Intelligent pedestrian tile safety event real-time response system

Abstract

The present application relates to the technical field of intelligent city infrastructure safety monitoring, and discloses a real-time response system for safety events of intelligent pedestrian tiles. The system comprises a real-time state monitoring module, a dynamic event detection module, a multi-dimensional event difference analysis module, a safety area correlation positioning module and a self-adaptive response generation module. The sensor array collects tile pressure distribution, surface humidity and temperature fluctuation data, and combines historical event data to construct a dynamic prediction model to output potential safety event probability values. A multi-dimensional difference analysis generates an event-level difference feature matrix, and a spatial topology network and environmental parameters are used to generate a safety event heat distribution map and locate high-risk areas. Finally, response parameters are adaptively configured. The system realizes real-time sensing, accurate positioning and dynamic response of safety events in urban pedestrian areas.

Description

Technical Field

[0001] This invention relates to the field of smart city infrastructure safety monitoring technology, specifically a real-time response system for safety incidents involving smart pedestrian pavers. Background Technology

[0002] The safety management of urban public spaces increasingly relies on intelligent technologies, especially in high-frequency areas such as sidewalks, where changes in their physical condition directly impact public safety. Traditional methods for monitoring the safety of paving stones are often limited to detecting single physical parameters, such as using pressure sensors to determine if a stone is damaged or using humidity sensors to detect surface slippage. These methods often lack the ability to collaboratively analyze multi-dimensional data, making it difficult to address the dynamic and sudden nature of safety incidents in complex environments.

[0003] In existing technologies, some systems employ fixed threshold-triggered alarm mechanisms, such as issuing an alarm when the pressure on the floor tiles exceeds a preset value. However, such static models cannot adapt to the effects of changes in time, climate, and pedestrian density, easily leading to false alarms or missed alarms. While some prediction models based on historical data can improve detection sensitivity to some extent, most do not incorporate real-time environmental parameters (such as material expansion due to temperature fluctuations, instantaneous humidity changes, etc.), resulting in discrepancies between prediction results and actual risks.

[0004] Furthermore, existing security response mechanisms are typically limited to reactive handling after an incident occurs, lacking analysis of the spatial correlation of risk areas. For example, when a pavement is partially damaged, the system may only mark that location, ignoring the cascading risks caused by stress transfer to surrounding pavements. This fragmented approach makes it difficult to form a regional-level security situation assessment, let alone achieve dynamic adjustments to response strategies.

[0005] The variability of environmental factors further increases the complexity of safety incident monitoring. For example, the interaction between humidity and temperature on the surface of paving stones during the rainy season may lead to a decrease in compressive strength, but existing systems rarely integrate multi-source environmental data for analysis, resulting in a lack of comprehensiveness in risk assessment. Simultaneously, most systems lack a network model based on spatial topology, making it impossible to correlate single-point events with regional physical layout (such as paving stone density and base structure), limiting the accurate location of high-risk areas. Therefore, a response system is needed that can integrate multi-dimensional real-time data and historical event characteristics, and combine environmental factors with spatial topological relationships for dynamic risk assessment, to improve the accuracy and timeliness of urban pedestrian paving stone safety management. Summary of the Invention

[0006] The purpose of this invention is to provide a real-time response system for safety incidents on intelligent pedestrian paving stones, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides a real-time response system for safety incidents on intelligent pedestrian paving stones, the system comprising: The real-time status monitoring module, based on a sensor array deployed on the pedestrian paving stones, collects real-time data on paving stone pressure distribution, surface humidity, and temperature fluctuation. The dynamic event detection module constructs a dynamic event prediction model based on historical security event data, and outputs the probability value of potential security events through the dynamic event prediction model. The multidimensional event difference analysis module performs multidimensional difference analysis on the probability values ​​of potential security events and real-time monitoring data to generate an event-level difference feature matrix. The safe area association and positioning module inputs the difference feature matrix into the spatial topology network, combines the floor tile layout parameters and environmental influencing factors to generate a heat map of safety events, and locates high-risk physical areas; The adaptive response generation module configures response parameters based on the heat map of the security events.

[0008] Preferably, the dynamic event detection module specifically includes: The historical data feature extraction unit performs multi-dimensional decomposition processing on historical safety event data, extracts the energy ratio of steady-state and transient components of pressure waveforms, establishes a correlation mapping between humidity fluctuations and event occurrence frequency, and aligns temperature fluctuation sequence patterns under different scenarios. The dynamic event prediction model building unit inputs the processed historical security event data into the prediction network to generate basic event prediction values, corrects prediction biases caused by environmental interference, and dynamically adjusts prediction weights based on real-time collected pressure distribution data. The event probability value calculation unit inputs real-time collected data into the dynamic event prediction model to obtain the probability value of potential security events.

[0009] Preferably, the multidimensional event difference analysis module specifically includes: The time-domain cumulative difference calculation unit performs a sliding comparison between the probability value of potential security events and real-time monitoring data within a preset time window, calculates the cumulative difference within each window, and generates a time-domain difference vector. The spatial distribution offset detection unit decomposes the spatial distribution of the probability value of potential safety events and real-time monitoring data into regions, extracts the distribution offset index of each region, and constructs a spatial offset vector. The event sequence similarity evaluation unit matches the distance distribution between the probability values ​​of potential security events and the real-time event sequence, quantifies the time synchronization error of the sequence mutation point, and generates an event similarity vector. The difference feature matrix generation unit performs multi-dimensional concatenation of the temporal difference vector, spatial offset vector, and event similarity vector, and outputs an event-level difference feature matrix through feature normalization processing.

