the second extended region, and the first virtual extended region, the second virtual region, and the third virtual extended region meets a preset condition. If it does, use the neural network model as the prediction model; if it does not meet the preset condition, adjust the parameters of the neural network model according to the difference.

[0025] The present invention is further configured such that the generation step of the optimization instruction in step S3 is as follows:

[0026] S301. Identify the current scenario category of the property based on the predicted change trend, and call the compensation strategy defined in the compensation rules;

[0027] S302. Extract the first feature data, second feature data, third feature data, fourth feature data, and fifth feature data associated with the compensation strategy from the rule base, and combine them with the compensation strategy to generate first instruction content, second instruction content, third instruction content, fourth instruction content, and fifth instruction content through a preset control logic template. The first instruction content, second instruction content, third instruction content, fourth instruction content, and fifth instruction content are all used as optimization instructions.

[0028] The present invention is further configured such that: the key management strategy in step S3 specifically includes: adopting a preset lightweight key encryption mechanism for the first instruction content, that is, setting a verification stage in the transmission channel between the first instruction content and the terminal device, the verification stage including key registration and key reconstruction, the key registration referring to binding the first instruction content with a preset initial key before transmission to generate instruction data with an initial key identifier; the key reconstruction referring to dynamically updating and transforming the initial key during transmission to form a new key, binding the new key with the instruction data and transmitting them together to a preset decryption module;

[0029] The second instruction content is encrypted using a preset delayed key encryption mechanism to generate an encryption layer that encapsulates the second instruction content;

[0030] The encryption layer includes a main encryption layer and several sub-encryption layers. The several sub-encryption layers are hidden when the main encryption layer successfully decrypts or fails to decrypt. The several sub-encryption layers are revealed layer by layer when the main encryption layer fails to decrypt or the decryption time exceeds a preset threshold.

[0031] The steps for generating the encryption layer are as follows: extracting the string from the second instruction content as the master node, and establishing a master connection index based on the association between the master nodes; weaving the master nodes into a first spherical mesh structure according to the master connection index to form a main encryption layer that wraps the second instruction content; randomly shuffling the string nodes of the second instruction content to generate child nodes, and establishing a sub-connection index between the child nodes; weaving the child nodes into a second spherical mesh structure according to the sub-connection index to form a sub-encryption layer that wraps around the outside of the main encryption layer;

[0032] A preset standard encryption mechanism is applied to the third, fourth, and fifth instruction contents.

[0033] The present invention is further configured such that: step S3 further includes maintaining the current operating state of the property's equipment if the predicted change trend deviates from the basic rules in the rule base and does not exceed the trigger threshold of the compensation rule;

[0034] Before sending the standard control command to the terminal device in step S4, a preset encryption operation is performed on the standard control command.

[0035] The present invention is further configured such that: the specific content of real-time decryption of the optimized instruction encrypted by the key management strategy in step S4 includes:

[0036] Execution of the first instruction content using the aforementioned lightweight key encryption mechanism:

[0037] The decryption module extracts feature values ​​from the new key and matches these feature values ​​with a preset key mapping rule to generate a seed value for the initial key. Based on the seed value, an initial key is generated using a preset key expansion algorithm. The instruction data is then decrypted using the initial key to obtain the content of the first instruction.

[0038] For the execution of the second instruction content using a delayed key encryption mechanism:

[0039] The decryption module extracts strings from the second instruction content, analyzes the semantic relationships and structural connections between strings, and establishes a multi-level index between strings based on the semantic relationships and structural connections between base strings. This multi-level index includes a main index and sub-indexes. The main index identifies the overall structure of the string, and the sub-indexes identify the internal logical relationships of the string. A main checksum is generated based on the main index, and a sub-checksum is generated based on the sub-indexes. The strings in the sub-encryption layer are randomly shuffled to establish sub-index logic. The order of the strings is shuffled using a preset randomization algorithm, and the shuffled strings are reconnected based on the sub-index logic to generate a sub-checksum. During decryption of the main encryption layer, the main checksum is used for verification. If the main checksum verification passes, the second instruction content is directly obtained. If verification fails or the decryption time exceeds the threshold, the string logic of the sub-encryption layer is gradually restored based on the sub-index logic and the sub-checksum until decryption is complete.

[0040] For the third, fourth, and fifth instructions that employ a preset standard encryption mechanism, decryption is performed using the corresponding standard decryption algorithm and key.

[0041] The beneficial effects of this invention are as follows:

[0042] 1. This invention employs a key management strategy to perform hierarchical encryption processing on optimized instructions: a lightweight key encryption mechanism is used for highly time-sensitive instructions, with a pre-set verification stage and dynamic key reconstruction in the transmission channel, significantly reducing encryption latency; a delayed key encryption mechanism is used for highly sensitive instructions, strengthening protection through a nested structure of main encryption layer and sub-encryption layer. When the main layer verification fails, the sub-layer decryption logic is activated to ensure the anti-attack capability of sensitive instructions. This mechanism solves the core contradiction of high-time-sensitive instructions being delayed due to excessive encryption and highly sensitive instructions being at risk of leakage due to insufficient protection under a fixed encryption strategy.

[0043] 2. This invention employs differentiated encryption logic for different command types: lightweight encryption uses a key expansion algorithm to quickly reconstruct commands, ensuring the real-time performance of emergency stop commands; delayed encryption establishes defense-in-depth through multi-level indexing, gradually reconstructing strings based on sub-indexes when the main verification fails, thus resisting brute-force attacks while avoiding the performance loss of high-strength encryption throughout the process. This design ensures the security of highly sensitive commands while optimizing the overall system response efficiency through dynamic adaptation of encryption strength, meeting the balance between security and timeliness requirements in IoT scenarios. Attached Figure Description

[0044] Figure 1 This is a flowchart of the method of the present invention;

[0045] Figure 2 This is a flowchart of the method for generating the prediction model of the present invention. Detailed Implementation

[0046] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby providing a clearer and more explicit definition of the scope of protection of the present invention.

