Intelligent vehicle door, shielding door safety detection method and system, electronic equipment and medium

Abstract

The invention relates to the field of rail traffic safety detection, and discloses an intelligent vehicle door and platform screen door safety detection method and system, electronic equipment and a medium, and the method comprises the following steps: collecting vehicle door and platform screen door vibration signals, environment temperature and humidity and passenger flow data; performing spectral analysis and extracting a time domain amplitude; dynamically generating a vibration threshold based on the environment and passenger flow data; when the frequency spectrum is abnormal and the time domain exceeds a threshold value, judging that the frequency spectrum is abnormal; maintaining or starting a standby door according to a timetable decision; the vehicle door is controlled in a linkage mode and prompts are pushed; the system comprises a data acquisition module, a spectrum analysis module, a dynamic threshold generation module, an abnormality judgment module and a maintenance decision module. According to the invention, through multi-source data fusion and frequency spectrum-time domain joint determination, the anomaly detection precision is improved, and the false alarm and missing alarm rate is reduced; and in combination with an environment dynamic threshold and a timetable driven decision, the adaptability to complex working conditions is enhanced, and the maintenance resource scheduling efficiency is optimized.

Description

Technical Field

[0001] The present invention relates to the field of rail transit safety detection, and specifically to a safety detection method, system, electronic device and medium for intelligent doors and platform screen doors. Background Art

[0002] In rail transit, the safety detection system of intelligent doors and platform screen doors is the core link to ensure operation safety. In the prior art, such systems mostly rely on single-dimensional sensor data for anomaly detection and determine faults by setting fixed thresholds. However, the complex operating environment can cause sensor signal drift or noise interference, making the traditional method have inherent defects in the data acquisition stage. For example, a single vibration sensor is difficult to distinguish mechanical anomalies from environmental interference and is prone to false alarms; at the same time, the setting of fixed thresholds is difficult to adapt to the signal baseline changes under different working conditions, resulting in a significant increase in the false positive rate in high-temperature and high-humidity scenarios.

[0003] In terms of anomaly detection algorithms, the prior art mostly adopts time-domain amplitude thresholds or fixed-frequency band energy analysis, lacking in-depth analysis of spectral features and cross-dimensional verification. Such methods are difficult to effectively identify composite fault features such as harmonic distortion and instantaneous shocks, resulting in missed detections or false alarms. For example, the early wear of the drive motor of an intelligent door may be manifested as energy anomalies in a specific frequency band, while traditional time-domain detection is difficult to capture such signs, leading to delays in the maintenance timing.

[0004] In addition, the maintenance decision-making and emergency response mechanisms of existing systems mostly rely on manual experience or static rules, and it is difficult to real-time associate the train operation schedule with the resource status. When an anomaly is detected, there are lags in the generation of maintenance work orders, spare parts scheduling, passenger guidance, etc., resulting in an extended operation interruption time and low utilization rate of standby doors. Especially during peak hours, traditional methods are difficult to balance the fault handling efficiency and passenger evacuation requirements, presenting potential safety hazards.

[0005] Therefore, the present invention proposes a safety detection method, system, electronic device and medium for intelligent doors and platform screen doors to solve the deficiencies of the prior art. Summary of the Invention

[0006] Aiming at the deficiencies of the prior art, the present invention provides a safety detection method, system, electronic device and medium for intelligent doors and platform screen doors, solving the problems of high false alarm and missed detection rates, low operation and maintenance efficiency, and insufficient passenger guidance efficiency caused by insufficient reliability of multi-source data, single anomaly detection dimension, poor adaptability to dynamic environments, and lagged emergency response in the existing door safety detection technology.

[0007] To achieve the above objectives, the present invention is realized through the following technical solutions: A safety detection method for intelligent doors and platform screen doors, including the following steps: Collect vibration signals, ambient temperature and humidity, and passenger flow density data of intelligent vehicle doors and platform screen doors; Perform spectral analysis on the vibration signals, detect abnormal spectral features, and extract the time-domain amplitude of the vibration signals; Generate a dynamic vibration amplitude threshold according to the joint distribution of the ambient temperature and humidity and the passenger flow density; When the abnormal spectral features meet the preset conditions and the time-domain amplitude exceeds the dynamic vibration amplitude threshold, it is determined as abnormal; Generate a maintenance decision based on the train operation schedule, and select immediate repair or enable the standby door; Execute the door control instruction according to the maintenance decision and push a prompt message to the passengers.

[0008] Preferably, the step of collecting vibration signals, ambient temperature and humidity, and passenger flow density data of intelligent vehicle doors and platform screen doors includes: Deploy vibration sensors at the door hinges to collect vibration signals, and the sampling frequency is not less than 1 kHz; Synchronously obtain ambient temperature and humidity data of the door area through temperature and humidity sensors; Real-time count the passenger flow density data of the door area through a lidar passenger flow counter; Perform wavelet denoising on the vibration signals and standardize them to a unified dimension.

