Intelligent park monitoring management system based on artificial intelligence

Abstract

The invention relates to the technical field of smart park monitoring and management, in particular to a smart park monitoring and management system based on artificial intelligence, which comprises a data acquisition module, a data preprocessing module, an intelligent analysis module, a knowledge graph construction module, a decision generation module, a visual interaction module and an execution feedback module which are directly or indirectly connected with one another, compared with the prior art, the smart park monitoring and management system based on the'collection-processing-analysis-decision-execution 'closed-loop management link is provided, the technologies of the convolutional neural network, the long and short-term memory network, the knowledge graph and the like are fused, and the intelligent park monitoring and management system has the advantages that the intelligent park monitoring and management system is simple in structure, convenient to use and high in practicability. Multi-dimensional data deep analysis and dynamic knowledge association are realized, an accurate diagnosis prediction result and an intelligent decision are generated, a self-optimization closed loop is formed through visual interaction and equipment control feedback, and the park safety protection, equipment operation and maintenance and resource utilization efficiency is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of smart park monitoring and management, and in particular to a smart park monitoring and management system based on artificial intelligence. Background Art

[0002] With the acceleration of urbanization and the popularization of digital technology, smart parks, as a key vehicle for urban intelligence, integrate technologies such as the Internet of Things, artificial intelligence, and big data to achieve collaborative management of people, equipment, and the environment within the park, becoming a core direction for improving park operational efficiency, safety, and resource utilization. Currently, traditional park management faces many challenges.

[0003] Existing smart park monitoring technology has many shortcomings: the deep correlation analysis mechanism of different types of data (such as environment, personnel, and equipment data) is not clear enough, which affects the efficiency of comprehensive judgment; when faced with large amounts of data or sudden anomalies, the real-time update and complex data processing capabilities of the data relationship network are relatively weak; the specific parameters of some core algorithms and the detailed descriptions of multi-algorithm coordination are vague, which may lead to unstable accuracy of functions such as fault prediction; the reverse optimization rules of the system model based on the status data after equipment control are unclear, which limits the system's self-improvement capabilities; in addition, the synchronization mechanism and priority strategy of multi-terminal warning information are not perfect, which may affect the timeliness and efficiency of manual processing.

[0004] In view of the above technical defects, a solution is now proposed. Summary of the Invention

[0005] The purpose of the present invention is to provide an artificial intelligence-based smart park monitoring and management system to solve the problems mentioned in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The AI-based smart park monitoring and management system includes a data acquisition module, a data pre-processing module, an intelligent analysis module, a knowledge graph construction module, a decision-making module, a visual interaction module, and an execution feedback module, which are directly or indirectly connected to each other, forming a closed-loop management chain of "acquisition-processing-analysis-decision-execution";

[0008] The data acquisition module is composed of sensors, cameras, Internet of Things devices and identification devices distributed in different locations in the park. It collects multi-dimensional basic data in the park in real time and transmits it to the data preprocessing module; the data preprocessing module unifies the format, filters noise and extracts features of the multi-dimensional basic data to generate standardized input data; the intelligent analysis module outputs diagnostic prediction structured results based on the standardized input data through machine learning and deep learning algorithms; the knowledge graph construction module constructs a dynamic knowledge graph containing the "personnel-equipment-environment-event" association relationship based on the diagnostic prediction structured results; the decision generation module combines the dynamic knowledge graph to generate equipment control instructions, safety warning signals and resource scheduling plans; the visual interaction module presents the structured results and decision content in the form of charts, monitoring screens and reports, and supports multi-terminal interactive operations; the execution feedback module receives the equipment control instructions from the decision generation module, automatically adjusts the park equipment, and feeds back the adjusted status data to the data preprocessing module.

[0009] Furthermore, the multi-dimensional basic data includes environmental status data, personnel behavior data, equipment operation data, safety parameter data and traffic data; the standardized data set generated by the data preprocessing module contains multi-dimensional basic data after cleaning, format unification, outlier detection and time-space calibration.