[0010] Preferably, the safe area association and positioning module specifically includes: The floor tile topology modeling unit constructs a spatial node connection topology diagram based on the floor tile layout parameters, labels the distance parameters between each node, and overlays environmental influencing factors to generate a topology model including a distance matrix and a regional weight matrix. The event propagation simulation unit maps the difference feature matrix to the corresponding nodes of the topology model, performs event propagation deduction, calculates the attenuation factor of abnormal events in nodes, captures cross-regional event correlation features, and simulates the diffusion path of abnormal events in the topology network. The heat map generation unit counts the frequency of events occurring along each path in the simulated propagation, calculates the event dwell probability value by combining distance parameters, and generates a heat map of safety events. The physical region positioning unit performs spatial clustering analysis on the heat map of the security events to identify areas where the event probability is concentrated and delineates high-risk physical boundaries.

[0011] Preferably, the adaptive response generation module specifically includes: The response parameter configuration unit enables a high-frequency monitoring mode based on the probability value region of the security event heat map. The disturbance test execution unit applies multi-band physical disturbance tests to the adjacent node regions with the largest probability gradient changes in the heat map of safety events.

[0012] Preferably, the system further includes an environmental factor reconstruction module: The environmental data acquisition unit obtains the location coordinates and timestamps of pedestrian paving stones, collects environmental humidity data and surface temperature data at the corresponding locations, associates the coordinate locations, calculates environmental suitability values, and generates an environmental suitability set.

[0013] Preferably, the system further includes a risk zoning labeling module: The spatial relationship sorting unit extracts the environmental suitability values ​​and coordinate data from the environmental suitability set, sorts adjacent point pairs according to spatial distance, and marks consistent and conflicting segments. The partition labeling generation unit generates a directionally consistent partition labeling set based on the difference rate of environmental suitability values ​​between adjacent point pairs.

[0014] Preferably, the system further includes a rate of change mapping module: The environmental fluctuation extraction unit filters directionally consistent segments from the directionally consistent partition label set and extracts the temperature time series and humidity time series. The impact intensity assessment unit calculates the rate of temperature rise and the rate of humidity fall, compares the magnitude of changes with the frequency of event occurrence, and generates the results of the event impact intensity analysis.

[0015] Preferably, the system further includes an event disturbance identification module: The disturbance point filtering unit identifies points in the event impact intensity analysis results whose impact intensity values ​​are greater than the average reference value, as well as points in directional conflict sections; The anomaly point generation unit, based on the dosage change ratio and area change ratio of the application point, combined with wind direction fluctuation data and humidity fluctuation data, marks disturbance points and generates a set of safety event disturbance anomalies.

[0016] Preferably, the system further includes a risk analysis output module: The joint judgment unit acquires the location information of the set of abnormal points of the security event disturbance and calculates the joint risk judgment value; The early warning result generation unit, based on the joint risk judgment value sequence, filters high-risk locations, marks the risk of spread, and generates safety incident detection and risk early warning results.

[0017] Compared with the prior art, the beneficial effects of the present invention are: This invention achieves real-time perception and precise response to safety incidents involving urban pedestrian paving stones through multi-sensor data fusion and dynamic modeling. The system integrates multi-dimensional data such as pressure, humidity, and temperature, overcoming the limitations of traditional single-parameter detection and significantly improving the comprehensiveness and reliability of safety incident identification in complex environments. The dynamic event prediction model built based on historical data enhances the system's ability to anticipate potential risks, avoiding false alarms or delayed responses caused by fixed threshold mechanisms.

[0018] The multidimensional event difference analysis module performs in-depth comparison between real-time monitoring data and predicted probabilities to generate an event-level difference feature matrix, effectively capturing subtle deviations between abnormal patterns and normal operation, and enhancing the system's sensitivity to emergencies. By introducing spatial topology networks and environmental influencing factors, the system can map discrete events into continuous regional heat map distributions, revealing the clustering patterns and diffusion trends of high-risk areas, and supporting the shift from single-point response to regional collaborative prevention and control.

[0019] The safe zone correlation and positioning module combines floor tile layout parameters with real-time environmental data to achieve precise correlation between physical space and risk situation, avoiding assessment bias caused by neglecting structural correlation in traditional methods. The adaptive response generation module dynamically configures response parameters based on heat map, realizing closed-loop management from early warning to response, and improving the system's decision-making flexibility and execution efficiency in different scenarios.

