Adaptive alarm monitoring and alarm control system for rail transit regional security

Abstract

The present application relates to the field of rail transit safety protection technology, and discloses a kind of adaptive alarm monitoring and alarm control system of rail transit area security, comprising: multi-source interference variable acquisition unit, for obtaining the non-invasive environment state data of station in-state, passenger flow pulse counting and electromechanical equipment start-stop state characterized by train in and out station state;Dynamic threshold operation unit is configured to linearly weight modulation static reference threshold according to environment state data, generate real-time dynamic decision threshold;Alarm decision logic unit is used to generate alarm instruction when original detection signal strength value continuously exceeds dynamic decision threshold, the present application introduces train, passenger flow and electromechanical three-dimensional interference factor to dynamically compensate background noise reference, effectively solve the contradiction of high false alarm rate and weak signal missing report of traditional fixed threshold scheme under strong interference working condition, realize accurate alarm identification and adaptive gating under complex dynamic environment.

Description

Technical Field

[0001] This invention belongs to the field of rail transit safety and security technology, and in particular relates to an adaptive alarm monitoring and control system for rail transit area security. Background Technology

[0002] In the current rail transit operation system, the security of stations and track areas is a fundamental link in ensuring train operation safety and the safety of passengers' lives and property. Existing rail transit area security systems typically consist of various physical detectors deployed at the front end, such as infrared beam detectors, fiber optic grating vibration cables, and video motion detection units. These detectors detect potential illegal intrusions based on changes in specific physical quantities. In conventional system designs, the alarm triggering logic mainly relies on preset static signal thresholds. When the strength of the analog signal collected by the sensor exceeds the factory setting or the fixed reference level value during the debugging phase, the system determines that an intrusion event has occurred and sends an alarm signal to the control center. This monitoring mechanism based on static thresholds can maintain a high detection accuracy in relatively stable enclosed environments.

[0003] However, rail transit stations are typical non-steady-state environments with strong physical interference. Trains entering and leaving the station cause severe transient fluctuations in air pressure in the platform area, low-frequency mechanical vibrations of the building structure, and periodic abrupt changes in the electromagnetic field strength of the overhead contact line. These physical field fluctuations caused by normal train operations often directly impact the sensitive units of security detectors, causing a sudden and significant increase in their output background noise level. To address these issues, existing technologies attempt to improve the intelligence level of security systems through system integration and high-level algorithm analysis, aiming to avoid the limitations of single physical detection methods. For example… Chinese invention patent application CN115359622A discloses an artificial intelligence-based rail transit security integrated system and method. By integrating a non-contact security inspection system, a video algorithm platform, and a security integration platform, it uses millimeter-wave imaging and video AI analysis technology to identify and alarm passenger behavior. Under the existing technology system, in order to suppress such false alarms caused by operating conditions, engineering practice is often forced to adopt compromise strategies such as time-based shielding or crude desensitization. This involves reducing the sensitivity of detectors during peak operating hours or forcibly shutting down the alarm function of specific zones during train arrival at stations through administrative orders.

[0004] Therefore, the technical problem to be solved by this invention is how to establish a system that can sense changes in the physical field of the environment in real time and dynamically adjust the signal decision logic accordingly, so as to effectively eliminate false alarms caused by deterministic physical interference without sacrificing detection coverage. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: An adaptive alarm monitoring and control system for security in rail transit areas, comprising:

[0006] The multi-source interference parameter acquisition unit is used to connect to the rail transit integrated monitoring system and synchronously collect environmental status data within the station area. The environmental status data includes: train position status signal representing the train's entry and exit sequence in the track area, passenger flow pulse count data representing the passage rate of the entry and exit gates, and electromechanical equipment switch feedback signals representing the start and stop status of the station's fans and platform screen doors.

[0007] The front-end detection signal interface unit is used to connect to security detectors deployed in the station area and read the original detection signal strength value output by the security detectors in real time.

[0008] The dynamic threshold calculation unit is used to execute linear weighted modulation logic based on environmental noise reference. This unit is configured to call preset train vibration, passenger flow thermal radiation and electromechanical electromagnetic interference coefficients respectively, normalize and weight the train position status bit signal, passenger flow pulse count data and electromechanical equipment switch feedback signal, and use the weighted sum as a gain variable to apply to the static reference threshold of the security detector to generate a real-time dynamic decision threshold that floats positively correlated with the intensity of environmental state data.

[0009] The alarm judgment logic unit is used to execute the time-domain differential judgment logic of the signal amplitude. This unit is configured to calculate the difference between the original detection signal strength value and the real-time dynamic judgment threshold in real time. When the difference is continuously positive and the duration exceeds the preset de-jitter time window, a valid alarm trigger command of the corresponding level is generated according to the amplitude range to which the absolute value of the difference belongs. The levels are quantified and divided into the following categories from high to low: emergency alarm (level 1), important alarm (level 2), general alarm (level 3), and non-critical alarm (level 4).

[0010] The alarm linkage execution unit is used to respond to valid alarm trigger commands and drive the associated video surveillance equipment to perform preset position focusing actions or drive the on-site audible and visual alarms to sound actions according to the pre-stored device mapping relationship.