[0047] Please see Figure 1 - Figure 2 A smart property management multi-functional data fusion and predictive optimization control method includes the following steps:

[0048] S1. Collect multi-source data of the property through a preset sensor network and generate a fused dataset;

[0049] S2. Build a rule base based on the fused dataset: Generate basic rules based on the fused dataset, configure preset compensation rules for the basic rules, and calculate the trigger threshold of the compensation rules;

[0050] S3. Analyze the fused dataset based on the preset prediction model to generate a predicted trend. If the predicted trend conforms to the basic rules in the rule base, a standard control instruction is generated. If the predicted trend deviates from the basic rules in the rule base and exceeds the trigger threshold of the compensation rule, an optimization instruction is generated according to the compensation rule, and a preset key management strategy is executed on the optimization instruction.

[0051] S4. Send standard control commands to terminal devices; decrypt optimized commands encrypted by the key management strategy in real time and transmit them to preset terminal devices.

[0052] S5. Collect the execution results of standard control commands and optimization commands, as well as feedback data from terminal devices, and update compensation rules and key management strategies.

[0053] Based on steps S1-S5, real-time fusion and intelligent prediction of multi-source property data are realized. Adaptive control instructions are generated through a rule base. When the predicted trend deviates from the basic rules and exceeds the threshold, encryption optimization instructions are automatically triggered. Combined with key management and execution feedback closed-loop optimization, the accuracy of equipment control and system security are significantly improved. This solves the problems of instruction delay and insufficient protection caused by traditional static encryption strategies, while significantly reducing operation and maintenance complexity and costs.

[0054] One embodiment of the present invention is as follows: the multi-source property data in step S1 includes security monitoring data, equipment operation status data, environmental sensor data, energy pipeline network data and personnel activity data that are independent of each other in terms of functional dimensions.

[0055] Specifically, the steps for generating the fused dataset in step S1 are as follows:

[0056] S11. Based on the sensor network, real-time security monitoring data (including video streams and access control logs), equipment operation status data (including elevator vibration signals and air conditioner power consumption curves), environmental sensing data (including temperature, humidity and PM2.5 readings), energy pipeline data (including water current pulses) and personnel activity data (including Bluetooth beacon trajectories and gate passage counts) are acquired, aligned and mapped to a preset storage space through timestamps. The security monitoring data, equipment operation status data, environmental sensing data, energy pipeline data and personnel activity data are stored in each storage unit of the storage space, and the storage units are stacked to construct the original data cube. The storage space is a memory, which is set in the terminal device.

[0057] Sensor networks include five types of sensor clusters: security sensing clusters, equipment status monitoring clusters, environmental sensing clusters, energy metering clusters, and personnel dynamic tracking clusters;

[0058] The security sensing cluster includes video surveillance nodes (ultra-high-definition cameras integrating behavior recognition algorithms to capture abnormal behavior in real time), intelligent access control terminals (equipped with dual-frequency RFID and liveness detection facial recognition modules to simultaneously record the identity, time and biometrics of passers-by), and perimeter security sensors (using distributed fiber optic vibration sensors and deploying sound and light linkage alarm devices along the perimeter wall).

[0059] The equipment status monitoring cluster allocates an elevator operation monitoring module (a three-axis vibration sensor is installed at the traction machine bearing of the elevator to synchronously collect current harmonic characteristics), an air conditioning efficiency sensor (an embedded meter captures the compressor power consumption curve in real time, and evaluates the air conditioning cooling efficiency in conjunction with the air outlet temperature and humidity probe), and a water pump / distribution cabinet monitoring terminal (integrating a multi-parameter fusion diagnostic unit for temperature, noise, and insulation resistance).

[0060] The environmental sensing cluster includes microclimate monitoring stations (with grid-based deployment of temperature and humidity composite probes, PM2.5 composite probes, and VOC composite probes) and light intensity sensors (installing illuminance meters in public areas and linking them to the property's intelligent lighting system).

[0061] The energy metering cluster includes a hydroelectric pulse acquisition unit (using a magnetoresistive flow meter (error ±0.5%) and a LoRaWAN communication module) and a steam / gas metering unit (turbine flow meter);

[0062] The personnel dynamic tracking cluster includes a Bluetooth beacon positioning network (deploying beacons based on the iBeacon protocol and drawing personnel heat maps through RSSI fingerprint positioning) and a gate access counter;

[0063] S12. Preprocess the original data cube and extract data from each storage unit: For security monitoring data, extract personnel activity frequency (number of times people pass through per unit time) and object movement speed (calculated based on inter-frame displacement of the video stream) to generate the first feature data; For equipment operating status data, extract equipment operating time (cumulative working time / day) and equipment power change rate (standard deviation of equipment power fluctuation per unit time) to generate the second feature data; For environmental sensing data, extract temperature and humidity change gradient (hourly temperature and humidity difference) and PM2.5 concentration fluctuation value (between adjacent sampling points) to generate the second feature data. The first, second, third, fourth, and fifth feature data are generated by extracting the absolute value of the concentration difference. For energy network data, the peak frequency (maximum amplitude of the pulse signal) and abnormal fluctuation periods (time windows where the amplitude exceeds a threshold) of the hydroelectric current pulses are extracted to generate the fourth feature data. For personnel activity data, the hotspot areas of Bluetooth beacon trajectories (coordinates of areas where Bluetooth beacons stay for more than 5 minutes) and the time-varying characteristics of gate traffic density (the ratio of peak to off-peak traffic volume) are extracted to generate the fifth feature data. These feature data are then fused to generate a fused dataset. The fusion process can employ methods such as weighted averaging and principal component analysis (existing technologies will not be elaborated upon here) to ensure that the fused dataset comprehensively reflects the key information from the multi-source property data.