[0020] The entire system, through data-driven and model-based collaboration, forms a complete chain from perception, analysis, location to response. This not only strengthens the ability to respond to known risks but also improves the level of discovery and intervention for unknown hidden dangers, providing continuously optimized technical support for the safety management of urban public spaces. Attached Figure Description

[0021] Figure 1 This is a schematic diagram illustrating the working principle of the intelligent pedestrian pavement safety event real-time response system described in this invention. Figure 2 This is a schematic diagram illustrating the working principle of the multidimensional event difference analysis module. Figure 3 Schematic diagram of the working principle of the safe area association and positioning module; Figure 4 This is a diagram illustrating the working principle of the risk zoning and labeling module. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figure 1 This invention provides a real-time response system for safety incidents on intelligent pedestrian paving stones, the system comprising: By integrating a sensor array and a data analysis model, the system enables real-time monitoring, prediction, and response to safety incidents in pedestrian paving areas. The system includes a real-time status monitoring module, a dynamic event detection module, a multi-dimensional event difference analysis module, a safe area correlation and location module, and an adaptive response generation module. The real-time status monitoring module, based on a sensor array deployed on the pedestrian paving stones, collects real-time data on paving stone pressure distribution, surface humidity, and temperature fluctuations. The dynamic event detection module constructs a dynamic event prediction model based on historical safety event data, outputting the probability value of potential safety events. The multi-dimensional event difference analysis module performs multi-dimensional difference analysis between the potential safety event probability value and the real-time monitoring data, generating an event-level difference feature matrix. The safe area correlation and location module inputs the difference feature matrix into a spatial topology network, combining paving stone layout parameters and environmental influencing factors to generate a safety event heat map, locating high-risk physical areas. The adaptive response generation module configures response parameters based on the safety event heat map.

[0024] Example 1: See Figure 2The implementation of the dynamic event detection module involves the collaborative work of multiple processing units. Its core lies in extracting effective features from historical data and constructing a dynamically adjustable prediction model. The historical data feature extraction unit is responsible for multi-dimensional decomposition of the stored historical security event dataset, which contains time-series information on pressure waveforms, humidity fluctuations, and temperature fluctuations. Pressure waveform processing employs the discrete wavelet transform method, decomposing the continuous pressure signal into sub-signals of different frequency bands, and calculating the energy ratio of the steady-state and transient components respectively. The steady-state component corresponds to the low-frequency sub-band, and the transient component corresponds to the high-frequency sub-band. The formula for calculating the energy ratio is: ; in: This indicates the ratio of energy percentages. This represents the k-th coefficient of the steady-state component. This represents the k-th coefficient of the transient component. This represents the total number of coefficients. Humidity fluctuation data is processed using time series analysis methods to calculate the correlation coefficient between the variance within the sliding window and the number of historical events, establishing a correlation mapping table. Temperature fluctuation sequences are aligned using a dynamic time warping algorithm to eliminate sequence length differences caused by asynchronous acquisition times, forming a standardized temperature feature vector.

[0025] The dynamic event prediction model building unit employs a Long Short-Term Memory (LSTM) neural network as the core structure of the prediction network. Processed historical safety event data includes pressure energy features, humidity correlation features, and temperature alignment features, which are normalized and used as the network's input layer. The network contains three hidden layers with 128, 64, and 32 neurons per layer, respectively, using the ReLU activation function. The output layer generates a basic event prediction value, a continuous value between 0 and 1, representing the probability of the event occurring. An environmental interference correction module is integrated into the network structure, using a Kalman filter to estimate prediction bias caused by environmental factors in real time. The filter calculates state estimates and adjusts the prediction output based on real-time collected environmental temperature and humidity data. The dynamic weight adjustment mechanism is based on the variance changes of real-time pressure distribution data. When the pressure data variance exceeds a set threshold, a backpropagation algorithm is triggered to update the network weights, with the adjustment magnitude proportional to the variance change.

[0026] The event probability calculation unit executes a real-time data input and probability calculation process. The real-time collected pressure distribution data first undergoes the same feature extraction processing as historical data to calculate the steady-state and transient energy ratios at the current moment. Surface humidity data is converted into event frequency correlation values ​​by querying a pre-established association mapping table. Temperature fluctuation data is matched with standard patterns using a dynamic time warping algorithm to generate temperature feature vectors. These real-time feature vectors are input into a trained dynamic event prediction model, and after forward propagation calculation via a neural network, the potential safety event probability value is output. During the probability calculation process, a Kalman filter continuously operates, correcting the model output based on real-time environmental data to ensure the accuracy of the probability values.

[0027] The implementation of the multidimensional event difference analysis module involves the cascaded operation of four processing units. The time-domain cumulative difference calculation unit sets a 5-second time window and a 1-second sliding step size to perform a sliding window comparison between the probability values ​​of potential safety events and real-time monitoring data. Within each window, the mean square error between the probability value and the real-time monitoring data is calculated as the cumulative difference, calculated by dividing the sum of the squares of the errors of all data points within the window by the window length. These cumulative differences are arranged in chronological order to form a time-domain difference vector, with the vector dimension matching the number of window slides. The spatial distribution offset detection unit divides the monitoring area into several grid regions based on the tile layout, with each grid containing multiple sensor nodes. For each grid region, the ratio of the standard deviation of the spatial distribution of the probability values ​​of potential safety events to the distribution of the real-time monitoring data is calculated to obtain the distribution offset index for that region. The offset indices of all grid regions are arranged in spatial order to form a spatial offset vector, with the vector length equal to the number of grids.

[0028] The event sequence similarity assessment unit handles the matching problem of time series. Potential security event probability value sequences and real-time event sequences are aligned with their timestamps using interpolation methods, and the Euclidean distance between the two sequences at each time point is calculated. Sequence mutation points are identified using a sliding window variance detection algorithm, and the time synchronization error, i.e., the absolute value of the time difference between corresponding mutation points in the two sequences, is calculated for the identified mutation points. The similarity vector consists of normalized distance values ​​and error values, with normalization using a min-max scaling method. The difference feature matrix generation unit performs multi-dimensional data concatenation and standardization. The temporal difference vector, spatial offset vector, and event similarity vector are concatenated along the feature dimensions to form a high-dimensional feature vector. Each vector undergoes individual feature normalization before concatenation, ensuring its values ​​are uniformly between 0 and 1. The normalized vectors are then matrixed to form an event-level difference feature matrix, where the number of rows corresponds to the number of time windows, and the number of columns corresponds to the total number of feature dimensions. This matrix serves as the input data for further analysis and localization of security events.