[0009] Preferably, the abnormal spectral features include at least one of the following: The proportion of the energy of the non-harmonic components in the total energy exceeds 10%. The proportion of the energy of the non-harmonic components in the total energy is calculated through the following steps: Perform a fast Fourier transform on the vibration signal to extract the fundamental frequency And the energy of the harmonic components. The calculation formula for the proportion of the non-harmonic component energy is: ; Wherein, is the total energy of the vibration signal in the frequency domain, is the fundamental frequency And the sum of the energies of its integer multiple harmonic components, is the proportion of the non-harmonic energy; Determine the proportion of the non-harmonic energy in the total energy; The number of mutations of the phase of the vibration signal on the Poincaré section exceeds 3 times.

[0010] Preferably, the step of extracting the time-domain amplitude of the vibration signal includes the following steps: Segment the vibration signal by a fixed time window, and the duration of each segment is 100 ms; Calculate the time-domain amplitude of each segment of the signal, and the time-domain amplitude is the effective value or the peak value; The calculation formula for the effective value is as follows: ; Wherein, represents the amplitude of the vibration signal at the th sampling point, represents the total number of sampling points of a single-segment signal, reflects the energy intensity of the vibration signal within the window; The calculation formula for the peak value is as follows: ; Wherein, represents the absolute value of the discrete sampling points of the vibration signal, is the maximum value of the absolute values of all sampling points within the window, characterizes the instantaneous impact intensity of the vibration signal within the window; Normalize the time-domain amplitude and map it to the range of 0-1.

[0011] Preferably, the steps of generating the dynamic vibration amplitude threshold include: Divide the temperature, humidity, and passenger flow density into preset intervals respectively, and establish a three-dimensional joint distribution data set; Train a Copula function model based on historical data to fit the joint probability distribution of temperature, humidity, and passenger flow density; Extract the upper limit of the 95% confidence interval of the joint probability distribution and convert it into a vibration amplitude threshold through linear mapping; According to the temperature, humidity, and passenger flow density data collected in real time, query the upper limit value of the corresponding confidence interval of the joint distribution as the dynamic threshold; Retrain the Copula function model every 24 hours to update the joint distribution parameters.

[0012] Preferably, the steps of determining an anomaly include: When the following conditions are simultaneously met, trigger an anomaly determination: The proportion of the energy of the non-harmonic components in the abnormal frequency spectrum characteristics exceeds 10% or the number of phase mutations in the Poincaré section exceeds 3 times; The time-domain amplitude continuously exceeds the dynamic vibration amplitude threshold for 5 seconds.

[0013] Preferably, the generation of the maintenance decision includes: Obtain the arrival time of the next train and calculate the latest maintenance start time. The calculation formula for the latest maintenance start time is as follows: ; Wherein, is the planned arrival time of the next train, The standard time required to complete the door repair, is the safety buffer time, and is the latest repair start time; If the current time exceeds , the standby door is enabled.

[0014] The present invention also provides an intelligent door and platform screen door safety detection system, including: A data acquisition module for acquiring vibration signals, ambient temperature and humidity, and passenger flow density data in the door area; A spectrum analysis module for performing spectrum analysis on the vibration signals to detect abnormal spectrum features and extracting the time-domain amplitude of the vibration signals; A dynamic threshold generation module for generating a dynamic vibration amplitude threshold according to the joint distribution of ambient temperature and humidity and passenger flow density; An abnormality determination module for determining abnormality when the abnormal spectrum features meet preset conditions and the time-domain amplitude exceeds the dynamic vibration amplitude threshold; A maintenance decision-making module for generating a maintenance decision based on the train operation schedule and selecting immediate repair or enabling the standby door; A control execution module for executing the door control instruction according to the maintenance decision and pushing a prompt message to passengers.

[0015] The present invention also provides an electronic device, including a memory and a processor. The memory stores a computer program executable by the processor, and when the processor executes the computer program, it can implement the intelligent door and platform screen door safety detection method.

[0016] The present invention also provides a storage medium storing a computer program, and when the computer program is executed by a processor, it can implement the intelligent door and platform screen door safety detection method.

[0017] The present invention provides an intelligent door and platform screen door safety detection method, system, electronic device and medium. It has the following beneficial effects: 1. Through the joint determination technical solution of spectrum analysis and time-domain amplitude statistics, the present invention significantly improves the reliability of abnormal detection. The limitation of relying on a single detection dimension in the prior art is broken through, and the technical problems of harmonic leakage detection and instantaneous impact misjudgment are overcome.

[0018] 2. Based on the joint modeling technical solution of environmental parameters and historical data, the present invention realizes the dynamic adaptation of vibration thresholds. Compared with the traditional fixed threshold method, it solves the problem of insufficient detection sensitivity under complex working conditions and enhances the adaptability of the system to temperature, humidity and passenger flow changes.

[0019] 3. The present invention adopts a maintenance decision-making and resource dynamic matching technical solution driven by a train timetable, which significantly shortens the emergency response time. The deficiencies of lagging manual scheduling or low resource utilization rate in the prior art are eliminated, and the closed-loop coordination of door control and passenger guidance is achieved.