[0064] Based on the collaborative deployment and feature-oriented extraction mechanism of five types of sensor clusters, millimeter-level data coverage of the entire property operation status is achieved. A multimodal data cube is constructed by aligning timestamps, and key feature values ​​such as the frequency of personnel activities, the rate of change of equipment power, and the temperature and humidity gradient are accurately extracted. This ensures that the fused dataset is deeply coupled with the dynamics of physical space and business logic, providing a high-fidelity, low-latency, and strongly correlated data foundation for predictive optimization control, and significantly improving the accuracy of system situational awareness and the reliability of decision-making.

[0065] One embodiment of the present invention is as follows: the construction of the rule base in step S2 includes:

[0066] S21. Combine the frequency of personnel activity and the speed of object movement in the first feature data into a security behavior rule; associate the device runtime and the device power change rate in the second feature data into a device health rule; map the temperature and humidity change gradient and PM2.5 concentration fluctuation value in the third feature data into an environmental anomaly rule; convert the frequency peak and abnormal fluctuation period of the water current pulse in the fourth feature data into an energy failure rule; integrate the dwell hotspot area of ​​the Bluetooth beacon trajectory and the time-varying characteristics of the gate passage density in the fifth feature data into a human flow density rule; and use the security behavior rule, device health rule, environmental anomaly rule, energy failure rule, and human flow density rule as the basic rules.

[0067] Security behavior rule generation steps: By analyzing the frequency of personnel activity (number of passages per unit time) and the speed of object movement (calculated based on the inter-frame displacement of the video stream) in the first feature data in real time, a dynamic behavior threshold model is established: When the frequency of personnel activity exceeds the preset safety threshold and the speed of object movement continues to accelerate abnormally (such as speed of intrusion into a restricted area > 2m / s), the "abnormal intrusion" judgment rule is automatically triggered, and the behavior judgment boundary is optimized by combining historical security event data to ensure that the rule responds accurately to risks such as intrusion and gathering;

[0068] Equipment health rule generation steps: Based on the second feature data, the equipment running time (cumulative working time / day) and the equipment power change rate (standard deviation of power fluctuation per unit time), a multi-dimensional health assessment model is constructed: by dynamically associating the running time attenuation coefficient with the power fluctuation threshold (e.g., elevator running for more than 200 hours / day and power fluctuation > 15%), the equipment health index is quantified, and the failure rate (failure statistics of similar equipment) and maintenance frequency are integrated to generate health level rules (health / early warning / failure);

[0069] Environmental anomaly rule generation steps: The environmental anomaly rule establishes an environmental risk coupling model by mapping the temperature and humidity change gradient (hourly difference value) of the third feature data with the PM2.5 concentration fluctuation value (absolute value of the concentration difference between adjacent sampling points): When the temperature and humidity gradient changes abruptly (such as temperature difference > 5℃ within 10 minutes) and the PM2.5 fluctuation value continuously exceeds the standard, the "environmental anomaly event" rule is triggered, and the fluctuation sensitivity is calibrated based on historical environmental accident data;

[0070] Energy Fault Rule Generation Steps: The energy fault rule uses the peak frequency of the water current pulse (maximum amplitude of the pulse signal) and the abnormal fluctuation period (amplitude exceeding the threshold time window) of the fourth feature data to construct a fault feature time series model: by detecting the attenuation of the frequency peak (such as a 40% decrease in amplitude) and the duration of abnormal fluctuation (>30 minutes), and associating it with the pipeline pressure sensor data, a "leakage / blockage" fault rule is generated, and the fault judgment threshold is optimized by combining equipment maintenance records;

[0071] The steps for generating pedestrian density rules are as follows: The pedestrian density rules integrate the Bluetooth beacon trajectory hotspot area (coordinates of areas where the stay is >5 minutes) of the fifth feature data with the time-varying characteristics of the gate passage density (peak / off-peak passage ratio) to construct a spatial-temporal density model: When the population density in the hotspot area exceeds the safe capacity (e.g., >3 people / ㎡) and the peak passage ratio suddenly increases (>200% of the historical average), the "exceeding pedestrian limit" rule is triggered, and the trajectory data is smoothed by Kalman filtering to improve the real-time performance of the rule;

[0072] S22. Configure preset compensation rules for various basic rules: Configure the first dynamic sub-rule, the second dynamic sub-rule, the third dynamic sub-rule, the fourth dynamic sub-rule and the fifth dynamic sub-rule for security behavior rules, equipment health rules, environmental anomaly rules, energy failure rules and people density rules respectively;

[0073] Specifically, the content of the first dynamic sub-rule in step S22 is as follows: Based on the real-time monitoring of personnel activity frequency and object movement speed in the security behavior rules, and superimposed with the coordinate clustering distribution of the Bluetooth beacon trajectory of the hotspot area generated by the crowd density rules, a spatial density model of the crowd heat map is constructed, and a crowd diversion compensation strategy is preset: Based on the real-time constructed spatial density model of the crowd heat map, when the monitoring shows that the personnel activity frequency exceeds the historical peak baseline and the standard deviation of the object movement speed exceeds the preset fluctuation range, the system automatically executes a graded response mechanism—first, it links the intelligent access control terminal and the gate access counter to start one-way flow restriction; at the same time, it pushes evacuation route navigation to the personnel in the hotspot area through the Bluetooth beacon positioning network, and publishes diversion routes and congestion warnings in real time on the electronic guide screen; further, it dynamically activates the emergency evacuation channel, forcibly unlocks the access control lock of non-safety exits and activates the sound and light guidance device; finally, it synchronously dispatches security personnel to key nodes to implement manual intervention, and tracks the diversion effect in real time through video monitoring nodes.