[0029] Example 2: See Figure 3The implementation of the secure area correlation and positioning module involves precise modeling of the physical space and simulation analysis of event propagation. The paving tile topology modeling unit constructs a digital spatial node network based on the actual layout parameters of the laid pedestrian paving tiles. The center point of each standard-sized paving tile is set as a network node, and connecting edges are established between adjacent paving tile nodes according to their actual physical distance. The weight of the connecting edges is dynamically calculated based on the actual spacing and relative orientation of adjacent paving tiles. In the application scenario of a commercial plaza, the 600mm×600mm granite paving tiles are arranged in a strict grid pattern. The node spacing is precisely recorded into three types based on their relative positional relationships: 600mm for directly adjacent paving tiles, 848mm for diagonally adjacent paving tiles, and 1200mm for alternately adjacent paving tiles. The system assigns different connection weights to each spacing type, with directly adjacent tiles having the highest weight, followed by diagonally adjacent tiles, and alternately adjacent tiles having the lowest weight. Environmental influencing factors include differences in solar radiation intensity and pedestrian flow density in different areas of the plaza. These data are collected in real time through an environmental sensor network installed on the columns and ceiling. Solar radiation intensity data is updated every five minutes, and pedestrian flow density is collected once per second using an infrared thermal imaging sensor. After normalization, this environmental data is converted into distance attenuation coefficients and region weight coefficients, ultimately forming a spatial topology model that includes all node connectivity relationships and weighting parameters. This topology model employs a dual representation of adjacency matrices and weight matrices, simultaneously recording the absolute coordinates and relative spatial relationships of the nodes.

[0030] The event propagation simulation unit maps each feature vector in the difference feature matrix to the corresponding node in the topological model. In a shopping mall morning rush hour example, the system detected abnormal pressure characteristics on three consecutive paving stones at the northwest entrance. The simulation process uses a graph-based propagation algorithm, starting from the initial abnormal node and performing a breadth-first search along the topological connecting edges to calculate the propagation intensity of the event's impact. The calculation of propagation intensity comprehensively considers the weight coefficient of the connecting edges and the distance attenuation coefficient, where the distance attenuation coefficient is dynamically adjusted based on environmental data. During the simulation, the system identified that the event correlation between two paving stones spaced 1200mm apart was significantly weaker due to the larger distance, with the propagation intensity attenuation rate reaching about 60% of the standard value. Nodes located downwind, affected by airflow, exhibited stronger propagation characteristics despite the greater physical distance, with propagation intensity about 20% higher than expected. Through multiple rounds of iterative calculations, the system captured the trend of the abnormal event potentially spreading southeastward along the main flow direction of people. Each iteration updates the propagation intensity value of each node and records all possible propagation paths and their intensity parameters. This path information includes detailed data such as the direction of propagation, the sequence of nodes traversed, and the intensity value change curves of each node, providing a complete event propagation archive for analysis.

[0031] The heat map generation unit performs statistical analysis on the simulation results. In the deployment case of an office building lobby, the system counts the frequency of abnormal events occurring at nodes along each propagation path. Frequency statistics are based on the number of times each node is traversed by the propagation path during the simulation, combined with the actual physical distance between nodes to calculate the event dwell probability. Nodes closer together have lower propagation resistance due to tighter connections, resulting in a higher dwell probability; conversely, nodes farther apart experience greater propagation resistance, leading to a lower dwell probability. Nodes near ventilation openings exhibit additional probability increments due to environmental factors, such as accelerated airflow. The system converts the calculation results into a visualized heat map, using a 256-color gradient spectrum to represent different levels of event probability, transitioning from dark blue for low-probability areas to bright red for high-probability areas. In the generated distribution map, the central lounge area in the lobby, with its high population density, forms a clear probability cluster, appearing in a dark red hue. Meanwhile, the passageway areas near the edges, with faster population movement and shorter event dwell times, exhibit a low color temperature dominated by blue. The heat map is updated every 30 seconds to ensure that the current risk distribution is reflected in real time.

[0032] The physical area positioning unit performs automated analysis of the heat map. In a pedestrian walkway monitoring example in an urban park, the system uses a density-based clustering algorithm to identify high-probability event clusters. The algorithm first sorts all nodes by probability value, then uses density accessibility analysis to group spatially close high-probability nodes into the same cluster. The first cluster contains 15 closely adjacent nodes, forming a compact circular distribution covering an area of ​​approximately 8 square meters. The system automatically delineates its boundary as a high-risk physical area. The second cluster exhibits a strip-like distribution, with nodes arranged linearly along the walkway, extending approximately 6 meters. The system marks this as a potential risk corridor. During boundary delineation, the system comprehensively considers the spatial gradient changes in node probability values ​​and uses an edge detection algorithm to identify areas of abrupt change in probability distribution. Boundary lines are set at points where the probability value drops above a set threshold, ensuring that the boundary of each risk area is located at a significant inflection point in probability change. The final output of high-risk area coordinates is represented in a geodetic coordinate system, perfectly corresponding to the actual location on-site with centimeter-level accuracy, providing precise positioning data for subsequent response measures.