[0011] Preferably, the dynamic threshold calculation unit is configured to calculate the real-time dynamic decision threshold according to the following formula: ,in, For real-time dynamic judgment thresholds; The static reference threshold for the security detector under standard operating conditions without interference; α, β, and γ are the preset train vibration influence factor, pyroelectric interference influence factor, and electromagnetic interference influence factor, respectively. The normalized state value of the train position status bit signal. The density value is the normalized value of the passenger flow pulse count data. The normalized state value of the switch feedback signal of the electromechanical equipment is used. If the physical detection principle of the security detector is not related to specific interference parameters, the corresponding influence factors α, β or γ are set to 0 during the parameter initialization stage to eliminate the contribution of unrelated independent variables to the dynamic decision threshold and improve the robustness of the system calculation.

[0012] Preferably, the multi-source interference parameter acquisition unit includes a train signal parsing subunit; the train signal parsing subunit is configured to parse the train operation message data packets of the automatic train monitoring system; when the parsed status bit indicates that the train is in the station entry state or the platform is occupied, the train position status bit signal is locked to a high level state to drive the dynamic threshold calculation unit to linearly increase the real-time dynamic decision threshold; when the parsed status bit indicates that the interval is idle, the train position status bit signal is reset to a low level state.

[0013] Preferably, the multi-source interference parameter acquisition unit includes a passenger flow density quantization subunit; the passenger flow density quantization subunit is configured to perform sliding window integration on the passenger flow pulse count data to calculate the cumulative number of people passing through per unit time; and based on a preset density-noise linear mapping table, convert the cumulative number of people passing through into the corresponding background thermal noise compensation coefficient, which is used as the input parameter of the dynamic threshold calculation unit to offset the pyroelectric background noise fluctuations caused by crowd movement.

[0014] Preferably, the alarm judgment logic unit includes a signal dejitter processing subunit; the signal dejitter processing subunit is configured to start a millisecond-level timer when the original detection signal strength value is detected to exceed the real-time dynamic judgment threshold for the first time; if the original detection signal strength value falls back and is lower than the real-time dynamic judgment threshold within the dejitter time window, a reset command is sent to the timer, and the current over-limit behavior is determined to be invalid pulse interference, and no valid alarm trigger command is generated.

[0015] Preferably, the system also includes a breakpoint resume buffer unit; the breakpoint resume buffer unit is configured to periodically monitor the network connection status between the system and the upper-level line center management platform; when a network connection interruption is detected, a first-in-first-out message queue is established in the local non-volatile memory, and the valid alarm trigger command and its generated timestamp are serialized and written into the message queue; when the network connection is detected to be restored, the data in the message queue is read in batches according to the timestamp order and sent to the line center management platform.

[0016] Preferably, the alarm linkage execution unit includes a timing strategy loading subunit; the timing strategy loading subunit internally stores two sets of linkage logic tables corresponding to the operating mode and non-operating mode respectively; this subunit is configured to read the current system clock, and when the system clock is within the operating period range, load the operating mode linkage logic table and drive the associated video surveillance equipment to perform the screen pop-up action; when the system clock is within the non-operating period range, load the non-operating mode linkage logic table and drive the on-site audible and visual alarm to perform the sounding action.

[0017] Preferably, the alarm linkage execution unit includes a pan-tilt attitude control subunit; this subunit is configured to store a one-to-one correspondence between the physical position coordinates of the security detector and the preset position number of the camera; after receiving a valid alarm trigger command, it retrieves the preset position number corresponding to the alarm source position and sends a call command containing the preset position number to the camera pan-tilt controller, driving the camera to mechanically rotate to the alarm coordinate area.

[0018] Preferably, the system further includes a device fault shielding unit; the device fault shielding unit is configured to poll the heartbeat signal of the security detector at a preset frequency; when no heartbeat signal is received within three consecutive polling cycles, or when the received device status word indicates a hardware fault, the input channel of the security detector is marked as a bypass state, and the signal from the channel is forcibly set to zero in the dynamic threshold calculation unit until a fault recovery signal is received.

[0019] Preferably, the multi-source interference parameter acquisition unit is also configured to monitor the starting inrush current signal of the large wind turbine; the dynamic threshold calculation unit is also configured to: within a time interval of 500ms to 2000ms after receiving the starting inrush current signal, superimpose a transient pulse suppression component on the real-time dynamic decision threshold to filter out the electromagnetic pulse interference generated at the moment of motor start-up.

[0020] Compared with existing technologies, the adaptive alarm monitoring and control system for security in rail transit areas of this invention has the following advantages:

[0021] 1. In rail transit area security, dynamic threshold modulation based on multi-dimensional physical condition vectors achieves real-time decoupling between alarm decision logic and unsteady environmental fields. This invention establishes a linear mapping relationship between alarm triggering benchmarks and specific physical conditions of rail transit stations through a dynamic threshold modulation unit. The system collects in real time the train operation state component representing the vibration field of train operation and the passenger flow density component representing the thermal radiation field of passenger flow, and maps them into a normalized environmental interference state vector. When receiving the original analog signal from the front-end detector, the system performs real-time weighted calculation on the static preset threshold based on the vector to generate a dynamic decision threshold that fluctuates with the intensity of environmental interference. This enables the alarm controller to automatically raise the signal decision threshold during deterministic physical interference such as airflow impact or ground vibration caused by train entering the station, and restore high-sensitivity detection during the quiet period between train operations. This processing method eliminates technical false alarms caused by the inability of static thresholds to adapt to dynamically changing physical fields from the underlying logic of signal processing, ensuring effective identification of real intrusion signals under strong interference conditions.