[0010] Furthermore, the structured diagnostic prediction results include warning status of combustible gas leakage in equipment rooms, warning status of intrusion into restricted areas, warning status of risk of gathering of people, warning status of abnormal running, warning status of abnormal access to authority, warning status of equipment pre-failure, confirmation status of fire incidents, confirmation status of violent intrusion incidents, warning status of malicious unlocking, warning status of parking space shortage, warning status of abnormal traffic flow during non-peak hours, and warning status of road congestion.

[0011] Furthermore, the intelligent analysis module performs real-time and historical analysis on the standardized data set, specifically including: environmental status data analysis: real-time collection of equipment room temperature, CO2 concentration, and combustible gas concentration values, generating a time series by sliding average of a 5-minute time window, and outputting the predicted value for the next 24 hours through long short-term memory network (LSTM) modeling; when the real-time temperature value is greater than 35°C or the predicted value is greater than 35°C, an "equipment room high temperature warning" is generated; when the real-time CO2 concentration value is greater than 1000ppm or the predicted value is greater than 1000ppm, an "equipment room air quality deterioration warning" is generated; and when the real-time combustible gas concentration value is greater than 35°C or the predicted value is greater than 35°C, an "equipment room air quality deterioration warning" is generated. When the value is greater than 20% of the lower explosion limit or the predicted value is greater than 20% of the lower explosion limit, a warning situation for combustible gas leakage in the equipment room is generated; Personnel behavior data analysis: Analyze the video stream of the campus gate camera, extract the personnel position, speed, and gathering density values, and output the identity matching result through face recognition and comparison with the permission library; When a person enters a restricted area, the gathering density is ≥3 people / ㎡ and lasts for more than 10 minutes, the movement speed of unauthorized personnel is greater than 5m / s, and authorized personnel enter the area without authorization, "Restricted Area Intrusion Warning Situation", "Personnel Gathering Risk Warning Situation", "Abnormal Running Warning Situation", and "Abnormal Permission Access Warning Situation" are generated respectively;

[0012] Further, equipment operation data analysis: normalize the equipment voltage, current, vibration frequency, energy consumption, and cumulative operating time, train the fault classification model through the random forest algorithm to output the fault probability value, and set the average energy consumption during non-working hours (22:00-6:00) as the benchmark value; when the failure probability of a single device is ≥80% for three consecutive 15-minute cycles, a device pre-failure warning situation is generated, and when the energy consumption of the equipment cluster during non-working hours is greater than 150% of the benchmark value, an "abnormal energy consumption warning during non-working hours" is generated; security parameter data analysis: obtain access control, fire smoke detection, and infrared radiation equipment signals in real time, and generate a "fire incident confirmation situation" when the smoke sensor is triggered and the video detects smoke pixels ≥5% or flame pixels ≥3%. When the access control is abnormally unlocked and the infrared radiation is triggered, a violent intrusion event confirmation situation is generated, and when the access control is invalidly unlocked for three consecutive times, a "malicious unlocking warning situation" is generated.

[0013] Furthermore, traffic data analysis: vehicle flow, speed, and parking space occupancy rate are obtained through license plate recognition and ground sensor coils, and a road network model is constructed to output the forecast value for the next hour; when the speed on the main road is ≤10km / h and lasts for more than 15 minutes, the parking lot occupancy rate is ≥90% and there are ≥50 reserved vehicles, and the traffic flow during non-peak hours (10:00-16:00) is >120% of the daily average, a "road congestion warning situation", "parking space shortage warning situation", and "non-peak traffic abnormality warning situation" are generated respectively.

[0014] Furthermore, the knowledge graph construction module constructs a dynamic knowledge graph according to the following steps: entity recognition: extracting personnel ID, equipment ID, area coordinates, and event type from the diagnosis prediction structured results; attribute definition: personnel nodes contain work ID and authorized area, equipment nodes contain numbers and maintenance responsible persons, environment nodes contain coordinates and prohibited area coordinates, and event nodes contain types and occurrence time; relationship construction: establishing "operation authority", "location association", and "trigger condition" relationships; dynamic update: storing node relationships in real time through the graph database, updating attributes or adding new relationships when new results are input, and outputting the complete graph to the decision generation module.