[0033] The entire safety zone association and positioning process demonstrated adaptability to complex environments during actual operation. In the scenario of diverse floor tile layouts in subway station concourses, the system successfully handled mixed topologies containing tiles of different shapes and sizes. The subway station concourse uses a mix of rectangular, square, and hexagonal tiles. The system automatically identified the center point of each type of tile using a feature matching algorithm and established an irregular spatial node network. The system accurately identified an irregular high-risk area at the escalator entrance, composed of twelve tiles of different shapes. The system calculated the optimal boundary representation of this area using an optimization algorithm. The system's real-time updated heat map reflects risk changes at different times. For example, during morning and evening rush hours, the high-risk area expands and intensifies due to increased passenger flow, while the risk area shrinks accordingly during off-peak hours. The accuracy of the positioning results was confirmed through cross-validation of multi-sensor data. The system fused and analyzed the heat map with video surveillance data and infrared thermal imaging data to ensure a high degree of consistency between the marked high-risk areas and the actual locations of safety incidents. The system also has a self-calibration function, which regularly optimizes and adjusts the topology model and propagation algorithm based on actual event data to continuously improve positioning accuracy and reliability.

[0034] The system demonstrates excellent computational efficiency when handling large-scale scenarios. In a large plaza deployment with over 10,000 nodes, the entire process from data acquisition to heat map generation can be completed within three minutes. The system employs a distributed computing architecture, dividing the spatial area into multiple computing units for parallel processing, significantly improving computational speed. For data storage, the system maintains a complete historical topology model and propagation records, supporting retrospective analysis and pattern recognition of historical events. The user interface provides an interactive heat map display, allowing security managers to zoom and rotate the view using gestures to observe risk distribution from different angles. The system also provides an early warning push function, automatically sending alerts to relevant personnel when high-risk areas are detected, including the area's location, risk level, and recommended measures. Through the integration of these functions, the security area association and positioning module provides comprehensive and efficient technical support for the security management of large public places.

[0035] Example 3: Implementation Process of the Adaptive Response Generation Module. The response strategy is automatically configured based on spatial probability data provided by the security event heatmap. The response parameter configuration unit dynamically adjusts the operating parameters of the monitoring system by analyzing the event probability values ​​of each region in the heatmap. When the system detects that the event probability value in a certain region exceeds the threshold of 0.7, it automatically triggers a high-frequency monitoring mode, increasing the sampling frequency of the sensors in that region from the usual 1 time / second to 5 times / second. In practical applications at large transportation hubs, the system identifies areas with persistently high probability values ​​near the security checkpoint and adjusts the monitoring frequency of 128 paving stones in that area to high-frequency mode, while simultaneously shortening the data upload interval from 60 seconds to 20 seconds. This parameter adjustment involves not only changes in the sampling frequency but also adjustments to the data compression algorithm and optimization of the transmission protocol to ensure stable transmission of high-frequency data streams.

[0036] The disturbance test execution unit performs multi-band physical disturbance tests on areas with the most significant changes in probability gradient on the heat map. The system first calculates the probability gradient values ​​between adjacent nodes and identifies boundary areas where the gradient change exceeds 0.15 / m. In the example of a subway platform, the system detected a probability gradient of 0.2 / m at the connection between the platform and the passageway, and then applied a series of physical disturbance tests to this area. The test sequence includes mechanical vibration tests in three frequency bands: low-frequency tests (5-10Hz) simulate the vibration of heavy equipment in operation, mid-frequency tests (20-30Hz) correspond to the vibration generated by pedestrians walking, and high-frequency tests (40-50Hz) simulate the vibration of small equipment. Simultaneously, acoustic tests are conducted, covering a frequency range of 1kHz to 10kHz, with test durations set between 3 and 10 seconds depending on the gradient magnitude. All test data is uploaded to the analysis system in real time to verify the effectiveness of the event response mechanism.

[0037] The environmental factor reconstruction module quantifies and processes external environmental factors through an environmental data acquisition unit. This unit acquires the precise coordinates and millisecond-level timestamps of each paving stone, simultaneously collecting relative humidity and surface temperature data at the corresponding locations. In the application scenario of a park walkway, the system collects environmental data from 1200 monitoring points at a 30-second interval, calculating the environmental suitability value using the following formula: ; in: This represents the environmental suitability value at the i-th point. This is the measured surface temperature value at that point. This is a relative humidity measurement. and These represent the upper and lower limits of the system's permissible temperature, respectively. and Indicates the upper and lower limits of humidity. and The weighting coefficients were determined to be 0.6 and 0.4 based on historical data statistical analysis. All data points were organized according to time series to form an environmental suitability set containing coordinates, timestamps, and suitability values.

[0038] During real-time operation, the environmental data acquisition unit and response generation module form a closed-loop control system. In the implementation case of a commercial complex, the system detected that the afternoon temperature rise in the atrium area led to a decrease in the suitability value. When this value fell below 0.4, a cooling response mechanism was automatically triggered. Simultaneously, the system continuously collects environmental data to establish environmental baseline models for different time periods, distinguishing between normal environmental fluctuations and abnormal events. In terms of data processing, the system uses a sliding window mechanism to maintain the environmental data for the most recent 24 hours, ensuring the timeliness of suitability value calculations. All environmental data is time-series labeled to facilitate trend analysis and pattern recognition.