[0074] The second dynamic sub-rule is as follows: It incorporates the Shannon entropy fluctuation characteristics of the cumulative wear coefficient of equipment runtime and the equipment power change rate from the associated equipment health rules, combined with the equipment physical connection topology provided by the energy fault rules, to construct an equipment health decay chain function; and pre-sets a device fault isolation protocol compensation strategy: First, based on the equipment physical connection topology, it locates the faulty equipment and its associated nodes (such as air conditioning units associated with a distribution cabinet); second, it remotely cuts off the power supply to the faulty equipment through pre-set intelligent power devices; simultaneously, it pushes fault location information and a spare parts replacement list to maintenance personnel; finally, it automatically updates the rule base, lowering the power fluctuation threshold of the faulty equipment by 20% to enhance monitoring sensitivity, until repair is completed and the health baseline is reset.

[0075] The third dynamic sub-rule is as follows: It integrates the temperature and humidity change gradient difference sequence of the environmental anomaly rule with the extreme value of PM2.5 concentration fluctuation, and injects the time-varying characteristic ratio of the gate passage density in real time feedback of the crowd density rule; and presets a fresh air system linkage control compensation strategy: based on the sudden change of the temperature and humidity gradient difference sequence (such as temperature difference > 5℃ within 10 minutes) superimposed with the extreme value of PM2.5 concentration exceeding the standard and the sudden increase of the gate passage density ratio (peak / off-peak > 200%), the system starts linkage control: dynamically adjusts the fresh air return air ratio, and increases the fresh air volume of the air conditioner to 120% of the design value in densely populated areas; intelligently matches the air conditioner output, and automatically increases the outlet temperature (winter +3℃) or relaxes the room temperature setting value (summer +2℃) according to the fresh air handling load to neutralize the increase in energy consumption caused by the increased ventilation;

[0076] The fourth dynamic sub-rule includes: the peak frequency spectrum characteristics of water current pulse frequency and the duration of abnormal fluctuations in the energy failure rule; the real-time power fluctuation entropy value fed back by the health rule of associated equipment; and a preset emergency pressure reduction operation compensation strategy for the water and electricity network: graded disconnection of non-critical loads, priority shutdown of landscape water system and floodlight circuits in the property intelligent lighting system to ensure power supply to core equipment such as elevators / fire pumps; activation of reverse voltage regulation mechanism to temporarily increase the bus voltage by 5% through the power distribution room AVC system to compensate for the end voltage drop caused by pipeline leakage; dynamic adjustment of pump frequency, identification of blockage points based on spectrum characteristics, allocation of flow in faulty pipe sections to redundant branches, and issuance of pressure reduction commands through the LoRaWAN network to prevent pipe bursts;

[0077] The fifth dynamic sub-rule is as follows: It integrates the Bluetooth beacon hotspot clustering index of the pedestrian density rule with the ratio of the time-varying characteristics of the turnstile passage density, and superimposes the spatial correlation weight of the object movement speed in the security behavior rule; and presets a graded flow restriction plan compensation strategy: Level 1 flow restriction: close secondary entrance and exit turnstiles and forcibly activate one-way circulation channels; Level 2 diversion: publish heat maps and diversion paths through the property electronic guide screen, and push Bluetooth beacons to target low-density area navigation; Level 3 control: link the security system to unlock emergency exits and activate sound and light guidance, and simultaneously dispatch security robots to cluster coordinate points to implement manual intervention; Dynamic optimization: iteratively update the flow restriction threshold and diversion strategy to the rule base based on real-time feedback to improve the accuracy of future early warnings;

[0078] S23. Calculate the trigger threshold for each compensation rule based on various basic rules. Complete the construction of the rule base by using various basic rules, the compensation rules configured by various basic rules, and the trigger thresholds of each compensation rule.

[0079] Among them: the threshold for the first dynamic sub-rule to respond to sudden peaks in pedestrian flow; and the associated parameters for the second dynamic sub-rule to respond to cascading equipment failure scenarios.

[0080] The specific details of calculating the trigger threshold for each compensation rule based on various basic rules in step S23 are as follows:

[0081] a. Extract the historical peak value of personnel activity frequency and the abnormal fluctuation range of object movement speed from the security behavior rules, and combine the spatial distribution characteristics of real-time pedestrian flow heat map with the time decay factor (peak period weight coefficient) to generate dynamic offset of sudden peak pedestrian flow scenarios; when the real-time dynamic offset exceeds the historical baseline offset × safety margin coefficient (preferably 0.9), determine the trigger threshold (first dynamic sub-rule).

[0082] b. Construct an equipment health decay function by using the cumulative wear coefficient of equipment runtime and the Shannon entropy fluctuation value of equipment power change rate in the associated equipment health rules; calculate the chain risk transmission coefficient between the target equipment and associated equipment based on the equipment fault propagation topology map of energy pipeline data: when the equipment health decay function value is lower than the preset health threshold and the chain risk transmission coefficient exceeds the risk threshold, determine the trigger threshold (second dynamic sub-rule).