[0015] Furthermore, the decision generation module generates instructions and plans in combination with the dynamic knowledge graph: Equipment control instructions: When the equipment is in a pre-failure state, the maintenance responsible person number and spare parts inventory number are obtained according to the knowledge graph, and a maintenance instruction containing the equipment number, downtime window (22:00-6:00 during non-working hours), and maintenance personnel number is generated; When the energy consumption is abnormal during non-working hours, a power-off instruction for non-critical equipment or a power adjustment instruction for critical equipment is generated (adjustment parameter = current power × 80%); Security warning signal: When abnormal access is made, a signal containing the warning level, unauthorized area, and processing suggestions is generated in combination with the historical record of personnel unauthorized access; When a violent intrusion event is confirmed, a first-level security warning signal is generated containing the intrusion time, access control device number, and surveillance video link; Resource scheduling plan: When parking spaces are tight, a parking guidance strategy containing the parking lot number and diversion route (the three shortest routes) is generated based on the real-time vacancy data of surrounding parking lots; The generated equipment control instructions are sent to the execution feedback module, and the security warning signal and resource scheduling plan are sent to the visualization interaction module along with the structured results.

[0016] Furthermore, the visualization interaction module supports multi-format presentation and multi-terminal interaction: Multi-format presentation: Use line graphs to display the time series changes of equipment room temperature and CO2 concentration, use thermal maps to mark the coordinates and density values of areas where people gather, call up warning-related monitoring screens in real time and overlay event details, generate reports summarizing warning historical data and resource scheduling execution effects; Multi-terminal interaction: Support PC-side screening of warning types, mobile-side real-time warning push, and large-screen terminals to dynamically display the security situation of the park, allowing users to manually trigger temporary instructions through the interactive interface, which will be executed after verification by the decision-making generation module.

[0017] Furthermore, the execution feedback module completes closed-loop management according to the following process: instruction parsing and execution: parsing the device number, instruction type and parameters in the device control instruction, and regulating the device through Modbus / TCP protocol signals; status monitoring: collecting the voltage, current, vibration frequency, energy consumption, and cumulative operating time parameters of the regulated device, generating standardized parameters in the 0-1 interval according to the normalization rules of the data preprocessing module, and comparing the standardized parameter range of the normal operation of the device to determine the status; feedback output: encapsulating the device status data and the equipment room temperature and CO2 concentration change values, and after format unification and noise filtering by the data preprocessing module, input them into the intelligent analysis module for model training or early warning rule optimization to form a closed-loop iteration.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] 1. The patent of this invention realizes the intelligent monitoring and management of smart parks by constructing a closed-loop management chain of "acquisition-processing-analysis-decision-execution". The data acquisition module obtains multi-dimensional basic data in real time, generates standardized input data through preprocessing, and provides a high-quality data source for intelligent analysis; the intelligent analysis module uses algorithms such as convolutional neural networks and long short-term memory networks to achieve in-depth analysis of data such as environmental status, personnel behavior, and equipment operation, and outputs accurate diagnostic and predictive structured results; the knowledge graph construction module constructs a dynamic knowledge graph containing the "personnel-equipment-environment-event" association relationship based on the structured results, providing multi-dimensional association information for decision-making; the decision-making module combines the knowledge graph to generate equipment control instructions, safety warning signals and resource scheduling plans, realizing precise control of park equipment and rational allocation of resources; the visual interaction module supports multi-format presentation and multi-terminal interaction, which improves the intuitiveness and convenience of monitoring and management; the execution feedback module feeds back the equipment status data to the data preprocessing module through closed-loop management, forming iterative optimization to ensure the continuous and efficient operation of the system.

[0020] 2. The advantages of the present invention lie in the integration of multiple technologies and full-process closed-loop management. The LSTM network is used to perform time-series analysis and prediction of the environmental parameters of the equipment room, and the random forest algorithm is used to calculate the probability of equipment failure, which can detect potential risks in advance and generate early warnings; the graph database is used to build a dynamic knowledge graph, which realizes the real-time update and deep mining of multi-entity relationships, providing comprehensive information support for decision-making; in equipment control and resource scheduling, the maintenance person in charge, spare parts inventory and surrounding parking space data are accurately located based on the knowledge graph, which improves the intelligent level of management; the execution feedback module realizes real-time monitoring and data feedback of equipment status through the Modbus / TCP protocol, ensuring the reliability and self-optimization capability of the system. Overall, the system significantly improves the security protection capability, equipment operation and maintenance efficiency, and resource utilization efficiency of the smart park, and has the technical advantages of comprehensiveness, accuracy and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings;

[0022] Figure 1 This is a system block diagram of the present invention.