[0022] 2. A self-interpretive alarm data structure incorporating physical context was constructed, enhancing the traceability of security evidence chains. This invention changes the traditional alarm system's single data mode of only recording the triggering event itself. When generating alarm data frames, the environmental interference state vector at the triggering moment is encapsulated simultaneously. The system uses the train's station status, passenger flow density level, and electromechanical equipment operation status at the time of the alarm as metadata, rigidly binding it with the alarm signal and its associated media images. This allows the backend management platform or reviewers to read the physical environment context at the moment of alarm triggering when reviewing alarm events, thereby determining whether the alarm originated from human intrusion or signal drift under special conditions through objective data. This mechanism provides quantifiable physical evidence for false alarm analysis and secondary calibration of threshold parameters in security systems, enhancing the legal validity and technical traceability of alarm data.

[0023] 3. By utilizing differential comparison and signal gating mechanisms at the edge, the effective information entropy density of the transmission channel in weak network environments is improved. This invention sets up a pre-positioned signal cleaning logic in the station-level security platform. By performing differential comparison between the original signal strength and the dynamic decision threshold locally, environmental noise signals that have not exceeded the dynamic threshold are directly clamped to the station. Only high-confidence alarm data that has been verified by physical validity are uploaded to the line or network center. Under the objective conditions of limited bandwidth resources or packet loss risk in the dedicated transmission network of rail transit, this mechanism filters out a large amount of invalid oscillation data caused by environmental fluctuations, ensuring that the limited transmission bandwidth is only occupied by security business data. By reducing the transmission and processing of redundant data, the system reduces the concurrent load of the central server and ensures the real-time alarm response under sudden high interference conditions at multiple stations. Attached Figure Description

[0024] Figure 1 This is a logical structure and module connection diagram of the adaptive alarm monitoring system of the present invention;

[0025] Figure 2 This is an interactive flowchart of the dynamic threshold generation and alarm linkage control of the present invention. Detailed Implementation

[0026] The technical solutions in the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0027] An adaptive alarm and security monitoring system for rail transit areas includes:

[0028] The multi-source interference parameter acquisition unit is used to connect to the rail transit integrated monitoring system and synchronously collect environmental status data within the station area. The environmental status data includes: train position status signal representing the train's entry and exit sequence in the track area, passenger flow pulse count data representing the passage rate of the entry and exit gates, and electromechanical equipment switch feedback signals representing the start and stop status of the station's fans and platform screen doors.

[0029] The front-end detection signal interface unit is used to connect to security detectors deployed in the station area and read the original detection signal strength value output by the security detectors in real time.

[0030] The dynamic threshold calculation unit is used to execute linear weighted modulation logic based on environmental noise reference. This unit is configured to call preset train vibration, passenger flow thermal radiation and electromechanical electromagnetic interference coefficients respectively, normalize and weight the train position status bit signal, passenger flow pulse count data and electromechanical equipment switch feedback signal, and use the weighted sum as a gain variable to apply to the static reference threshold of the security detector to generate a real-time dynamic decision threshold that floats positively correlated with the intensity of environmental state data.

[0031] The alarm decision logic unit is used to execute the time-domain differential decision logic of the signal amplitude. This unit is configured to calculate the difference between the original detection signal strength value and the real-time dynamic decision threshold in real time. When the difference is continuously positive and the duration exceeds the preset de-jitter time window, a valid alarm trigger command is generated.

[0032] The alarm linkage execution unit is used to respond to valid alarm trigger commands. Based on the pre-stored device mapping relationship, it drives the associated video surveillance devices to perform video linkage responses, including capturing and saving alarm context video, capturing alarm images, and triggering emergency plans, or drives the on-site audible and visual alarms to sound.

[0033] Preferably, the dynamic threshold calculation unit is configured to calculate the real-time dynamic decision threshold according to the following formula: ,in, For real-time dynamic judgment thresholds; The static reference threshold for the security detector under standard operating conditions without interference; α, β, and γ are the preset train vibration influence factor, pyroelectric interference influence factor, and electromagnetic interference influence factor, respectively. The normalized state value of the train position status bit signal. The density value is the normalized value of the passenger flow pulse count data. This refers to the normalized state value of the switch feedback signal from electromechanical equipment.

[0034] Preferably, the multi-source interference parameter acquisition unit includes a train signal parsing subunit; the train signal parsing subunit is configured to parse the train operation message data packets of the automatic train monitoring system; when the parsed status bit indicates that the train is in the station entry state or the platform is occupied, the train position status bit signal is locked to a high level state to drive the dynamic threshold calculation unit to linearly increase the real-time dynamic decision threshold; when the parsed status bit indicates that the interval is idle, the train position status bit signal is reset to a low level state.