[0039] The collaborative operation of multiple modules is achieved through a unified data bus, with control commands output by the adaptive response generation module directly acting on the actuators of the floor tile terminals. In stadium applications, the system identifies high-risk areas at the stands exits based on heat map data. It not only activates a high-frequency monitoring mode but also simultaneously adjusts the lighting intensity and broadcast volume in that area, forming a multi-dimensional safety response. The entire system adopts a distributed architecture design, allowing each functional unit to operate and expand independently, ensuring reliability and stability during large-scale deployments. The data processing flow uses a pipeline approach, with the average latency from data acquisition to response execution controlled within 2 seconds, meeting real-time response requirements.

[0040] Example 4: See Figure 4 The risk zoning and labeling module is implemented based on spatial relationship analysis and regional division using data provided by the environmental suitability set. The spatial relationship sorting unit extracts data records containing coordinate information and environmental suitability values ​​from the environmental suitability set. In an application example in a large urban square, the system processed data from over 2000 monitoring points. These monitoring points are distributed in a grid pattern, with a standard spacing of 1.5 meters between adjacent points. The system first calculates the Euclidean distance between each pair of adjacent monitoring points and sorts them according to the distance value from smallest to largest, forming an ordered list of adjacent point pairs. The difference in environmental suitability values ​​between each pair of points is calculated using absolute values. When the difference is less than 0.1, the pair is marked as a consistent segment; when the difference is greater than 0.3, it is marked as a conflict segment. In the area on the east side of the square, the system identified 32 consistent segments and 15 conflict segments, and the distribution of these segments exhibits a clear spatial regularity.

[0041] The zoning label generation unit calculates and processes the environmental suitability value difference rate between adjacent point pairs. The difference rate is defined as the absolute value of the difference in suitability values ​​between two points divided by the average of the suitability values ​​of the two points. In the deployment case of Central Park, the system calculates the difference rate for all identified adjacent point pairs and generates a directional consistency zoning label set based on the magnitude and direction of the difference rate. This label set records the type identifier, spatial location information, and directional index data for each segment. The directional index is determined by comparing the trend of suitability value changes at three consecutive monitoring points; an upward trend is marked as positive, and a downward trend is marked as negative. The zoning label set generated by the system contains detailed descriptions of 287 segments in the entire area, providing basic data support for analysis.

[0042] The rate of change mapping module extracts and analyzes environmental fluctuation characteristics through further processing of the directional consistency partition label set. The environmental fluctuation extraction unit filters out all directionally consistent segments from the partition label set; in the example of a commercial pedestrian street, the system identified 89 directionally consistent segments from 156 segments. For the monitoring points contained in these segments, the system extracts the most recent 30-minute temperature and humidity time series data, with a sampling interval of 1 minute. The sequence data for each monitoring point contains 30 consecutive sampled values. The system performs quality checks on these data, removing invalid data points caused by sensor anomalies, and uses linear interpolation to fill in missing values, ensuring the integrity and continuity of the time series.

[0043] The impact intensity assessment unit calculates the rate of change and analyzes the impact intensity of the extracted time-series data. In the case of an urban transportation hub, the system analyzed 45 directional sections in the waiting hall area. The rate of temperature increase was obtained by calculating the slope of a linear regression of the time series, and the rate of humidity decrease was calculated using the same method. The system correlated the calculated rate values ​​with historical event frequency data to compare the variation characteristics of different sections. During the analysis, the system established a table of correspondence between rate values ​​and event frequencies, and generated the event impact intensity analysis results for each monitoring point based on this table. These results are expressed in numerical form, reflecting the potential impact of environmental changes on the occurrence of safety events. For the impact intensity analysis of monitoring points in the commercial center area, see Table 1.

[0044] Table 1: Data on the impact intensity analysis of the commercial center area are as follows:

[0045] The system's data processing workflow employs a distributed computing architecture, enabling simultaneous data analysis tasks across multiple areas. In the sports stadium example, the system simultaneously performs zoning and impact intensity assessments on the playing field, stands, and entrances / exits, processing data from over 5,000 monitoring points. All analysis results are updated in real time and displayed to administrators through a visual interface, supporting their monitoring and decision-making regarding the security status of different areas. The entire process is fully automated, with an average time delay of less than 3 seconds from data acquisition to analysis result output, ensuring the system's timeliness and effectiveness. Analysis results data is stored in a time-series database, supporting historical data queries and trend analysis functions, providing data support for long-term security management.

[0046] Example 5: The implementation process of the event disturbance identification module is based on the impact intensity analysis results and zoning labeling data. Potential abnormal events are identified by filtering and labeling specific data points. The disturbance point filtering unit processes the analysis results from the impact intensity assessment unit, which include environmental parameters and event correlation indicators for each monitoring point. In practical applications at large transportation hubs, the system first performs time-series analysis on the historical impact intensity data of all monitoring points to calculate the average impact intensity benchmark value for all points. This benchmark value is calculated by using a moving average of the intensity data of all points over the past 24 hours. A sliding window technique is used to ensure the timeliness of the data, and the benchmark value is automatically updated hourly. The system sets a dynamic threshold mechanism to automatically identify monitoring points whose impact intensity values ​​consistently exceed the average benchmark value by a certain range, while also including all points in directional conflict zones in the filtering scope. In the subway station concourse case, the system collects real-time data from 856 monitoring points and, through multiple rounds of filtering, finally determines 47 points that meet the criteria. These points are mainly concentrated in densely populated areas such as security checkpoints and escalator entrances. The system generates detailed feature descriptions for each point, including intensity deviation values, duration length, and spatial distribution characteristics.