[0020] 4. The present invention adopts a multi-sensor fusion and anti-interference filtering technical solution to achieve high-precision data acquisition and noise suppression. Compared with the defect that a single sensor in the prior art is vulnerable to environmental interference, the core problems of signal distortion and false triggering are solved, providing a stable input for subsequent analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a flowchart of the method of the present invention; Figure 2 is a system architecture diagram of the present invention; Figure 3 is a schematic diagram of the equipment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0022] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0023] Please refer to the attached Figure 1 , the embodiments of the present invention provide a method for safety detection of intelligent doors and platform screen doors, including the following steps: S1. Collect vibration signals, ambient temperature and humidity, and passenger flow density data of intelligent doors and platform screen doors; S2. Perform spectrum analysis on the vibration signals, detect abnormal spectrum features, and extract the time-domain amplitude of the vibration signals; S3. Generate a dynamic vibration amplitude threshold according to the joint distribution of the ambient temperature and humidity and the passenger flow density; S4. When the abnormal spectrum features meet the preset conditions and the time-domain amplitude exceeds the dynamic vibration amplitude threshold, it is determined as abnormal; S5. Generate a maintenance decision based on the train operation timetable, and select immediate maintenance or enable the standby door; S6. Execute the door control instruction according to the maintenance decision and push a prompt message to the passengers.

[0024] For step S1, in this embodiment, the step of collecting vibration signals, ambient temperature and humidity, and passenger flow density data of the intelligent door and the platform screen door is achieved through the collaborative work of multi-source sensors. Specifically, the data collection process needs to meet the input requirements of subsequent modules such as spectrum analysis and dynamic threshold calculation to ensure spatio-temporal synchronization and data validity.

[0025] In some embodiments, a piezoelectric ceramic accelerometer is selected as the vibration sensor and is deployed 3 - 5 cm inside the door hinge, with its axis perpendicular to the door movement direction. The sampling frequency of the vibration signal is set to 1.2 kHz, covering the transient impact frequency band (0 - 500 Hz) during door opening and closing. As an option, after the sensor signal is converted by a 24-bit ADC, it is transmitted to the microcontroller through the SPI interface and a timestamp is added.

[0026] In a possible implementation, an SHT35 model is used as the temperature and humidity sensor, which is installed near the ventilation opening at the top of the door to avoid direct sunlight. The data collection interval is set to 10 seconds, and the measurement range covers -40°C to 85°C (temperature) and 0% to 100% RH (humidity). The timestamp error between the ambient data and the vibration signal does not exceed ±0.1 second, and synchronous sampling is triggered by a hardware interrupt.

[0027] For the passenger flow density data, a TOF lidar scans a rectangular area of 1.5 m × 2 m in front of the door at a scanning frequency of 10 Hz. The number of people is counted using the background difference method: ; where, is the passenger flow density (persons / m 2 ); is the number of pedestrians; is the monitored area; The specific steps for wavelet denoising processing of the vibration signal are as follows: 1. Select the Daubechies4 wavelet basis for 5-layer decomposition, retain the low-frequency coefficients, and perform threshold processing on the high-frequency coefficients: ; where, is the wavelet threshold; is the noise standard deviation (calculated through the signal stationary section); is the signal length; 2. Perform soft threshold shrinkage on the high-frequency coefficients exceeding the threshold λ: ; ; where, is the wavelet coefficient after threshold processing; is the original wavelet coefficient.

[0028] 3. Reconstruct the signal and standardize it to a unified dimension: ; wherein, is the mean of historical data; is the standard deviation; The signal after standardization.

[0029] In some embodiments, the standardization process further includes range mapping: linearly transforming the amplitude of the vibration signal to the range of [-1, 1], mapping the ambient temperature and humidity to the interval of [0, 100], and mapping the passenger flow density to the interval of [0, 10], so as to eliminate the interference of dimension differences on subsequent analysis.

[0030] In terms of hardware deployment, the vibration sensor and the lidar are networked using the RS-485 bus, and the temperature and humidity sensor is directly connected to the main control unit through the I 2 C interface. The data packet is encapsulated in JSON format, including the device ID, timestamp, data value and check code, and is uploaded to the cloud server through the MQTT protocol.

[0031] It should be noted that when the door is in a completely closed and stationary state, the baseline noise amplitude collected by the vibration sensor should be less than 0.02g. If the noise exceeds the limit, trigger the self-check program: disconnect the load and collect the no-load signal for 10 seconds, calculate the noise power spectral density, and automatically adjust the ADC reference voltage or gain parameter.

[0032] For data storage, the original signal is stored in segments of 5 minutes on the local SD card, and at the same time, the compressed feature data (including the frequency domain energy distribution, the mean temperature and humidity, and the peak passenger flow density) is uploaded. The compression algorithm uses Huffman coding, and the compression ratio is not less than 4:1.

[0033] In summary, this step effectively solves the integrity and consistency problems of multi-source data acquisition under complex working conditions through hardware deployment optimization and algorithm-level preprocessing, laying a technical foundation for subsequent anomaly detection and decision-making control.