[0035] Preferably, the multi-source interference parameter acquisition unit includes a passenger flow density quantization subunit; the passenger flow density quantization subunit is configured to perform sliding window integration on the passenger flow pulse count data to calculate the cumulative number of people passing through per unit time; and based on a preset density-noise linear mapping table, convert the cumulative number of people passing through into the corresponding background thermal noise compensation coefficient, which is used as the input parameter of the dynamic threshold calculation unit to offset the pyroelectric background noise fluctuations caused by crowd movement.

[0036] Preferably, the alarm judgment logic unit includes a signal dejitter processing subunit; the signal dejitter processing subunit is configured to start a millisecond-level timer when the original detection signal strength value is detected to exceed the real-time dynamic judgment threshold for the first time; if the original detection signal strength value falls back and is lower than the real-time dynamic judgment threshold within the dejitter time window, a reset command is sent to the timer, and the current over-limit behavior is determined to be invalid pulse interference, and no valid alarm trigger command is generated.

[0037] Preferably, the system also includes a breakpoint resume buffer unit; the breakpoint resume buffer unit is configured to periodically monitor the network connection status between the system and the upper-level line center management platform; when a network connection interruption is detected, a first-in-first-out message queue is established in the local non-volatile memory, and the valid alarm trigger command and its generated timestamp are serialized and written into the message queue; when the network connection is detected to be restored, the data in the message queue is read in batches according to the timestamp order and sent to the line center management platform.

[0038] Preferably, the alarm linkage execution unit includes a timing strategy loading subunit; the timing strategy loading subunit internally stores two sets of linkage logic tables corresponding to the operating mode and non-operating mode respectively; this subunit is configured to read the current system clock, and when the system clock is within the operating period range, load the operating mode linkage logic table and drive the associated video surveillance equipment to perform the screen pop-up action; when the system clock is within the non-operating period range, load the non-operating mode linkage logic table and drive the on-site audible and visual alarm to perform the sounding action.

[0039] Preferably, the alarm linkage execution unit includes a pan-tilt attitude control subunit; this subunit is configured to store a one-to-one correspondence between the physical position coordinates of the security detector and the preset position number of the camera; after receiving a valid alarm trigger command, it retrieves the preset position number corresponding to the alarm source position and sends a call command containing the preset position number to the camera pan-tilt controller, driving the camera to mechanically rotate to the alarm coordinate area.

[0040] Preferably, the system further includes a device fault shielding unit; the device fault shielding unit is configured to poll the heartbeat signal of the security detector at a preset frequency; when no heartbeat signal is received within three consecutive polling cycles, or when the received device status word indicates a hardware fault, the input channel of the security detector is marked as a bypass state, and the signal from the channel is forcibly set to zero in the dynamic threshold calculation unit until a fault recovery signal is received.

[0041] Preferably, the multi-source interference parameter acquisition unit is also configured to monitor the starting inrush current signal of the large wind turbine; the dynamic threshold calculation unit is also configured to: within a time interval of 500ms to 2000ms after receiving the starting inrush current signal, superimpose a transient pulse suppression component on the real-time dynamic decision threshold to filter out the electromagnetic pulse interference generated at the moment of motor start-up.

[0042] Example 1: This example describes the specific configuration and signal processing of an adaptive alarm monitoring and control system for security in rail transit areas under strong composite interference conditions. This system addresses the contradiction between false alarms and missed alarms faced by fixed threshold alarm mechanisms when train entry vibrations and high-density passenger flow heat radiation coexist. The system connects to the rail transit integrated monitoring system via a hard-wired interface, sets the sampling period of the multi-source interference parameter acquisition unit to 100ms, and synchronously reads environmental status data within the station. During parameter initialization, a static reference threshold is set for the security detector under interference-free standard operating conditions. The voltage is 1.5V. Based on the calibration results of the on-site physical environment, the preset train vibration influence factor α is 0.6, the pyroelectric interference influence factor β is 0.3, and the electromagnetic interference influence factor γ is 0.2. The system operates under superimposed interference conditions where trains are entering the station and passenger flow is dense. At this time, the train signal analysis subunit analyzes the train operation message data packets of the automatic train monitoring system, identifies that the train in the upward direction is in the station braking stage, and locks the train position status bit signal to a high level. The corresponding normalized state value of the train position status bit signal is... The value is set to 1.0. Simultaneously, the passenger flow density quantization subunit performs sliding window integration on the passenger flow pulse count data from the entrance and exit gates. If the cumulative number of passengers passing through per unit time is in the high range, based on a preset density-noise linear mapping table, it outputs the normalized density value of the passenger flow pulse count data. The value is 0.8. At this time, the electromechanical equipment in the station is operating smoothly, and no large fan starting inrush current is detected. The normalized state value of the electromechanical equipment switch feedback signal is... Maintain at the baseline level of 0.1.