[0023] Reference numeral: None DETAILED DESCRIPTION

[0024] 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.

[0025] Example 1: Figure 1 As shown, an AI-based smart park monitoring and management system includes a data acquisition module, a data preprocessing module, an intelligent analysis module, a knowledge graph construction module, a decision generation module, a visual interaction module, and an execution feedback module that are directly or indirectly connected to each other, forming a closed-loop management chain of "acquisition-processing-analysis-decision-execution";

[0026] The data acquisition module is composed of sensors, cameras, IoT devices and identification devices distributed in different locations in the park. It collects multi-dimensional basic data in the park in real time and transmits it to the data pre-processing module;

[0027] The data preprocessing module unifies the format, filters noise, and extracts features from multi-dimensional basic data to generate standardized input data;

[0028] The intelligent analysis module outputs structured diagnostic and predictive results based on standardized input data through machine learning and deep learning algorithms;

[0029] The knowledge graph construction module constructs a dynamic knowledge graph containing the "personnel-equipment-environment-event" association relationship based on the structured results of the diagnosis prediction;

[0030] The decision generation module combines dynamic knowledge graphs to generate equipment control instructions, safety warning signals, and resource scheduling plans;

[0031] The visual interaction module presents structured results and decision-making content in the form of charts, monitoring screens and reports, and supports multi-terminal interactive operations;

[0032] The execution feedback module receives the equipment control instructions from the decision generation module, automatically adjusts the park equipment, and feeds back the adjusted status data to the data preprocessing module.

[0033] The multi-dimensional basic data includes environmental status data, personnel behavior data, equipment operation data, safety parameter data and traffic data. The data preprocessing module receives the multi-dimensional basic data from the data acquisition module, cleans it, unifies the format, detects outliers, and calibrates it in time and space to generate a standardized data set, and sends it to the intelligent analysis module.

[0034] The structured diagnostic and predictive results include warnings for combustible gas leaks in equipment rooms, abnormal access to authorized resources, equipment failure predictions, confirmation of violent intrusions, malicious unlocking, parking shortages, and abnormal traffic flow during off-peak hours.

[0035] The intelligent analysis module receives standardized data sets and performs real-time and historical analysis of the data based on convolutional neural networks and long short-term memory networks.

[0036] The intelligent analysis module analyzes environmental status data as follows: It collects real-time temperature values from temperature sensors deployed in the equipment room, as well as CO2 concentration and combustible gas concentration values from gas sensors. It then performs a sliding average of these values over a 5-minute window to generate a time series of four parameters.

[0037] The long short-term memory network (LSTM) is used to model the temperature time series, CO2 concentration time series, and combustible gas concentration time series, and output the temperature forecast value, CO2 concentration forecast value, and combustible gas concentration forecast value for the next 24 hours;

[0038] When the real-time temperature value is greater than 35°C or the predicted temperature value is greater than 35°C, a structured result of "Equipment Room High Temperature Warning" is generated, including the warning time and the corresponding equipment room number;

[0039] When the real-time CO2 concentration is greater than 1000ppm or the predicted CO2 concentration is greater than 1000ppm, a structured "Equipment Room Air Quality Deterioration Warning" result is generated, including the warning time and the area of the equipment room where ventilation equipment needs to be activated.

[0040] When the real-time combustible gas concentration is greater than 20% of the lower explosion limit or the predicted combustible gas concentration is greater than 20% of the lower explosion limit, a combustible gas leakage warning situation is generated in the equipment room, including the warning time, leakage gas type, and the location of related fire-fighting equipment.