[0034] For step S2, in this embodiment, the step of performing spectral analysis on the vibration signal and extracting the time-domain amplitude is based on the preprocessed vibration signal (see step S1) to perform frequency-domain feature mining and time-domain statistic calculation. Specifically, the spectral analysis module and the dynamic threshold generation module form a cascaded processing link, and anomaly determination is realized through abnormal spectral feature recognition and time-domain overrun detection.

[0035] In some embodiments, the spectral analysis is implemented using the fast Fourier transform (FFT). The vibration signal after noise reduction is intercepted to a length of 1024 points, and the Hanning window is used for weighting to reduce spectral leakage. The fundamental frequency Determined by the peak search method: find the frequency point corresponding to the maximum amplitude within the range of 5 - 50 Hz. The harmonic components are defined as integer multiples of the frequency ( ), and the sum of their energies is calculated as: ; where represents the energy value at the th harmonic frequency; is the harmonic order, taking integer values from 1 to 5, representing 1 to 5 times the fundamental frequency; is the sum of harmonic energies, representing the cumulative energy of the fundamental frequency and its first 5 harmonic components.

[0036] The proportion of non - harmonic energy is calculated by the following formula: ; where is the total energy of the vibration signal in the frequency domain, is the fundamental frequency and the sum of the energies of its integer - multiple harmonic components, is the proportion of non - harmonic energy.

[0037] In a possible implementation, the Poincaré section phase mutation detection is performed according to the following steps: 1. Perform Hilbert transform on the vibration signal to extract the instantaneous phase ; 2. Intercept the phase points per period and plot them on the Poincaré section; 3. When the angular difference between adjacent phase points exceeds 60°, it is counted as a mutation, and the number of mutations within the unit time window is statistically counted.

[0038] For time - domain amplitude extraction, the vibration signal is processed in segments with a fixed time window. Generally, the window length is set to 100 ms, and with a sampling rate of 1.2 kHz, a single window contains 120 sampling points. There are two ways to calculate the time - domain amplitude: Root Mean Square (RMS): ; where represents the amplitude of the vibration signal at the th sampling point, represents the total number of sampling points in a single - segment signal; reflects the energy intensity of the vibration signal within the window.

[0039] Peak: ; Among them, represents the absolute value of the discrete sampling points of the vibration signal; is the maximum value of the absolute values of all sampling points within the window; characterizes the instantaneous impact intensity of the vibration signal within the window; .

[0040] In some embodiments, the time-domain amplitude normalization processing adopts linear mapping: ; Among them, is the original time-domain amplitude value; is the historical minimum amplitude value, and the lower limit is determined by statistically analyzing the time-domain amplitude data of the past 30 days; is the historical maximum amplitude value, and the upper limit is determined by statistically analyzing the data in the same time period; is the normalized amplitude value, which is mapped to the [0, 1] interval to eliminate the influence of magnitude differences on subsequent threshold determination.

[0041] It should be noted that when the peak value is used as the time-domain amplitude, additional pulse interference suppression is required: if the peak value in three consecutive windows exceeds 3 times the standard deviation of the historical mean, data resampling verification is triggered.

[0042] At the hardware implementation level, the FFT operation is accelerated by the FPU unit of the STM32H7 microcontroller, and the single 1024-point FFT calculation time ≤ 2 ms. The phase mutation detection and the time-domain amplitude calculation are executed in parallel by independent threads, and data blocking is avoided through a double-buffer mechanism.

[0043] In summary, this step combines spectrum analysis and time-domain amplitude joint calculation, combines hardware-level acceleration (FFT parallelization, dynamic boundary update) and frequency-domain - time-domain cross-verification strategy, significantly improves the anomaly detection sensitivity and anti-interference ability, effectively solves the technical bottlenecks of harmonic leakage detection, instantaneous impact misjudgment and dimension inconsistency in traditional methods, and provides a highly reliable input for dynamic threshold generation and multi-level anomaly decision-making.

[0044] For step S3, in this embodiment, the dynamic vibration amplitude threshold generation method is based on the cross-domain coupling relationship of environmental temperature and humidity, passenger flow density and vibration signal, and realizes threshold adaptive adjustment by establishing a multi-dimensional joint probability model, solving the problem of high false alarm rate caused by environmental interference of traditional fixed thresholds. The method receives the environmental data collection in step S1 and the vibration feature extraction results in step S2, and generates a threshold boundary matching the current working condition through probability statistics and non-linear mapping, providing a dynamic determination benchmark for subsequent anomaly decision-making.

[0045] Three-dimensional joint distribution modeling: In some embodiments, the temperature, humidity, and passenger flow density are respectively divided into preset intervals to construct a three-dimensional joint distribution data set. Specifically, the temperature interval is divided into , , the humidity interval , , the passenger flow density interval , to form discretized grid cells. The maximum value and the minimum value of the historical vibration amplitude are stored in each grid cell. The data collection period is 30 days, covering day and night and different weather conditions.

[0046] Copula function model training: In a possible implementation, the Copula function model is trained based on historical data to fit the joint probability distribution of temperature, humidity, and passenger flow density. The model construction process includes: 1. Marginal distribution modeling: The temperature follows a normal distribution ; where is the mean of the daily average temperature, is the standard deviation of temperature fluctuation.