[0043] The dynamic threshold calculation unit receives the above real-time environment vector and calculates it according to the formula. By executing the linear weighted modulation logic and substituting the aforementioned real-time parameters into the calculation, the real-time dynamic decision threshold is obtained. It equals the sum of 1.5 multiplied by 1, 0.6 multiplied by 1.0, 0.3 multiplied by 0.8, and 0.2 multiplied by 0.1, that is... The calculation results show that the system has automatically raised the alarm trigger threshold to adapt to the current strong interference physical field. During the period when this dynamic threshold is in effect, the front-end detection signal interface unit reads the original detection signal strength value output by the security detectors deployed at the perimeter of the platform. The original detection signal strength value was affected by ground vibrations and airflow impacts caused by the train entering the station. When the voltage fluctuates to 2.4V, the alarm judgment logic unit calculates the original detection signal strength value in real time. With real-time dynamic decision threshold The difference, 2.4V minus 2.79V, is negative. The system determines that the current signal exceeding the limit is due to environmental background noise, performs signal suppression, and does not generate a valid alarm trigger command. This processing logic filters out interference from the same source signal caused by the vibration of the train entering the station at the physical level. A physical intrusion event occurs during the train's entry into the station, causing the original detection signal strength value output by the detector to change. The voltage surges to 3.5V. The alarm decision logic unit recalculates the difference, which is 0.71V. This difference is positive, triggering the signal debounce processing subunit to start a millisecond-level timer. Within the preset 500ms debounce time window, if the difference remains positive and does not decrease, the system confirms a valid intrusion. The alarm decision logic unit generates a valid alarm trigger command and encapsulates the trigger time. parameter, Parameters and The parameters serve as metadata for the operating context. The alarm linkage execution unit responds to the command, retrieves the pre-stored device mapping relationship, and drives the associated video surveillance device to perform the preset position focusing action. This process verifies that while the system dynamically raises the decision threshold to suppress false alarms, it still retains the ability to respond to high-intensity real intrusion signals.

[0044] Example 2: This example details the testing process on a simulated high-interference underground rail transit station security monitoring experimental platform. It verifies the false alarm suppression capability and intrusion detection reliability of the adaptive alarm monitoring and control system proposed in this invention under complex physical field interference. The experimental platform consists of a programmable vibration table, a thermal infrared radiation array, and an electromagnetic pulse generator, used to reproduce ground vibrations caused by train entering the station, pyroelectric background noise generated by high-density passenger flow, and electromagnetic surges during the start-up and shutdown of large electromechanical equipment, respectively. The experimental data acquisition system uses a high-precision multi-channel oscilloscope recorder with a sampling frequency set to 10kHz to ensure the complete waveform characteristics of the sensor output signal are captured. To simulate signal transmission loss and aging noise in a real industrial environment, Gaussian white noise with a signal-to-noise ratio of 18dB is superimposed at the input of the front-end detection signal interface unit through a hardware noise-adding circuit. During the test preparation phase, the system parameter calibration procedure is executed to determine the core operational parameters. The environment is placed in an interference-free silent state, and the peak background noise output of the security detector is measured to be 0.45V. Based on the principle of a 3-fold safety margin, the static reference threshold is determined. With a voltage of 1.5V, the vibration table and thermal radiation array were independently turned on to their maximum design conditions: under the vibration condition of a fully loaded train entering the station, the noise floor of the detected signal rose to 1.35V, and the train vibration influence factor was calibrated based on this. The value is 0.6; under the simulated passenger flow density of 4 people / square meter, the thermal noise rises to 0.85V, based on which the pyroelectric interference influence factor is calibrated. The value is 0.3; at the moment the simulated station fan starts, the amplitude of the induced electromotive force fluctuation is 0.25V, and based on this, the electromagnetic interference influence factor γ is calibrated to be 0.2.

[0045] The experiment was conducted in two parallel logic channels: a control group and the prototype of this invention. The control group used traditional fixed threshold decision logic, with its alarm decision threshold always locked at a static reference threshold of 1.5V. The prototype of this invention activated the multi-source interference parameter acquisition unit and the dynamic threshold calculation unit, according to the formula... The floating threshold was calculated in real time. In the first phase of the experiment, a complex and strong interference scenario of trains entering the station during the evening rush hour was simulated. The vibration table was controlled to be loaded with a variable frequency vibration waveform of 25Hz to 60Hz to simulate train braking, while the thermal radiation array was adjusted to an equivalent flow density of 0.9 (i.e., ), and trigger an electromagnetic pulse (i.e. After the system detects the rising edge of the pulse, it superimposes a bias voltage sequence with an initial value of 2.0 volts, decreasing in steps at a rate of 200 millivolts / 200 milliseconds, onto the threshold to cover the attenuation tail of electromagnetic interference. At this time, the normalized state value of the train position status bit signal output by the train signal analysis subunit is... The signal strength value jumps to 1.0. Due to the aforementioned superimposed interference, the original detection signal strength value output by the front-end security detector... It exhibits severe nonlinear fluctuations, with its peak value climbing to 2.68V.