[0047] The anomaly point generation unit further analyzes the selected points, combining multi-source data to comprehensively determine their degree of anomaly. In the implementation case of a commercial center, the system established a comprehensive data acquisition network to obtain pesticide application records for each point. This data includes parameters such as application time, dosage concentration, and effective range. The system calculates the rate of change in pesticide dosage and the rate of change in affected area per unit time. The rate of change is calculated using a relative comparison method, obtained by comparing with historical data from the same period, reflecting the degree of anomaly in the current period. The system integrates wind direction change data and humidity fluctuation monitoring data provided by a meteorological monitoring network. This data comes from environmental sensors distributed throughout the area and is updated at a minute-level frequency to ensure the timeliness and accuracy of the data. The system establishes a multi-parameter evaluation model containing twelve evaluation dimensions, assigning a weighted score to each candidate point. The scoring algorithm adopts an adaptive weight allocation mechanism, dynamically adjusting the weight ratio of each dimension according to different environmental conditions. In the shopping mall example, the system, through comprehensive analysis of 32 candidate points, collected more than 200 feature parameters, ultimately identifying 18 points constituting a set of anomaly points for security event disturbances. This point set uses a standardized data structure, containing precise coordinates, anomaly type identifiers, and risk level information for each point, while also recording data acquisition time and processing logs.

[0048] The risk analysis output module performs in-depth processing of the anomaly set to complete the final risk assessment and early warning generation. The joint judgment unit acquires detailed information on all points in the anomaly set, including spatial coordinates, anomaly index data, and environmental parameters. In its deployment in urban parks, the system performs a joint risk assessment on 26 identified anomalies, using a multi-factor weighted algorithm to calculate the joint risk judgment value for each point. This algorithm design includes eight core assessment factors, comprehensively considering multiple dimensions such as the anomaly intensity, spatial clustering degree, environmental influencing factors, and correlation with historical events. Each dimension has a different weight coefficient, which is obtained by training historical data using a machine learning algorithm. The training dataset includes event records and environmental data from the past three years. The system is optimized and adjusted quarterly, updating the weight coefficients based on the latest data to ensure the accuracy of the assessment model. The calculated risk judgment value is standardized, ranging from 0 to 1, with higher values ​​indicating a greater risk level. The system generates a risk assessment report for each point, including a detailed calculation process and analysis of influencing factors.

[0049] The early warning result generation unit filters and marks high-risk locations based on a joint risk assessment value sequence. In the case of sports stadiums, the system sorts and analyzes the risk assessment values ​​of 45 anomaly points and uses a clustering algorithm to identify high-risk locations. These locations not only show high risk values ​​but also exhibit obvious clustering characteristics in spatial distribution. The system further analyzes the spatial relationships of these locations and establishes a propagation model to simulate the possible paths and ranges of the abnormal event's spread. The propagation model considers environmental factors and spatial distance, calculates the possible directions and speeds of event propagation, and marks potentially affected areas for spread risk. The final generated safety event detection and risk warning results are in a structured data format, containing a detailed list of risk locations, risk level assessment, spatial distribution characteristics, and spread risk prediction. The warning results are displayed to safety management personnel in real time through a graphical interface that supports multi-dimensional data visualization. Simultaneously, the system pushes the information to relevant emergency response systems via a standard API interface to ensure timely information transmission and processing.

[0050] The implementation process employs a fully automated data processing workflow, with the entire process from anomaly identification to early warning generation completed automatically by the system. In an application example at an airport terminal, the system is deployed over an area of ​​more than 50,000 square meters, processing data streams from over 2,000 monitoring points per minute. The system utilizes a distributed computing architecture, enabling it to complete the entire process from data acquisition to early warning output within 10 seconds. A robust data quality control mechanism is established, encompassing multiple stages including data verification, outlier filtering, and integrity checks. Sensor data is verified in real-time to ensure the accuracy of the analysis results. All generated risk warning results are accompanied by precise timestamps and spatial coordinates, stored in a standardized data format, and support historical backtracking and trend analysis. The system also provides an early warning feedback mechanism, allowing safety managers to confirm and evaluate warning results through an interactive interface. This feedback data is automatically fed into the model training database to optimize the system's judgment algorithm, forming a closed-loop management system for continuous improvement. Through multi-source data fusion and intelligent analysis, the system effectively identifies potential security risks, providing comprehensive decision support for safety management. Detailed operation logs and performance indicators are generated during system operation, facilitating system optimization and functional upgrades.

[0051] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0052] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent paving tile safety event real-time response system, characterized in that, Comprise: Real-time state monitoring module, based on the sensor array deployed in the pedestrian floor tile, real-time acquisition of floor tile pressure distribution data, surface humidity data and temperature fluctuation data; Dynamic event detection module, based on historical safety event data to build a dynamic event prediction model, and output the potential safety event probability value through the dynamic event prediction model; Multi-dimensional event difference analysis module, the potential safety event probability value is carried out multi-dimensional difference analysis with real-time monitoring data, and the difference characteristic matrix of event level is generated; Safety area correlation positioning module, input the difference characteristic matrix into the space topology network, combine the floor tile layout parameters and environmental influence factors, generate safety event heat distribution map, locate high-risk physical area; Adaptive response generation module, according to the safety event heat distribution map, configure response parameters.