[0041] The intelligent analysis module analyzes the personnel behavior data as follows: by parsing the video stream of the camera at the park gate frame by frame, it extracts the personnel location coordinates, movement speed value, and gathering density value (number of personnel per unit area);

[0042] Use the face recognition camera to collect facial feature data of personnel, perform a 1:1 comparison with the facial feature data of authorized personnel in the park personnel authority database, and output the personnel identity matching result (authorized personnel / unauthorized personnel);

[0043] When the coordinates of a person's location fall within the coordinate set of the restricted area defined by the electronic fence, a "restricted area intrusion warning" structured result is generated, including the intrusion time, the person's real-time image, and the restricted area number. When the gathering density value is ≥3 people / ㎡ and the duration is ≥10 minutes, a "personnel gathering risk warning" structured result is generated, including the gathering start time, gathering area coordinates, and the recommended number of security personnel to be dispatched.

[0044] When the person's identity matching result is "unauthorized personnel" and their moving speed value is greater than 5m / s, an "abnormal running warning" structured result is generated, including a video clip of the person's trajectory and the number of the nearest access control device;

[0045] When an authorized person enters an area outside of his or her authority (an ordinary employee enters the computer room area), an abnormal access warning is generated, including the person's employee number, the name of the unauthorized area, and the time of access.

[0046] The intelligent analysis module analyzes the device operation data as follows: the device voltage value, device current value, vibration frequency value, energy consumption value, and cumulative operating time collected by the IoT device are normalized using the formula: Where X is the original equipment operating parameter, Xmin is the historical minimum value of the parameter, and Xmax is the historical maximum value of the parameter), generating standardized parameters in the range of 0-1. A random forest algorithm is used to build an equipment fault classification model. The model inputs historical equipment voltage values, current values, vibration frequency values, cumulative operating time, and corresponding fault types (bearing wear / circuit overheating / component looseness) for training, and outputs the failure probability value of the current equipment. The equipment energy consumption values are clustered according to the time dimension (working days / non-working days, working hours / non-working hours), and the average energy consumption during the non-working hours (22:00-6:00) is set as the baseline value.

[0047] When the failure probability of a single device is ≥80% for three consecutive 15-minute periods, a pre-failure warning status for the device is generated, including the fault type, recommended repair time, and related spare parts inventory number;

[0048] When the energy consumption value of a device cluster in a certain area during non-working hours is greater than 150% of the baseline value, a structured result of "abnormal energy consumption warning during non-working hours" is generated, including the abnormal area number and a list of devices with excessive energy consumption (sorted by voltage and current values).

[0049] The intelligent analysis module analyzes security parameter data as follows: real-time acquisition of the switch status of the access control sensor (unlocked / locked), the alarm signal of the fire smoke detector (triggered / untriggered), and the trigger signal of the infrared radiation device (triggered / untriggered);

[0050] When the fire smoke detector alarm signal is "triggered" and the proportion of smoke pixels detected in the corresponding area camera video stream is ≥5% or the proportion of flame pixels detected is ≥3%, a "fire incident confirmation" structured result is generated, including the alarm time, the three-dimensional coordinates of the fire area, and the nearest fire hydrant number;

[0051] When the access control sensor switch state is "unlocked" and the unlocking time is not within the user's authorized period (0:00-6:00 in the morning), and the infrared radiation device trigger signal is "triggered", a violent intrusion event confirmation situation is generated, including the intrusion time, access control device number, and the location sequence of the triggering infrared radiation device;

[0052] When the access control sensor receives three invalid unlocking signals in a row (wrong password / fingerprint mismatch), a malicious unlocking warning situation is generated, which includes the unlocking time, the facial image of the person attempting to unlock, and the link to the associated surveillance video clip.

[0053] The intelligent analysis module analyzes traffic data as follows:

[0054] The license plate recognition camera captures vehicle flow (number of vehicles passing per minute) and speed (km / h), and the ground sensor coil captures parking space occupancy (number of occupied parking spaces / total number of parking spaces). This model then constructs a campus road network model, inputting historical vehicle flow, speed, and parking space occupancy to output predicted speeds for each road section and the number of remaining parking spaces for the next hour.

[0055] When the vehicle speed on a certain section of the main road is ≤10km / h and the duration is ≥15 minutes, a structured "road congestion warning" result is generated, including the name of the congested section, the start time of the congestion, and a recommended detour route (three alternative routes are generated based on the principle of shortest distance);

[0056] When the parking lot occupancy rate is ≥90% and the number of vehicles reserved in the next hour is ≥50, a parking shortage warning is generated, including the parking lot number, the current number of remaining parking spaces, and the parking guidance strategy to be activated (diverting traffic to surrounding parking lots);

[0057] When the vehicle flow value during the off-peak period (10:00-16:00) is greater than 120% of the daily average, an off-peak period traffic abnormality warning situation is generated, including the abnormal period and the entrance / exit number where the vehicle flow exceeds the limit.