[0046] Under this operating condition, the decision logic of the control group detected... When the voltage (2.68V) exceeds the fixed threshold (1.5V), a false alarm record is generated. In the sample of this invention, the dynamic threshold calculation unit performs calculations by inputting the current environmental parameters in real time. The calculation process executes hardware saturation protection logic in parallel: if the calculation result exceeds the system supply voltage of 5.0 volts, the threshold is forcibly clamped at 4.8 volts (i.e., 96% of the range) to prevent the operational amplifier from entering the nonlinear saturation region. The alarm judgment logic unit compares and finds that the current signal peak of 2.68V is less than the real-time dynamic judgment threshold of 2.835V, and determines that the over-limit is a background noise fluctuation. The signal clamping operation is executed, and no alarm command is triggered, which confirms the system's physical layer suppression capability against false alarms under strong interference background. In the second stage of the test, while keeping the above strong interference background unchanged, a simulated physical intrusion action is introduced. A standard test human model (weighing 75kg, surface temperature 37℃) is thrown into the detection area by a pneumatic ejection device to simulate illegal climbing behavior. At this time, the original detection signal strength value output by the detector is... A transient pulse was superimposed on a background noise base of 2.68V, causing the total amplitude to surge to 4.12V instantaneously. At this instant, the alarm decision logic unit of the present invention calculated the difference: 4.12V minus the current dynamic threshold of 2.835V, resulting in a difference of 1.285V. This positive difference exceeded the preset sensitivity dead zone and remained positive within the 500ms dejitter time window. The system then confirmed the intrusion event and generated a valid alarm trigger command. By comparison, while simply raising the fixed threshold to a high threshold of 3.0V could suppress false alarms in the first stage, this approach also proved effective. However, during periods of low interference (such as nighttime rest periods when the background noise is only 0.4V), false alarms will occur when faced with weak signal intrusions (such as amplitude 1.8V). In contrast, the sample of this invention, under nighttime operating conditions, automatically drops the threshold to 1.5V because all environmental parameters return to zero, and can capture the weak signal intrusion of 1.8V. The stress test data of 72 consecutive hours shows that in the test cycle including 120 simulated train entry and 50 simulated intrusions, the false alarm rate of the sample of this invention is 0.8% and the false alarm rate is 0%, compared with the false alarm rate of 35.2% of the control group, which achieves an order of magnitude improvement in alarm reliability.

[0047] Example 3: This example details the parameter calibration and initialization procedures for the adaptive alarm monitoring and control system for security in rail transit areas before its formal operation. It clarifies the normalization logic for input parameters. During system deployment, it executes an automatic acquisition process for static baseline thresholds. The multi-source interference parameter acquisition unit continuously collects background noise data from front-end security detectors for 30 minutes during non-operational periods—a quiet period with no trains running, no passenger flow, and non-essential electromechanical equipment shut down. The dynamic threshold calculation unit calculates the mean μ and standard deviation σ of the original detection signal strength values ​​within this time window. Based on the 3-times-standard-deviation criterion, the system automatically sets the static baseline threshold. The value is set to μ+3σ to ensure that the false alarm rate in a clean environment is below the statistical level. The system enters a univariate disturbance calibration mode to determine the specific values ​​of the train vibration impact factor α, the pyroelectric interference impact factor β, and the electromagnetic interference impact factor γ. When calibrating the train vibration impact factor, the system coordinates the automatic train monitoring system to dispatch an empty train to pass through the platform at a standard entry speed, keeping the passenger flow count at zero and the electromechanical equipment silent. The multi-source interference parameter acquisition unit records the maximum peak value of the detected signal strength during the train's entry into the station. Based on the physical definition, the system calculates the train vibration influence factor α as the ratio of the voltage increment caused by the train to the static reference value, i.e. .

[0048] During the electromechanical equipment calibration process, the system instructs the fans and shielded doors within the station to perform a full-load start-stop cycle, recording the peak electromagnetic pulse value during this process. And the electromagnetic interference influence factor γ was determined to be To address passenger flow interference characterized by continuous variation, the system executes a linear regression fitting procedure based on historical data. This involves referencing passenger flow pulse count data from the past 30 operating days and the corresponding detector background noise baseline values. The passenger flow density quantization subunit then performs normalization processing on the passenger flow data. ,in This represents the cumulative number of people passing through the current sliding window. The maximum capacity constant designed for the station, when Exceed hour With a clamping value of 1.0, the system operates at a normalized density. The noise floor increment ratio is used as the independent variable to detect the signal strength. As the dependent variable, the least squares method is used for linear fitting. Before fitting, data cleaning logic is performed: all sampling points with accompanying train vibration signals (i.e., vibration sensor voltage greater than 0.2 volts) or strong electromagnetic interference (i.e., electromagnetic sensor voltage greater than 0.5 volts) are removed, retaining only noise data under a clean passenger flow background. The slope of the fitted line is determined as the pyroelectric interference influence factor β. After completing the above calibration, the dynamic threshold calculation unit will determine the... α, β, and γ are solidified into non-volatile memory to construct a defined linear weighted modulation model. During real-time operation, if detector aging causes the mean static background noise μ to drift by more than 10%, the system will trigger a recalibration warning. This procedure transforms the abstract weighted logic into quantifiable and reproducible engineering configuration steps, ensuring the universality and accuracy of the adaptive algorithm in different physical structure station environments.