[0047] The humidity follows a Beta distribution ; where is the first shape parameter of the Beta distribution; is the second shape parameter of the Beta distribution; the parameters are determined by the method of moment estimation.

[0048] The passenger flow density follows a Poisson distribution ; where is the Poisson distribution intensity parameter; is calibrated by the historical peak period data.

[0049] 2. Copula function selection: The Gaussian Copula function is used to describe the non-linear correlation between variables, and its expression is: ; where , , are the marginal distribution probability values of temperature, humidity, and passenger flow density; is the correlation coefficient matrix, is the multivariate normal distribution function; is the inverse function of the standard normal distribution.

[0050] 3. Parameter estimation and verification: Optimize by maximum likelihood estimation The Kolmogorov-Smirnov test was used to verify the goodness of fit of the joint distribution (p value < 0.05) Confidence interval extraction and threshold mapping: Specifically, the upper limit of the 95% confidence interval of the joint probability distribution is extracted and converted into a vibration amplitude threshold through linear mapping. The threshold generation formula is defined as: ; in, is the 95% quantile of the joint distribution, representing the extreme vibration risk under the current environmental combination; is the slope coefficient, with a value of 0.8-1.2, calibrated according to the stiffness of the door material; is the offset, determined by the normalized amplitude baseline of step S2 (default ).

[0051] For example, when the real-time temperature is 25°C, the humidity is 60%RH, and the passenger flow density is 2 people / square meter, the query joint distribution is ,like , , then the dynamic threshold is .

[0052] Dynamic threshold update mechanism: As an option, the Copula function model is retrained every 24 hours to update the joint distribution parameters. The update trigger conditions include: Environmental data distribution deviation (Calculate the difference between historical and new data through KL divergence); The cumulative number of vibration amplitude exceeding limit events is ≥ 3 times / day; Manual forced calibration command input.

[0053] During the update process, a sliding window mechanism is used to retain the data of the last 30 days and remove outliers ( in principle).

[0054] In summary, this step solves the technical defects of high false alarm rate and poor adaptability of traditional fixed thresholds due to environmental interference through cross-domain joint modeling of environmental parameters and vibration amplitude, combined with Copula probability distribution fitting and dynamic confidence interval mapping mechanism, and realizes the adaptive optimization of vibration thresholds under working conditions, providing a dynamic judgment benchmark for multi-level abnormal decision-making.

[0055] For step S4, in this embodiment, the anomaly determination method is based on the spectrum-time domain feature extraction result of step S2 and the dynamic vibration amplitude threshold generated in step S3, and realizes high-confidence anomaly detection through a multi-condition joint decision mechanism, solving the technical problems of high false positive rate of a single index and sensitivity to environmental interference. The determination process is strictly coupled with the output of the upstream module to ensure logical coherence and technical closed-loop.

[0056] Anomaly trigger conditions: Specifically, when the following conditions are met simultaneously, anomaly determination is triggered: 1. The spectrum anomaly feature meets the preset condition: The proportion of the energy of non-harmonic components exceeds 10%, that is: ; The number of phase mutations in the Poincaré section exceeds 3 times: Through the Hilbert instantaneous phase sequence analysis in step S2, the number of phase difference mutations between adjacent vibration periods within the unit time window is statistically analyzed to be ≥ 3 times.

[0057] 2. The time-domain amplitude continuously exceeds the limit: The root mean square (ARMS) or peak value (Apeak) of the time domain continuously exceeds the dynamic threshold Athreshold generated in step S3 for 5 seconds. The calculation formula is: ; In a possible implementation, the anomaly determination is executed through the following process: Conditional parallel detection: The detection threads of the spectrum feature and the time-domain amplitude run independently, and the data buffer adopts a double-queue structure to store the spectrum feature marks and the time-domain amplitude sequences in the most recent 10 seconds respectively.

[0058] Temporal consistency verification: When the spectrum anomaly mark appears, start the sliding window check of the time-domain amplitude (window length: 5 seconds, step size: 1 second), and require that the duration of exceeding the limit covers at least 80% of the window length (that is, ≥ 4 seconds of exceeding the limit within 5 seconds).

[0059] Dynamic threshold coupling: Threshold is updated in real time. For example, when the ambient temperature rises from 25 °C to 35 °C, the threshold is adjusted from 1.85 g to 2.1 g (based on the output of the Copula model in step S3), ensuring that the determination conditions are adaptively adjusted according to the working conditions.

[0060] Implementation example: Suppose the following data is detected for a subway door during the morning rush hour: Spectrum feature: , number of phase mutations = 4 times / second; Time-domain amplitude: (Dynamic threshold ), continuous over-limit duration = 5.3 seconds; According to the determination conditions, when the spectrum anomaly ( and the phase mutation occurs 23 times) and the time-domain continuous over-limit (5 seconds) are satisfied simultaneously, an abnormal alarm in step S4 is triggered, and an emergency braking control in step S5 is linked.