[0058] Example 2: The knowledge graph construction module receives the equipment room combustible gas leakage warning situation, abnormal access warning situation, equipment pre-failure warning situation, violent intrusion event confirmation situation, malicious unlocking warning situation, parking space shortage warning situation, and off-peak traffic abnormality warning situation output by the intelligent analysis module, and constructs a dynamic knowledge graph containing the "personnel-equipment-environment-event" association relationship according to the following steps:

[0059] Entity recognition: Extract employee ID, equipment ID, area coordinates, and event type from each diagnostic prediction structured result;

[0060] Attribute definition: assign work number and authorization area attributes to personnel nodes, assign equipment number and maintenance responsible person attributes to equipment nodes, assign area coordinates and prohibited area coordinates attributes to environment nodes, and assign event type and occurrence time attributes to event nodes;

[0061] Relationship building: Establish the "operation permission" relationship between personnel nodes and equipment nodes, the "location association" relationship between equipment nodes and environment nodes, and the "trigger condition" relationship between event nodes and personnel / equipment nodes;

[0062] Dynamic update and output: The above nodes and relationships are stored in real time through the graph database. When new diagnostic prediction structured results are received, the corresponding node attributes are updated or new association relationships are added, and the complete dynamic knowledge graph is output to the decision generation module.

[0063] The decision generation module receives the dynamic knowledge graph from the knowledge graph construction module and generates equipment control instructions, safety warning signals, and resource scheduling plans according to the following logic:

[0064] Device control command generation:

[0065] For equipment pre-failure warning situations (failure probability ≥ 80%), the equipment maintenance responsible person number and spare parts inventory number are obtained from the knowledge graph, and a shutdown maintenance instruction is generated containing the equipment number, downtime window (22:00-6:00 during non-working hours), and maintenance personnel number;

[0066] For abnormal energy consumption warnings during non-working hours (energy consumption value > 150% of the baseline value), the abnormal device type is located from the knowledge graph, and power-off instructions for non-critical devices or power adjustment instructions for critical devices are generated (adjustment parameter = current power × 80%);

[0067] Safety early warning signal generation:

[0068] For abnormal access warning trends, combined with the historical records of the person exceeding their authority in the knowledge graph, a security warning signal is generated that includes the warning level, the name of the area exceeding authority, and handling suggestions.

[0069] To confirm the situation of violent intrusion incidents, the access control device number and the triggered infrared radiation device location sequence are obtained from the knowledge graph to generate a first-level security warning signal containing the intrusion time and the associated surveillance video link; resource scheduling plan generation:

[0070] In response to parking shortage warnings (parking space occupancy rate ≥ 90% and number of reservations ≥ 50 vehicles), the system obtains real-time vacancy data from surrounding parking lots from the knowledge graph and generates a parking guidance strategy that includes parking lot numbers and diversion routes (the three shortest routes).

[0071] The generated equipment control instructions are sent to the execution feedback module, and the safety warning signals and resource scheduling plans are sent to the visualization interaction module along with the structured results.

[0072] The visualization interaction module receives the diagnostic and prediction structured results from the intelligent analysis module and the safety warning signals and resource scheduling plans from the decision-making module, and presents and interacts with them in the following ways:

[0073] Multiple forms of presentation:

[0074] Use line graphs to display the time-series changes in equipment room temperature and CO2 concentration, and use thermal maps to mark the coordinates and density values of areas where people gather. Access monitoring images associated with warnings in real time, overlaying event details. Generate reports summarizing historical data on equipment pre-failure warnings and the effectiveness of resource scheduling plans.

[0075] Multi-terminal interaction:

[0076] It supports filtering specific types of warnings on PC, receiving real-time warning push on mobile devices, and dynamically displaying the overall security situation of the park on large-screen terminals; it allows users to manually trigger temporary instructions through the interactive interface, and the instructions are executed after verification by the decision-making generation module.