2. The intelligent pedestrian tile safety event real-time response system according to claim 1, characterized in that, The dynamic event detection module specifically comprises: Historical data feature extraction unit, multidimensional decomposition processing is carried out on historical safety event data, the energy proportion of steady component and transient component of pressure waveform is extracted, the correlation mapping of humidity fluctuation and event frequency is established, and temperature fluctuation sequence mode in different scenes is aligned; Dynamic event prediction model construction unit, the processed historical safety event data is input into the prediction network, the basic event prediction value is generated, the prediction deviation caused by environmental interference is corrected, and the prediction weight is dynamically adjusted according to the real-time acquisition pressure distribution data; Event probability value calculation unit, input real-time acquisition data into the dynamic event prediction model, and obtain the potential safety event probability value.

3. The intelligent pedestrian tile safety event real-time response system according to claim 2, wherein, The multi-dimensional event difference analysis module specifically comprises: Time domain cumulative difference calculation unit, the potential safety event probability value and real-time monitoring data are compared in a sliding manner in a preset time window, the cumulative difference amount in each window is calculated, and a time domain difference vector is generated; Space distribution offset detection unit, the space distribution of the potential safety event probability value and the real-time monitoring data is regionally decomposed, the distribution offset index of each region is extracted, and a space offset vector is constructed; Event sequence similarity evaluation unit, match the distance distribution of potential safety event probability value and real-time event sequence, quantify the time synchronization error of sequence mutation point, generate event similarity vector; Difference characteristic matrix generation unit, the time domain difference vector, the space offset vector and the event similarity vector are multi-dimensionally spliced, the feature is normalized, and the difference characteristic matrix of event level is output.

4. The intelligent pedestrian tile safety event real-time response system according to claim 3, wherein, The safety area correlation positioning module specifically comprises: Floor tile topology modeling unit, the space node connection topology graph is constructed according to the floor tile layout parameters, the distance parameters between nodes are labeled, and the environmental influence factors are superimposed to generate the topology model including distance matrix and area weight matrix; Event propagation simulation unit, the difference characteristic matrix is mapped to the corresponding node of the topology model, the event propagation deduction is executed, the attenuation factor of node abnormal event is calculated, the cross-region event correlation characteristics are captured, and the diffusion path of abnormal event in the topology network is simulated; Heat distribution map generation unit, the event occurrence frequency of each path in the simulation propagation is counted, the event residence probability value is calculated combined with the distance parameter, and the safety event heat distribution map is generated; A physical area positioning unit performs spatial clustering analysis on the safety event heat distribution map, identifies an event probability aggregation area, and delineates a high-risk physical boundary.

5. The intelligent pedestrian tile safety event real-time response system according to claim 4, wherein, The adaptive response generation module specifically includes: A response parameter configuration unit enables a high-frequency monitoring mode according to a probability value area of the safety event heat distribution map; A disturbance test execution unit applies a multi-frequency physical disturbance test to an adjacent node area with the largest probability gradient change in the safety event heat distribution map.

6. The intelligent pedestrian tile safety event real-time response system according to claim 1, wherein, Further comprising an environmental factor reconstruction module: An environmental data acquisition unit acquires human walking tile position coordinates and timestamps, collects environmental humidity data and ground temperature data at corresponding positions, associates coordinate positions and calculates environmental suitability values, and generates an environmental suitability set.

7. The intelligent pedestrian tile safety event real-time response system according to claim 6, wherein, Further comprising a risk partition labeling module: A spatial relationship sorting unit extracts environmental suitability values and coordinate data in the environmental suitability set, sorts adjacent point pairs according to spatial distance, and labels consistent sections and conflict sections; A partition labeling generation unit generates a directional consistency partition labeling set based on the environmental suitability value difference rate of adjacent point pairs.

8. The intelligent pedestrian tile safety event real-time response system according to claim 7, wherein, Further comprising a change rate mapping module: An environmental fluctuation extraction unit filters directional consistent sections from the directional consistency partition labeling set, extracts air temperature time series and humidity time series; An influence intensity evaluation unit calculates air temperature rise rate and humidity drop rate, compares change amplitudes in combination with event occurrence frequency, and generates event influence intensity analysis results.

9. The intelligent pedestrian tile safety event real-time response system according to claim 8, wherein, Further comprising an event disturbance identification module: A disturbance point screening unit identifies points with an influence intensity value greater than an average benchmark value in the event influence intensity analysis results, as well as points in directional conflict sections; An abnormal point set generation unit labels disturbance points based on drug dosage change ratio and area change ratio, in combination with wind direction amplitude data and humidity fluctuation data, and generates a safety event disturbance abnormal point set.

10. The intelligent pedestrian tile safety event real-time response system according to claim 9, wherein, Further comprising a risk analysis output module: A joint determination unit acquires point information in the safety event disturbance abnormal point set and calculates a joint risk determination value; An early warning result generation unit screens high-risk points based on a joint risk determination value sequence, labels diffusion risk, and generates safety event detection and risk early warning results.

Images