[0061] In summary, this step significantly improves the reliability of abnormal determination and environmental robustness through the joint decision-making mechanism of spectrum non-harmonic characteristics and time-domain continuous over-limit conditions, combined with dynamic threshold mapping and multi-level timing verification strategies, effectively solves the technical defects of false alarms of single indicators, sensitivity to transient interference, and serious determination lag in traditional methods, and provides an accurate trigger basis for multi-level alarms and closed-loop control.

[0062] For step S5, in this embodiment, the maintenance decision-making generation method is dynamically associated with the abnormal determination result of step S4 and the train operation schedule, and through the maintenance time window prediction and resource scheduling optimization algorithm, a quick response decision under abnormal conditions is realized, solving the technical problems of low efficiency of traditional manual scheduling and insufficient fault tolerance redundancy. The decision-making logic strictly depends on the output data of the upstream module and the real-time operation status to ensure system-level safety and operation continuity.

[0063] Calculation of the latest maintenance time window: Specifically, obtain the planned arrival time of the next train , combined with the standard maintenance man-hours and the safety buffer time, generate the latest maintenance start time , and the calculation formula is: ; Among them, is the planned arrival time of the next train, sourced from the train operation schedule database; is the standard maintenance man-hours; is the safety buffer time, used to offset uncertain factors such as maintenance delays and equipment preheating; is the latest maintenance start time. If the current time , it is determined that the maintenance cannot be completed on time, and the standby door is forcibly enabled.

[0064] Decision execution and resource scheduling: In a possible implementation manner, the maintenance decision execution process includes: Maintenance feasibility assessment: If , generate an immediate maintenance instruction, synchronously schedule the nearest maintenance personnel (based on GPS positioning) and the spare part inventory status (such as the guide rail slider inventory ≥ 2). Spare door enabling conditions: When occurs, send a spare door activation request to the station control center and lock the power supply of the faulty door to prevent misoperation.

[0065] Dynamic adjustment mechanism: In some embodiments, dynamically optimize according to historical maintenance data , and the formula is: ; Where is the optimized standard maintenance man-hour; is the historical data weight coefficient; is the moving average of historical maintenance time consumption; is the preset standard maintenance man-hour. For example, if the historical average maintenance time is 520 seconds, the new standard man-hour is: ; As an option, when the spare door enabling fails (such as the self-check of the spare door fails), perform the following operations: Start the train approaching speed reduction protocol (speed limit 5 km / h) and extend the passenger boarding and alighting time to 60 seconds; Temporarily relax the vibration detection threshold ( ) through the dynamic threshold model in step S3 to avoid operation interruption caused by false alarms; Dispatch the mobile maintenance platform to the location of the faulty door for rapid repair.

[0066] In summary, this step significantly improves the timeliness and reliability of maintenance decisions through the dynamic coupling analysis of the train operation schedule and real-time maintenance man-hours, combined with the optimization of maintenance resource scheduling and redundancy tolerance strategies, effectively solves the technical bottlenecks of lagging maintenance response, low utilization rate of spare resources, and rigid decision-making in traditional methods, and provides intelligent decision-making support for the balance of operation safety and efficiency.

[0067] For step S6, in this embodiment, the door control and information push method is based on the maintenance decision result of step S5, and realizes rapid response and passenger guidance under abnormal conditions through multi-channel instruction execution and hierarchical alarm strategies, solving the technical problems of lagging control instructions, unfriendly human-computer interaction, and insufficient emergency coordination in traditional methods. The execution logic tightly couples the output data of the upstream module with the hardware state to ensure the reliability and safety of system-level operations.

[0068] Specifically, according to the decision result of step S5 (immediate repair or enable the spare door), perform the following operations: Immediate repair mode: Send a locking instruction to the faulty door controller, cut off the power supply of the drive motor, and synchronously activate the maintenance mode indicator light (red LED, flashing frequency 2Hz); Spare door enable mode: Send an activation instruction to the spare door controller, switch to the redundant power module, and unlock the mechanical linkage device.

[0069] In a possible implementation, if the self-check of the spare door fails (such as the return value of the position sensor exceeding the limit), perform the following operations: Retry the activation instruction (up to 3 times, with an interval of 5 seconds); If it still fails, start the degraded operation mode (door half-open limit, manual operation by passengers); Synchronously trigger the threshold relaxation mechanism of step S4 ( ) to avoid false alarms of secondary anomalies.

[0070] Passenger prompt message push The prompt message includes audible and visual alarms and mobile terminal notifications. The specific implementation method is as follows: Display screen prompt: On the in-carriage LED display screen, scroll and display the faulty door number (such as "Door 3 failure") and guiding information ("Please use the spare door"), with a refresh frequency of 1Hz and the font color red (RGB value #FF0000); Broadcast announcement: Generate a voice prompt ("Dear passengers, Door 3 is out of service. Please move to Door 5") through the TTS engine, with an audio sampling rate of 16kHz and a loop playback interval of 30 seconds; Mobile terminal push: Send a PUSH notification (JSON format) to the smartphones of in-station passengers, including the location of the faulty door (GPS coordinates with an accuracy of ±10m) and the recommended route (based on Bluetooth Beacon positioning).