[0083] c. Map the gradient difference sequence of temperature and humidity changes and the extreme value of PM2.5 concentration fluctuation in the environmental anomaly rule, inject the time-varying characteristic ratio of gate passage density in the real-time pedestrian flow density rule, and construct the environmental comfort risk index; when the environmental comfort risk index exceeds the historical mean + 3 times the standard deviation, determine the trigger threshold (third dynamic sub-rule).

[0084] d. Convert the peak frequency spectrum characteristics of the water current pulse and the duration of abnormal fluctuations in the energy fault rules, and associate them with the real-time power fluctuation entropy value fed back by the equipment health rules to generate an energy fault feature vector; extract the key components of the energy fault feature vector through principal component analysis, and determine the trigger threshold (fourth dynamic sub-rule) when the score of the key component exceeds the preset risk threshold.

[0085] e. Integrate the Bluetooth beacon hotspot clustering index of the pedestrian density rule with the time-varying characteristic ratio of the gate passage density, and superimpose the spatial correlation weight of the object movement speed in the security behavior rule to construct a pedestrian congestion risk assessment model; when the output value of the pedestrian congestion risk assessment model exceeds the preset congestion warning threshold, determine the trigger threshold (fifth dynamic sub-rule).

[0086] By calculating trigger thresholds for each compensation rule based on various basic rules, the system integrates these basic rules, their corresponding compensation rules, and the trigger thresholds for each compensation rule to construct a rule base. This rule base provides comprehensive and accurate rule support for multifunctional data fusion and predictive optimization control in smart property management. It enables the system to respond quickly and effectively to different real-time situations, further improving property management efficiency and service quality, and ensuring the safety and stable operation of personnel, equipment, and the environment within the property area. For example, during sudden peak periods of pedestrian traffic, the system can quickly activate pedestrian flow management compensation strategies based on relevant rules in the rule base to ensure the safe evacuation of personnel. When equipment malfunctions, it can promptly execute equipment fault isolation protocol compensation strategies to reduce the impact of the fault on the entire system. Simultaneously, the trigger threshold calculation method in the rule base fully considers the characteristics and interrelationships of various data, ensuring the scientific validity and practicality of the rules.

[0087] One embodiment of the present invention is as follows: the method for generating the prediction model in step S3 is as follows:

[0088] S31. Obtain the historical fusion dataset, extract the initial, intermediate, and late data of the first, second, third, fourth, and fifth feature data in the historical fusion dataset within the same first sampling period, and extract the initial, intermediate, and late data collected in the second sampling period adjacent to the same sampling period as the training set, validation set, and test set, respectively.

[0089] S32. Map the initial, intermediate, and late data from the previous sampling period to a preset neural network model, and divide the initial, intermediate, and late data into grid regions in the neural network model respectively. Through the neural network model, adaptively expand the boundaries of the initial, intermediate, and late data to obtain expanded training subsets, expanded validation subsets, and expanded test subsets. Map the expanded training subsets, expanded validation subsets, and expanded test subsets to each grid region to generate a first expanded region, a second expanded region, and a third expanded region connected to each grid region.

[0090] S33. Map the training set, validation set, and test set within the second sampling period to each grid region in the neural network model, generate a first virtual extended region, a second virtual region, and a third virtual extended region connected to each grid region, calculate whether the difference between the first extended region, the second extended region, the first virtual extended region, the second virtual extended region, and the third virtual extended region meets the preset conditions. If it does, use the neural network model as the prediction model. If it does not meet the preset conditions, adjust the parameters of the neural network model according to the difference.

[0091] Specifically, the steps for generating the optimization instructions in step S3 are as follows:

[0092] S301. Identify the current scenario category of the property based on the predicted change trend, and call the compensation strategy defined in the compensation rules;

[0093] S302. Extract the first feature data, second feature data, third feature data, fourth feature data and fifth feature data associated with the compensation strategy from the rule base, and combine them with the compensation strategy to generate the first instruction content, second instruction content, third instruction content, fourth instruction content and fifth instruction content through a preset control logic template. The first instruction content, second instruction content, third instruction content, fourth instruction content and fifth instruction content are all used as optimization instructions.

[0094] Specifically, the key management strategy in step S3 is as follows: a preset lightweight key encryption mechanism is adopted for the first instruction content, that is, a preset verification stage is set in the transmission channel between the first instruction content and the terminal device. The verification stage includes key registration and key reconstruction. Key registration refers to binding the first instruction content with the preset initial key before transmission to generate instruction data with the initial key identifier; key reconstruction is to dynamically update and transform the initial key during transmission (according to preset algorithms and rules) to form a new key, and bind the new key with the instruction data and transmit it together to the preset decryption module.

[0095] The second instruction content is encrypted using a preset delayed key encryption mechanism to generate an encryption layer that encapsulates the second instruction content;

[0096] The encryption layer includes a main encryption layer and several sub-encryption layers. The sub-encryption layers are hidden when the main encryption layer is successfully decrypted or not decrypted. The sub-encryption layers are revealed layer by layer when the main encryption layer fails to decrypt or the decryption time exceeds a preset threshold.

[0097] The steps for generating the encryption layer are as follows: extract the string from the second instruction content as the master node, and establish a master connection index based on the association between the master nodes; weave the master nodes into a first spherical mesh structure according to the master connection index to form the main encryption layer that wraps the second instruction content; randomly shuffle the string nodes of the second instruction content to generate child nodes, and establish a sub-connection index between the child nodes; weave the child nodes into a second spherical mesh structure according to the sub-connection index to form a sub-encryption layer that wraps around the main encryption layer.