[0077] The execution feedback module receives the device control instructions from the decision generation module and completes the closed-loop management according to the following process:

[0078] Instruction parsing and execution: parse the device number, instruction type and parameters in the device control instruction, and send Modbus / TCP protocol control signals to the device;

[0079] Condition monitoring: Collect the operating parameters of the regulated equipment, including equipment voltage, equipment current, vibration frequency, energy consumption, and accumulated operating time. Based on the normalization method of the data preprocessing module for equipment operation data, the collected operating parameters are normalized to generate standardized parameters in the range of 0-1. These parameters are compared with the corresponding standardized parameter range when the equipment is operating normally to determine whether the equipment is operating normally.

[0080] Feedback output: Equipment status data (normal or abnormal) and environmental response data (equipment room temperature changes, CO2 concentration changes) are encapsulated as feedback data sets. After format unification, noise filtering, and feature extraction by the data preprocessing module, they are input into the intelligent analysis module for model training or early warning rule optimization, forming a closed-loop iteration of "collection-processing-analysis-decision-making-execution."

[0081] The above content is merely an example and explanation of the structure of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the claims, they should all fall within the scope of protection of the present invention.

[0082] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0083] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to specific embodiments. Obviously, many modifications and variations are possible based on the contents of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. The intelligent park monitoring and management system based on artificial intelligence is characterized by: It includes data acquisition modules, data preprocessing modules, intelligent analysis modules, knowledge graph construction modules, decision-making modules, visual interaction modules and execution feedback modules that are directly or indirectly connected to each other, forming a closed-loop management chain of acquisition, processing, analysis, decision-making and execution; The data acquisition module is composed of sensors, cameras, IoT devices and identification devices distributed in different locations in the park. It collects multi-dimensional basic data in real time within the park and transmits it to the data preprocessing module; the data preprocessing module unifies the format, filters noise and extracts features of the multi-dimensional basic data to generate standardized input data; the intelligent analysis module outputs diagnostic and predictive structured results based on the standardized input data through machine learning and deep learning algorithms; the knowledge graph construction module constructs a dynamic knowledge graph containing the relationship between personnel, equipment, environment and events based on the diagnostic and predictive structured results; the decision generation module generates equipment control instructions, safety warning signals and resource scheduling plans based on the dynamic knowledge graph; The visual interaction module presents the structured results and decision content in the form of charts, monitoring screens and reports, and supports multi-terminal interactive operations; The execution feedback module receives the equipment control instructions from the decision generation module, automatically adjusts the park equipment, and feeds back the adjusted state data to the data preprocessing module.

2. The intelligent park monitoring and management system based on artificial intelligence according to claim 1 is characterized in that: The multi-dimensional basic data includes environmental status data, personnel behavior data, equipment operation data, safety parameter data and traffic data; the standardized data set generated by the data preprocessing module contains multi-dimensional basic data after cleaning, format unification, outlier detection and time-space calibration.

3. The intelligent park monitoring and management system based on artificial intelligence according to claim 1 is characterized in that: The intelligent analysis module performs real-time and historical analysis on the standardized data set, specifically including: environmental status data analysis: real-time collection of equipment room temperature, CO2 concentration, and combustible gas concentration values, generating a time series by sliding average of a 5-minute time window, and outputting the predicted value for the next 24 hours through long-term and short-term memory network modeling; when the real-time temperature value is greater than 35°C or the predicted value is greater than 35°C, a high temperature warning for the equipment room is generated; when the real-time CO2 concentration value is greater than 1000ppm or the predicted value is greater than 1000ppm, a warning of decreased air quality in the equipment room is generated; when the real-time combustible gas concentration value is greater than explosion level, a warning of decreased air quality in the equipment room is generated. When the lower explosion limit is 20% or the predicted value is greater than 20% of the lower explosion limit, a flammable gas leakage warning situation is generated in the equipment room; personnel behavior data analysis: parse the video stream of the park gate camera, extract the personnel position, speed, and gathering density values, and output the identity matching result through face recognition and comparison authority library; when personnel enter the restricted area, the gathering density is ≥3 people / ㎡ and lasts for more than 10 minutes, the moving speed of unauthorized personnel is greater than 5m / s, and the authorized personnel enter the area without permission, a restricted area intrusion warning situation, a personnel gathering risk warning situation, an abnormal running warning situation, and an abnormal authority access warning situation are generated respectively.