[0049] Example 4: In a cold start scenario where the system is first deployed at a newly built station, the multi-source interference parameter acquisition unit executes the default parameter loading procedure. This procedure calls the pre-set general-purpose station interference coefficient matrix in the read-only memory, setting the train vibration influence factor α to 0.5, the pyroelectric interference influence factor β to 0.4, and the electromagnetic interference influence factor γ to 0.2. The above parameter combination is determined based on the statistical average of historical operating data of stations with similar geological conditions, aiming to ensure that the system dynamically determines the threshold during the first month of operation when there is no local data accumulation. The generation logic can still maintain basic false alarm suppression capabilities, avoiding logic paralysis caused by missing parameters.

[0050] The system enters the adaptive correction phase. After every 24 hours of operation, the dynamic threshold calculation unit automatically extracts all valid alarm records and corresponding original signal waveforms for the day. If the system detects consecutive false alarms during a specific period (such as the non-operational period from 23:00 to 06:00 the next day), it will automatically lower the static baseline threshold. The system adjusts the weights and fine-tunes the electromagnetic interference influence factor γ, with each adjustment limited to 5% of the current value. This iterative correction process continues until the localized false alarm rate calculated by the system is lower than the acceptance standard of 0.1%. After that, the system locks the current set of optimal parameters as the dedicated operating configuration for the station, completing a smooth transition from general preset to local adaptation.

[0051] Example 5: This example details an adaptive parameter matrix offline generation and verification procedure applicable to various heterogeneous rail transit environments. It provides a systematic parameter initialization and optimization method for stations with different physical structures (e.g., underground, elevated, ground) and different operational intensity levels, solving the problem of uncertainty in system cold start performance caused by a lack of on-site measured data. It constructs a parameterized simulation model library containing typical station physical characteristics and interference source attributes, covering three standard station topologies: underground island, elevated side, and ground hybrid. For each topology, a set of key physical parameters is set, including platform length (12...). Based on the parameters of 0m to 180m, tunnel cross-sectional area (30㎡ to 60㎡), and acoustic reverberation time (1.5s to 3.0s), parameterized waveform generators for train vibration, passenger flow thermal radiation, and electromagnetic interference were established. The train vibration waveform was set as a random vibration signal with a frequency range of 20Hz to 80Hz and an amplitude of 0.1g to 0.5g. The passenger flow thermal radiation model was set as a time-varying thermal field distribution with a density of 0 to 5 people / ㎡ and a temperature of 36℃ to 38℃. The electromagnetic interference was set as a transient pulse sequence with a pulse width of 10ms to 100ms and an intensity of 10V / m to 50V / m.

[0052] Based on the aforementioned model library, a large-scale Monte Carlo simulation experiment was conducted to generate the optimal parameter matrix. The input variables for the simulation experiment were set as the combination of the aforementioned physical parameter set and the interference source parameter set, and the output variables were the system's false alarm rate and false negative rate. For each combination of station topology and interference intensity, the system performed a grid search in the parameter space (α∈[0.1,1.0], β∈[0.1,1.0], γ∈[0.1,0.5]). For each group (α,β,γ), 1000 random interference superposition simulations were run, and the false alarm rate and false negative rate were statistically analyzed. Based on the selection criteria of a false alarm rate <1% and a false negative rate <0.1%, the optimal parameter combination for that operating condition was selected. Finally, all the selected optimal parameter combinations were compiled into a multidimensional lookup table. This table used station type, passenger flow level, and electromechanical configuration as index keys, and (α,β,γ) as output values. The parameter adaptability verification procedure based on small sample data from the field is executed before actual engineering deployment in the initial configuration module of the system. After the system is installed at the target station, a portable signal generator is used to inject standard test signals (such as infrared and vibration composite signals simulating intrusion) at key positions on the platform. The system records the response under the default parameter settings. If a missed or false alarm is detected, the parameter fine-tuning logic is initiated: the Euclidean distance between the feature vector of the test signal and the nearest neighbor sample in the simulation model library is calculated, and the parameter combination corresponding to the nearest simulation condition is selected as the recommended correction value. This injection-verification-correction process is repeated until the system responds correctly in 10 consecutive tests. This procedure ensures that the system can quickly converge from the general simulation optimal solution to the specific field optimal solution when facing unknown actual physical environments, realizing the standardization and efficiency of engineering deployment.

[0053] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.