[0098] The third, fourth, and fifth instructions are encrypted using a preset standard encryption mechanism, preferably the same as the delayed key encryption mechanism.

[0099] Specifically, step S3 also includes maintaining the current operating status of the property's equipment if the predicted trend of change deviates from the basic rules in the rule base but does not exceed the trigger threshold of the compensation rule;

[0100] In this embodiment, step S3 involves acquiring a historical fusion dataset, extracting the initial, intermediate, and late-stage data of the first to fifth feature data within the same sampling period, and using this data, along with data from adjacent sampling periods, as the training set, validation set, and test set, respectively. Subsequently, the data from the previous sampling period is mapped to a neural network model, dividing the data into grid regions and adaptively expanding the boundaries to generate expanded training, validation, and test subsets. These expanded subsets are then mapped to the grid regions, forming the first to third expanded regions. Next, the data from the second sampling period is mapped to the grid regions to generate a virtual expanded region. The difference between the expanded region and the virtual expanded region is calculated to determine if it meets a preset condition. If it does, the neural network model is used as the prediction model; otherwise, the model parameters are adjusted. The optimization instruction generation identifies the property scenario category, invokes the compensation strategy, extracts relevant feature data, and combines it with the control logic template to generate five instruction contents as optimization instructions. In the key management strategy, the first instruction uses lightweight key encryption, including key registration and reconstruction; the second instruction uses delayed key encryption, generating a main encryption layer and a sub-encryption layer; and the third to fifth instructions use standard encryption mechanisms. If the predicted trend deviates from the basic rules but does not exceed the threshold, the current state of the equipment is maintained. This approach ensures both the accuracy and adaptability of the prediction model, generates reasonable and effective optimization instructions based on different property scenarios, and guarantees the security of instruction transmission through diverse key management strategies. In practical applications, this method can effectively handle various complex situations in property operations, improving the operational efficiency and management level of property equipment.

[0101] One embodiment of the present invention is as follows: before sending the standard control command to the terminal device in step S4, a preset encryption operation is performed on the standard control command, preferably the encryption operation is the same as the delayed key encryption mechanism.

[0102] Specifically, the real-time decryption of the optimization instructions encrypted by the key management strategy in step S4 includes:

[0103] Execution of the first instruction, which employs a lightweight key encryption mechanism:

[0104] The decryption module extracts the feature value from the new key and matches the feature value with the preset key mapping rule to generate the seed value of the initial key. Based on the seed value, the preset key expansion algorithm generates the initial key. The instruction data is decrypted according to the initial key to obtain the first instruction content.

[0105] Execution of the second instruction content using a delayed key encryption mechanism:

[0106] The decryption module extracts strings from the second instruction content, analyzes the semantic relationships and structural connections between strings, and establishes a multi-level index between strings based on these relationships and connections. This multi-level index includes a main index and sub-indexes. The main index identifies the overall structure of the string, while the sub-indexes identify the internal logical relationships. A main checksum is generated based on the main index, and sub-checksums are generated based on the sub-indexes. The strings in the sub-encryption layer are randomly shuffled to establish sub-index logic. The order of the strings is shuffled using a preset randomization algorithm, and the shuffled strings are reconcatenated based on the sub-index logic to generate a sub-checksum. During decryption of the main encryption layer, the main checksum is used for verification. If the main checksum verification passes, the second instruction content is directly obtained. If verification fails or the decryption time exceeds a threshold, the string logic of the sub-encryption layer is gradually restored based on the sub-index logic and the sub-checksum until decryption is complete.

[0107] For the third, fourth, and fifth instructions that employ a preset standard encryption mechanism, decryption is performed using the corresponding standard decryption algorithm and key.

[0108] In this embodiment of the invention, step S4, which encrypts the standard control command, employs the same multi-level dynamic encryption system as the delayed key encryption mechanism: during encryption, the command string is extracted to establish a master index (identifying the overall structure) and a sub-index (identifying internal logic). A preset randomization algorithm scrambles the string order to generate sub-check codes, and the master / sub-check codes are encapsulated into layered ciphertext. During decryption, the integrity of the master check code is first verified. If the network latency exceeds a threshold or verification fails, the scrambled strings of the sub-encryption layers are gradually reassembled according to the sub-index logic. Finally, the initial command content is restored using a key expansion algorithm. This mechanism optimizes dynamic decryption efficiency through layered triggering while ensuring security.

[0109] After collecting the command execution results in step S5, the compensation rules are dynamically adjusted based on these results. Specifically, the actual effectiveness of the compensation strategy in different scenarios is analyzed, and the effectiveness of the compensation rules is evaluated by combining the basic rules and trigger thresholds in the rule base. If the compensation strategy achieves the expected effect in a certain scenario, such as successfully evacuating people safely by activating the crowd control compensation strategy during a sudden peak in pedestrian traffic, and no situation exceeding the expectations of the rule base occurs, then the compensation rule and its trigger threshold remain unchanged. If the execution result does not meet expectations, such as when equipment malfunctions, and the impact of the fault on the system is still significant after implementing the equipment fault isolation protocol compensation strategy, then a deeper analysis of the reasons is needed. It may be that the trigger threshold setting is unreasonable, leading to the compensation strategy being activated untimely or over-activated; or it may be that the compensation strategy itself has defects and cannot effectively cope with the current fault situation. To address these issues, the trigger threshold of the compensation rule is adjusted accordingly, or the specific content of the compensation strategy is optimized. For example, if it is found that the threshold settings of the equipment health decay function value and the cascading risk transmission coefficient cause the equipment fault isolation protocol to be activated too late, the preset health threshold of the equipment health decay function can be appropriately reduced, or the risk threshold of the cascading risk transmission coefficient can be increased. Simultaneously, by combining historical execution results and real-time data, a dynamic update mechanism for compensation rules is established. These rules are regularly evaluated and adjusted to ensure the rule base remains adaptable to various complex situations in property operations, further improving management efficiency and service quality, and guaranteeing the safety and stable operation of personnel, equipment, and the environment within the property area. Furthermore, the updating of compensation rules fully considers the correlation and synergy between different rules to avoid negative impacts on other rules due to adjustments in one rule. In this way, the rule base is continuously improved, becoming a solid foundation for multifunctional data fusion and predictive optimization control in smart property management.