[0071] Priority strategy: In some embodiments, the information push priority is dynamically adjusted according to the fault level. The formula is: ; Among them, is the information push priority. The smaller the value, the higher the priority; the priority is the highest level, preempting the resources of the display screen and broadcast channel.

[0072] In summary, this step significantly improves the emergency response speed and passenger guidance efficiency by maintaining the coordinated execution of decisions and multimodal instructions, combining hardware-level fault tolerance control (spare door redundancy switching, rapid response of solenoid valves) with a dynamic priority mapping strategy, effectively solving the technical bottlenecks of control lag, human-machine interaction fragmentation, and fault diffusion in traditional methods, and providing a closed-loop guarantee for train safety operation and passenger information services.

[0073] Please refer to Figure 2 , the present invention also provides an intelligent door and platform screen door safety detection system, including: A data acquisition module for collecting vibration signals, ambient temperature and humidity, and passenger flow density data in the door area; For the data acquisition module, multi-sensor fusion technology is adopted to collect vibration signals, ambient temperature and humidity, and passenger flow density data in the door area in real time. Through anti-interference circuit design and timestamp synchronization mechanism, the spatio-temporal consistency of multi-source data is ensured.

[0074] A spectrum analysis module for performing spectrum analysis on the vibration signal to detect abnormal spectrum features and extracting the time-domain amplitude of the vibration signal; For the spectrum analysis module, based on the fast Fourier transform (FFT) and Hilbert-Huang transform (HHT), the frequency-domain energy distribution analysis of the vibration signal is carried out to detect abnormal spectrum features such as non-harmonic components and phase mutations; synchronously extract statistical quantities such as the root mean square (RMS) and peak value in the time domain to provide two-dimensional feature inputs for abnormal determination.

[0075] A dynamic threshold generation module for generating a dynamic vibration amplitude threshold according to the joint distribution of ambient temperature and humidity and passenger flow density; For the dynamic threshold generation module, through the joint probability distribution modeling (Copula function) of environmental parameters (temperature and humidity, passenger flow density), combined with the historical vibration extreme value statistics, the vibration amplitude threshold is dynamically generated; the sliding window data update and covariance matrix optimization are adopted to realize the adaptive adjustment of the threshold with the working conditions.

[0076] An abnormal determination module for determining as abnormal when the abnormal spectrum features meet the preset conditions and the time-domain amplitude exceeds the dynamic vibration amplitude threshold; For the abnormal determination module, based on the joint decision mechanism of spectrum features (non-harmonic energy ratio > 10%, number of phase mutations ≥ 3 times / second) and time-domain amplitude (continuous overrun ≥ 5 seconds), multi-condition logic verification is performed; integrating a time-sequence sliding window and a hardware-level timer to ensure the real-time performance and fault tolerance of the determination result.

[0077] A maintenance decision module for generating a maintenance decision based on the train operation schedule and selecting immediate repair or enabling a spare door; For the maintenance decision-making module associated with the train operation schedule, calculate the latest maintenance start time window ( ), and generate decision instructions (immediate maintenance / enable spare door) by combining maintenance resource scheduling (personnel positioning, spare part inventory); support the dynamic optimization of maintenance man-hours driven by historical data.

[0078] The control execution module is used to execute the door control instruction according to the maintenance decision and push prompt information to passengers.

[0079] For the control execution module, according to the maintenance decision result, send locking, activation or downgrading instructions (such as relay control, solenoid valve drive) to the door controller through the CAN bus protocol; synchronously trigger multimodal passenger guidance information (LED screen display, voice broadcast, mobile terminal push), and dynamically allocate communication resource priorities according to the fault level.

[0080] Please refer to the appendix Figure 3 , and an electronic device described below can be correspondingly referred to the intelligent door and platform screen door safety detection method described above.

[0081] The present invention also provides an electronic device, including: a processor and a memory, the memory stores a computer program executable by the processor, and when the computer program is executed by the processor, it executes the method as described above.

[0082] The present invention also provides a storage medium, on which a computer program is stored, and when the computer program is run by a processor, it executes the method as described above.

[0083] Among them, the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (abbreviation: SRAM), electrically erasable programmable read-only memory (abbreviation: EEPROM), erasable programmable read-only memory (abbreviation: EPROM), programmable read-only memory (abbreviation: PROM), read-only memory (abbreviation: ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0084] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. Smart door and platform screen door safety detection method, characterized in that, Including the following steps: Collect vibration signals, ambient temperature and humidity, and passenger flow density data of intelligent vehicle doors and platform screen doors; Conduct spectral analysis on the vibration signals, detect abnormal spectral features, and extract the time-domain amplitude of the vibration signals; Generate a dynamic vibration amplitude threshold according to the joint distribution of the ambient temperature and humidity and the passenger flow density; When the abnormal spectral features meet the preset conditions and the time-domain amplitude exceeds the dynamic vibration amplitude threshold, it is determined as abnormal; Generate a maintenance decision based on the train operation schedule, and select immediate repair or activate the standby door; Execute the door control instruction according to the maintenance decision and push a prompt message to the passengers.