Claims

1. A self-adaptive alarm monitoring and alarm control system for rail transit area security, characterized in that, include: The multi-source interference parameter acquisition unit is used to connect to the rail transit integrated monitoring system and synchronously collect environmental status data within the station area; Environmental status data includes: train position status signals representing the timing of trains entering and leaving the station in the track area; passenger flow pulse count data representing the passage rate of the gates in and out of the station; and electromechanical equipment switch feedback signals representing the start and stop status of the fans and platform screen doors in the station. The front-end detection signal interface unit is used to connect to security detectors deployed in the station area and read the original detection signal strength value output by the security detectors in real time. The dynamic threshold calculation unit is used to execute linear weighted modulation logic based on environmental noise reference. This unit is configured to call preset train vibration, passenger flow thermal radiation and electromechanical electromagnetic interference coefficients respectively, normalize and weight the train position status bit signal, passenger flow pulse count data and electromechanical equipment switch feedback signal, and use the weighted sum as a gain variable to apply to the static reference threshold of the security detector to generate a real-time dynamic decision threshold that floats positively correlated with the intensity of environmental state data. The alarm decision logic unit is used to execute the time-domain differential decision logic of the signal amplitude. This unit is configured to calculate the difference between the original detection signal strength value and the real-time dynamic decision threshold in real time. When the difference is continuously positive and the duration exceeds the preset de-jitter time window, a valid alarm trigger command is generated. The alarm linkage execution unit is used to respond to valid alarm trigger commands and drive the associated video surveillance equipment to perform preset position focusing actions or drive the on-site sound and light alarm to perform sounding actions according to the pre-stored device mapping relationship. The dynamic threshold calculation unit is configured to calculate the real-time dynamic decision threshold according to the following formula: ,in, For real-time dynamic judgment thresholds; The static reference threshold for the security detector under standard operating conditions without interference; α, β, and γ are the preset train vibration influence factor, pyroelectric interference influence factor, and electromagnetic interference influence factor, respectively. The normalized state value of the train position status bit signal. The density value is the normalized value of the passenger flow pulse count data. The normalized state value of the switch feedback signal of electromechanical equipment; The multi-source interference parameter acquisition unit is also configured to monitor the starting inrush current signal of large wind turbines; the dynamic threshold calculation unit is also configured to superimpose a transient pulse suppression component on the real-time dynamic decision threshold within a time interval of 500ms to 2000ms after receiving the starting inrush current signal, so as to filter out the electromagnetic pulse interference generated at the moment of motor start-up.

2. The adaptive alarm monitoring and control system for security in rail transit areas according to claim 1, characterized in that, The multi-source interference parameter acquisition unit includes a train signal parsing subunit. The train signal parsing subunit is configured to parse the train operation message data packets of the automatic train monitoring system. When the parsed status bit indicates that the train is in the station entry state or the platform is occupied, the train position status bit signal is locked to a high level to drive the dynamic threshold calculation unit to linearly increase the real-time dynamic decision threshold. When the parsed status bit indicates that the interval is empty, the train position status bit signal is reset to a low level.

3. The adaptive alarm monitoring and control system for security in rail transit areas according to claim 1, characterized in that, The multi-source interference parameter acquisition unit includes a passenger flow density quantization subunit. The passenger flow density quantization subunit is configured to perform sliding window integration on the passenger flow pulse count data to calculate the cumulative number of people passing through per unit time. Based on a preset density-noise linear mapping table, the cumulative number of people passing through is converted into the corresponding background thermal noise compensation coefficient, which is used as the input parameter of the dynamic threshold calculation unit.

4. The adaptive alarm monitoring and control system for security in rail transit areas according to claim 1, characterized in that, The alarm judgment logic unit includes a signal dejitter processing subunit; the signal dejitter processing subunit is configured to start a millisecond-level timer when the original detection signal strength value is detected to exceed the real-time dynamic judgment threshold for the first time; If the original detection signal strength value drops below the real-time dynamic decision threshold within the dejittering time window, a reset command is sent to the timer, and the current out-of-limit behavior is determined to be invalid pulse interference, and no valid alarm trigger command is generated.

5. The adaptive alarm monitoring and control system for security in rail transit areas according to claim 1, characterized in that, The system also includes a breakpoint resume buffer unit; the breakpoint resume buffer unit is configured to periodically monitor the network connection status between the system and the upper-level line center management platform; when a network connection interruption is detected, a first-in-first-out message queue is established in the local non-volatile memory, and the valid alarm trigger command and its generated timestamp are serialized and written into the message queue; when the network connection is detected to be restored, the data in the message queue is read in batches according to the timestamp order and sent to the line center management platform.

6. The adaptive alarm monitoring and control system for security in rail transit areas according to claim 1, characterized in that, The alarm linkage execution unit includes a timing strategy loading subunit; the timing strategy loading subunit internally stores two sets of linkage logic tables corresponding to the operating mode and the non-operating mode respectively; this subunit is configured to read the current system clock, and when the system clock is within the operating time range, load the operating mode linkage logic table and drive the associated video surveillance equipment to perform the screen pop-up action; When the system clock is within the non-operational period range, the non-operational mode linkage logic table is loaded, driving the on-site audible and visual alarms to sound.

7. The adaptive alarm monitoring and control system for security in rail transit areas according to claim 1, characterized in that, The alarm linkage execution unit includes a pan-tilt attitude control subunit. This subunit is configured to store a one-to-one correspondence between the physical position coordinates of the security detector and the preset position number of the camera. After receiving a valid alarm trigger command, it retrieves the preset position number corresponding to the alarm source position and sends a call command containing the preset position number to the camera pan-tilt controller, driving the camera to mechanically rotate to the alarm coordinate area.

8. The adaptive alarm monitoring and control system for security in rail transit areas according to claim 1, characterized in that, The system also includes a device fault shielding unit; the device fault shielding unit is configured to poll the heartbeat signal of the security detector at a preset frequency; when no heartbeat signal is received within 3 consecutive polling cycles, or when the received device status word indicates a hardware fault, the input channel of the security detector is marked as bypassed, and the signal from the channel is forcibly set to zero in the dynamic threshold calculation unit until a fault recovery signal is received.

Images