[0110] And dynamically adjust the encryption mechanism:

[0111] If a certain type of optimization instruction is frequently intercepted but not cracked (e.g., three attacks are triggered within a month but fail), the current encryption strength can be maintained;

[0112] If the new attack method causes the decryption delay to exceed the threshold (e.g., >500ms), then the encryption strength of such optimized instructions will be enhanced. For example, for instructions using lightweight key encryption mechanisms, the key expansion algorithm will be adjusted to increase the complexity and randomness of the key; for instructions using delayed key encryption mechanisms, the number of sub-encryption layers will be increased or the weaving method of the main encryption layer and sub-encryption layers will be changed to make the encryption structure more complex; for instructions using standard encryption mechanisms, a higher-level standard encryption algorithm will be used.

[0113] The above are merely embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A smart property management multi-functional data fusion and predictive optimization control method, characterized by: Includes the following steps: S1. Collect multi-source data of the property through a preset sensor network and generate a fused dataset; S2. Construct a rule base based on the fused dataset: Generate basic rules based on the fused dataset, configure preset compensation rules for the basic rules, and calculate the trigger threshold of the compensation rules; S3. Analyze the fused dataset based on the preset prediction model to generate a predicted change trend. If the predicted change trend conforms to the basic rules in the rule base, a standard control instruction is generated. If the predicted change trend deviates from the basic rules in the rule base and exceeds the trigger threshold of the compensation rule, an optimization instruction is generated according to the compensation rule, and a preset key management strategy is executed on the optimization instruction. S4. Send the standard control command to the terminal device; decrypt the optimized command encrypted by the key management strategy in real time and transmit it to the preset terminal device; S5. Collect the execution results of the standard control instructions and the optimization instructions, as well as the feedback data from the terminal device, and update the compensation rules and the key management strategy.

2. The smart property multi-functional data fusion and predictive optimization control method according to claim 1, characterized in that: The multi-source property data in step S1 includes security monitoring data, equipment operation status data, environmental sensor data, energy pipeline data, and personnel activity data.

3. The smart property multi-functional data fusion and predictive optimization control method according to claim 2, characterized in that: The steps for generating the fused dataset in step S1 are as follows: S11. Based on the sensor network, security monitoring data, equipment operation status data, environmental sensing data, energy pipeline data and personnel activity data are acquired in real time, aligned and mapped to a preset storage space through timestamps, and the security monitoring data, equipment operation status data, environmental sensing data, energy pipeline data and personnel activity data are stored in each storage unit of the storage space, and the storage units are stacked to construct the original data cube. S12. Preprocess the original data cube and extract the data in each storage unit: For security monitoring data, extract the frequency of personnel activities and the speed of object movement to generate first feature data; For equipment operating status data, the operating time and power change rate of the equipment are extracted to generate the second feature data; for environmental sensing data, the temperature and humidity change gradient and PM2.5 concentration fluctuation value are extracted to generate the third feature data; for energy pipeline network data, the frequency peak value and abnormal fluctuation period of water current pulse are extracted to generate the fourth feature data; for personnel activity data, the time-varying features of Bluetooth beacon trajectory hotspot areas and gate passage density are extracted to generate the fifth feature data; the first feature data, the second feature data, the third feature data, the fourth feature data and the fifth feature data are fused to generate a fused dataset.

4. The smart property multi-functional data fusion and predictive optimization control method according to claim 3, characterized in that: The construction of the rule base in step S2 includes: S21. Combine the frequency of personnel activity and the speed of object movement of the first feature data into a security behavior rule; associate the device running time and the device power change rate of the second feature data into a device health rule; map the temperature and humidity change gradient and PM2.5 concentration fluctuation value of the third feature data into an environmental anomaly rule; convert the frequency peak value and abnormal fluctuation period of the water current pulse of the fourth feature data into an energy failure rule; integrate the dwell hotspot area of ​​the Bluetooth beacon trajectory and the time-varying feature of the gate passage density of the fifth feature data into a human flow density rule; and use the security behavior rule, the device health rule, the environmental anomaly rule, the energy failure rule, and the human flow density rule as the basic rules. S22. Configure preset compensation rules for each type of basic rule: Configure a first dynamic sub-rule, a second dynamic sub-rule, a third dynamic sub-rule, a fourth dynamic sub-rule and a fifth dynamic sub-rule for the security behavior rule, the equipment health rule, the environmental anomaly rule, the energy failure rule and the crowd density rule respectively. S23. Calculate the trigger threshold for each compensation rule according to the various basic rules, and complete the construction of the rule base by using the various basic rules, the compensation rules configured by the various basic rules, and the trigger thresholds of each compensation rule.