2. The intelligent vehicle door and platform screen door safety detection method according to claim 1, wherein The step of collecting vibration signals, ambient temperature and humidity, and passenger flow density data of intelligent vehicle doors and platform screen doors includes: Deploy vibration sensors at the door hinges to collect vibration signals, and the sampling frequency is not less than 1 kHz; Synchronously obtain the ambient temperature and humidity data of the door area through temperature and humidity sensors; Real-time count the passenger flow density data of the door area through a lidar passenger flow counter; Perform wavelet denoising processing on the vibration signals and standardize them to a unified dimension.

3. The intelligent vehicle door and platform screen door safety detection method according to claim 1, wherein, The abnormal spectral features include at least one of the following: The proportion of the non-harmonic component energy in the total energy exceeds 10%, and the proportion of the non-harmonic component energy in the total energy is calculated through the following steps: Perform a fast Fourier transform on the vibration signal to extract the fundamental frequency and the energy of harmonic components. The calculation formula for the proportion of the energy of non-harmonic components is as follows: ； Among them, is the total energy of the vibration signal in the frequency domain, is the fundamental frequency and the sum of the energies of its integer multiple harmonic components, is the proportion of non-harmonic energy; Determine the proportion of the non-harmonic energy in the total energy; The number of mutations of the phase of the vibration signal on the Poincaré section exceeds 3 times.

4. The intelligent vehicle door and platform screen door safety detection method according to claim 1, wherein The step of extracting the time-domain amplitude of the vibration signals includes: Segment the vibration signals according to a fixed time window, and the duration of each segment is 100 ms; Calculate the time-domain amplitude of each segment of the signal, and the time-domain amplitude is the effective value or the peak value; The calculation formula for the effective value is: ； Among them, represents the amplitude of the vibration signal at the th sampling point, represents the total number of sampling points of a single-segment signal, reflects the energy intensity of the vibration signal within the window; The calculation formula for the peak value is: ； Among them, represents the absolute value of the discrete sampling points of the vibration signal, is the maximum value of the absolute values of all sampling points within the window, which characterizes the instantaneous impact intensity of the vibration signal within the window; Perform normalization processing on the time-domain amplitude and map it to the range of 0-1.

5. The intelligent vehicle door and platform screen door safety detection method according to claim 1, characterized in that, The step of generating the dynamic vibration amplitude threshold includes: Divide the temperature, humidity, and passenger flow density into preset intervals respectively, and establish a three-dimensional joint distribution data set; Train a Copula function model based on historical data to fit the joint probability distribution of temperature, humidity, and passenger flow density; Extract the upper limit of the 95% confidence interval of the joint probability distribution and convert it into a vibration amplitude threshold through linear mapping; According to the real-time collected temperature, humidity, and passenger flow density data, query the upper limit value of the corresponding confidence interval of the joint distribution as the dynamic threshold; Retrain the Copula function model every 24 hours to update the joint distribution parameters.

6. The intelligent vehicle door and platform screen door safety detection method according to claim 1, characterized in that The step of determining as abnormal includes: When the following conditions are met simultaneously, trigger abnormal determination: The proportion of the non-harmonic component energy in the abnormal spectral features exceeds 10% or the number of phase mutations on the Poincaré section exceeds 3 times; The time-domain amplitude exceeds the dynamic vibration amplitude threshold for 5 consecutive seconds.

7. The intelligent vehicle door and platform screen door safety detection method according to claim 1, characterized in that The generation of the maintenance decision includes: Obtain the arrival time of the next train and calculate the latest maintenance start time, and the calculation formula for the latest maintenance start time is: ； wherein, is the planned arrival time of the next train, is the standard time required to complete the door repair, is the safety buffer time, is the latest repair start time; If the current time exceeds , then enable the standby door.

8. An intelligent door and platform screen door safety detection system, which is applied to the intelligent door and platform screen door safety detection method according to any one of claims 1-7, characterized in that, Including: A data acquisition module for collecting vibration signals, ambient temperature and humidity, and passenger flow density data of the door area; A spectrum analysis module, which is used to perform spectrum analysis on the vibration signal to detect abnormal spectrum features and extract the time-domain amplitude of the vibration signal; A dynamic threshold generation module, which is used to generate a dynamic vibration amplitude threshold according to the joint distribution of environmental temperature and humidity and passenger flow density; An abnormality determination module, which is used to determine an abnormality when the abnormal spectrum feature meets a preset condition and the time-domain amplitude exceeds the dynamic vibration amplitude threshold; A maintenance decision-making module, which is used to generate a maintenance decision based on the train operation schedule and select immediate repair or enable a spare door; A control execution module, which is used to execute a door control instruction according to the maintenance decision and push a prompt message to passengers.

9. An electronic device, characterized in that, It includes a memory and a processor. The memory stores a computer program executable by the processor. When the processor executes the computer program, it implements the intelligent door and platform screen door safety detection method according to any one of claims 1-7.

10. A storage medium, characterized in that, A computer program is stored, and when the computer program is executed by a processor, it implements the intelligent door and platform screen door safety detection method according to any one of claims 1-7.

Images