4. The intelligent park monitoring and management system based on artificial intelligence according to claim 3 is characterized in that: Equipment operation data analysis: Normalize the equipment voltage, current, vibration frequency, energy consumption, and cumulative operating time, train the fault classification model through the random forest algorithm to output the fault probability value, and set the average energy consumption during non-working hours as the baseline value; when the failure probability of a single device is ≥80% for three consecutive 15-minute cycles, a device pre-failure warning situation is generated, and when the energy consumption of the equipment cluster during non-working hours is greater than 150% of the baseline value, a non-working energy consumption abnormality warning is generated; security parameter data analysis: Real-time acquisition of access control, fire smoke detection, and infrared radiation equipment signals. When the smoke sensor is triggered and the video detects smoke pixels ≥5% or flame pixels ≥3%, a fire event confirmation situation is generated. When the access control is abnormally unlocked and the infrared radiation is triggered, a violent intrusion event confirmation situation is generated. When the access control is invalidly unlocked for three consecutive times, a malicious unlocking warning situation is generated.

5. The intelligent park monitoring and management system based on artificial intelligence according to claim 3 is characterized in that: Traffic data analysis: Vehicle flow, speed, and parking space occupancy are obtained through license plate recognition and ground sensor coils, and a road network model is constructed to output forecast values for the next hour. When the speed on the main road is ≤10km / h and lasts for more than 15 minutes, the parking lot occupancy rate is ≥90% and there are ≥50 reserved vehicles, and the traffic flow during non-peak hours is >120% of the daily average, a road congestion warning situation, a parking space shortage warning situation, and a non-peak hour traffic abnormality warning situation are generated respectively.

6. The intelligent park monitoring and management system based on artificial intelligence according to claim 1 is characterized in that: The knowledge graph construction module constructs a dynamic knowledge graph through entity recognition, attribute definition, relationship construction, and dynamic update, and outputs the complete graph to the decision generation module.

7. The artificial intelligence-based smart park monitoring and management system according to claim 6 is characterized in that: The decision generation module generates instructions and solutions in combination with the dynamic knowledge graph: Equipment control instructions: When equipment is in the process of failure, the maintenance responsible person number and spare parts inventory number are obtained according to the knowledge graph, and a maintenance instruction containing the equipment number, downtime window, and maintenance personnel number is generated; when energy consumption is abnormal during non-working hours, a power-off instruction for non-critical equipment or a power adjustment instruction for critical equipment is generated; Security warning signal: When abnormal access is made, a signal containing the warning level, the unauthorized area, and handling suggestions is generated in combination with the historical record of personnel unauthorized access; When a violent intrusion event is confirmed, a first-level security warning signal is generated containing the intrusion time, access control device number, and surveillance video link; Resource scheduling plan: When parking spaces are tight, a parking guidance strategy including parking lot numbers and diversion routes is generated based on real-time vacancy data of surrounding parking lots; The generated equipment control instructions are sent to the execution feedback module, and the safety warning signals and resource scheduling plans are sent to the visualization interaction module along with the structured results.

8. The artificial intelligence-based smart park monitoring and management system according to claim 7 is characterized in that: The visualization interaction module supports multi-format presentation and multi-terminal interaction: Multi-format presentation: Use line graphs to display the time series changes of equipment room temperature and CO2 concentration, use thermal maps to mark the coordinates and density values of areas where people gather, call up warning-related monitoring screens in real time and overlay event details, generate reports summarizing warning historical data and resource scheduling execution effects; Multi-terminal interaction: Support PC-side screening of warning types, mobile-side real-time warning push, and large-screen terminals to dynamically display the security situation of the park, allowing users to manually trigger temporary instructions through the interactive interface, which will be executed after verification by the decision-making generation module.

9. The artificial intelligence-based smart park monitoring and management system according to claim 8 is characterized in that: The execution feedback module completes closed-loop management according to instruction parsing and execution, status monitoring, and feedback output.

Images