5. The smart property multi-functional data fusion and predictive optimization control method according to claim 4, characterized in that: The method for generating the prediction model in step S3 is as follows: S31. Obtain the historical fusion dataset, extract the initial, intermediate, and late data of the first feature data, second feature data, third feature data, fourth feature data, and fifth feature data in the same first sampling period from the historical fusion dataset, and extract the initial, intermediate, and late data collected in the second sampling period adjacent to the same sampling period as the training set, validation set, and test set, respectively. S32. Map the initial data, intermediate data, and late data in the previous sampling period to a preset neural network model, and divide the initial data, intermediate data, and late data into grid regions in the neural network model respectively. Through the neural network model, adaptively expand the boundaries of the initial data, intermediate data, and late data to obtain expanded training subsets, expanded verification subsets, and expanded test subsets. Map the expanded training subsets, expanded verification subsets, and expanded test subsets to each of the grid regions to generate a first expanded region, a second expanded region, and a third expanded region connected to each of the grid regions. S33. Map the training set, validation set, and test set within the second sampling period to each of the grid regions in the neural network model, generate a first virtual extended region, a second virtual region, and a third virtual extended region connected to each of the grid regions, calculate whether the difference between the first extended region, the second extended region, and the first virtual extended region, the second virtual region, and the third virtual extended region meets a preset condition. If it does, use the neural network model as the prediction model; if it does not meet the preset condition, adjust the parameters of the neural network model according to the difference.

6. The smart property multi-functional data fusion and predictive optimization control method according to claim 5, characterized in that: The steps for generating the optimization instructions in step S3 are as follows: S301. Identify the current scenario category of the property based on the predicted change trend, and call the compensation strategy defined in the compensation rules; S302. Extract the first feature data, second feature data, third feature data, fourth feature data, and fifth feature data associated with the compensation strategy from the rule base, and combine them with the compensation strategy to generate first instruction content, second instruction content, third instruction content, fourth instruction content, and fifth instruction content through a preset control logic template. The first instruction content, second instruction content, third instruction content, fourth instruction content, and fifth instruction content are all used as optimization instructions.

7. The smart property multi-functional data fusion and predictive optimization control method according to claim 6, characterized in that: The specific content of the key management strategy in step S3 is as follows: a preset lightweight key encryption mechanism is adopted for the first instruction content, that is, a preset verification stage is set in the transmission channel between the first instruction content and the terminal device. The verification stage includes key registration and key reconstruction. Key registration refers to binding the first instruction content with a preset initial key before transmission to generate instruction data with the initial key identifier. Key reconstruction is to dynamically update and transform the initial key during transmission to form a new key, and bind the new key with the instruction data and transmit them together to the preset decryption module. The second instruction content is encrypted using a preset delayed key encryption mechanism to generate an encryption layer that encapsulates the second instruction content; The encryption layer includes a main encryption layer and several sub-encryption layers. The several sub-encryption layers are hidden when the main encryption layer successfully decrypts or fails to decrypt. The several sub-encryption layers are revealed layer by layer when the main encryption layer fails to decrypt or the decryption time exceeds a preset threshold. The steps for generating the encryption layer are as follows: extracting the string from the second instruction content as the master node, and establishing a master connection index based on the association between the master nodes; weaving the master nodes into a first spherical mesh structure according to the master connection index to form a main encryption layer that wraps the second instruction content; randomly shuffling the string nodes of the second instruction content to generate child nodes, and establishing a sub-connection index between the child nodes; weaving the child nodes into a second spherical mesh structure according to the sub-connection index to form a sub-encryption layer that wraps around the outside of the main encryption layer; A preset standard encryption mechanism is applied to the third, fourth, and fifth instruction contents.

8. The smart property multi-functional data fusion and predictive optimization control method according to claim 7, characterized in that: Step S3 further includes maintaining the current operating status of the property's equipment if the predicted change trend deviates from the basic rules in the rule base but does not exceed the trigger threshold of the compensation rule; Before sending the standard control command to the terminal device in step S4, a preset encryption operation is performed on the standard control command.

9. The smart property multi-functional data fusion and predictive optimization control method according to claim 8, characterized in that: The specific content of real-time decryption of the optimized instruction encrypted by the key management strategy in step S4 includes: Execution of the first instruction content using the aforementioned lightweight key encryption mechanism: The decryption module extracts feature values ​​from the new key and matches these feature values ​​with a preset key mapping rule to generate a seed value for the initial key. Based on the seed value, an initial key is generated using a preset key expansion algorithm. The instruction data is then decrypted using the initial key to obtain the content of the first instruction. For the execution of the second instruction content using a delayed key encryption mechanism: The decryption module extracts strings from the second instruction content, analyzes the semantic relationships and structural connections between strings, and establishes a multi-level index between strings based on the semantic relationships and structural connections between base strings. This multi-level index includes a main index and sub-indexes. The main index identifies the overall structure of the string, and the sub-indexes identify the internal logical relationships of the string. A main checksum is generated based on the main index, and a sub-checksum is generated based on the sub-indexes. The strings in the sub-encryption layer are randomly shuffled to establish sub-index logic. The order of the strings is shuffled using a preset randomization algorithm, and the shuffled strings are reconnected based on the sub-index logic to generate a sub-checksum. During decryption of the main encryption layer, the main checksum is used for verification. If the main checksum verification passes, the second instruction content is directly obtained. If verification fails or the decryption time exceeds the threshold, the string logic of the sub-encryption layer is gradually restored based on the sub-index logic and the sub-checksum until decryption is complete. For the third, fourth, and fifth instructions that employ a preset standard encryption mechanism, decryption is performed using the corresponding standard decryption algorithm and key.

Images