A port railway gate lever state identification method, device, equipment and medium

Abstract

The application provides a port railway gate lever state recognition method, device, equipment and medium, which is used in the field of railway intelligent management technology, and includes the following steps: calculating corresponding electrical confidence value, acceleration confidence value and visual confidence value based on signal jump number and level noise amplitude, three-axis acceleration and resting offset corresponding to the gate lever, and edge pixels and adjustment coefficient, and weighting and comparing each confidence value to obtain a target confidence value; determining a corresponding target state value, and determining a corresponding real-time gate lever state according to the target state value and a corresponding preset state threshold; calculating a corresponding trajectory deviation degree according to a corresponding real-time running trajectory, and determining a corresponding gate lever running state according to the trajectory deviation degree and a preset deviation degree; determining a corresponding target gate lever state of the gate lever according to the real-time gate lever state and the gate lever running state; in this way, the anti-interference ability and recognition accuracy of the gate lever state recognition are improved, and long-term monitoring of the health state of the gate lever is realized.

Description

Technical Field

[0001] This application relates to the field of railway intelligent management technology, and in particular to a method, device, equipment and medium for identifying the status of port railway gates. Background Technology

[0002] As the core hub for the collection and distribution of port materials, port railways undertake the core operational tasks of bulk cargo transshipment and container distribution. They are a key component of the port's intelligent and automated system. Port railway gates are deployed in the open-air operation area of ​​the port for a long time. Unlike ordinary highway crossings, port railways have a complex on-site operating environment, many interference factors, and dynamic and changing operating conditions. The frequent start-stop of large cranes and reach stackers on site generates high-intensity electromagnetic interference. The near-shore area has continuous sea breeze disturbance and a high salt spray corrosion environment, which places extremely high demands on the reliability, anti-interference and robustness of the gate status recognition system.

[0003] Existing gate arm status identification methods typically employ data comparison and conflict selection verification, without establishing an evaluation system. This results in low utilization of multi-source data, poor fault tolerance, and difficulty in guaranteeing the accuracy of status identification in complex scenarios. Furthermore, existing methods can only achieve binary state determination of the gate arm being fully raised or fully lowered, leading to low detection accuracy. Moreover, due to factors such as port wind disturbances, the gate arm may experience abnormal swinging, jamming, or even deformation, causing abnormal results from existing detection systems.

[0004] Therefore, existing gate arm status recognition methods suffer from low anti-interference capabilities, low data utilization, and low recognition accuracy. Summary of the Invention

[0005] This application provides a method, device, equipment, and medium for identifying the status of port railway gate arms, in order to solve the problems of low anti-interference ability, low data utilization rate, and low identification accuracy of existing gate arm status identification methods.

[0006] Firstly, this application provides a method for identifying the status of a port railway gate, the method comprising: Based on the number of signal transitions and the amplitude of level noise, the triaxial acceleration and resting offset of the gate arm, and the edge pixels and adjustment coefficient, the corresponding electrical confidence value, acceleration confidence value and visual confidence value are calculated respectively. Then, according to their respective target weights, the confidence values ​​are weighted and compared to obtain the target confidence value. Based on the target sensor corresponding to the target confidence value, determine the corresponding target state value, and based on the target state value and the corresponding preset state threshold, determine the corresponding real-time gate state. Based on the corresponding real-time running trajectory, calculate the corresponding trajectory deviation, and determine the corresponding gate arm operating status based on the trajectory deviation and the preset deviation. Based on the real-time gate arm status and gate arm operation status, determine the target gate arm status corresponding to the gate arm.

[0007] In some embodiments of this application, based on the number of signal transitions and the amplitude of level noise, the triaxial acceleration and resting offset corresponding to the gate arm, and the edge pixels and adjustment coefficients, the corresponding electrical confidence values, acceleration confidence values, and visual confidence values ​​are calculated, including: Based on their respective preset weights, the number of signal transitions and the amplitude of level noise are weighted and summed to obtain the first sum, and the variance of the triaxial acceleration and the resting offset are weighted and summed to obtain the second sum. The difference between the standard value and the first sum, as well as the difference between the standard value and the first sum, is calculated to obtain the electrical confidence value and the acceleration confidence value. The difference between the standard value and the multiplication value is calculated based on the gradient variance of the edge pixel and the adjustment coefficient. The visual confidence value is obtained by multiplying the model confidence value output by the difference in the target visual model output.

[0008] In some embodiments of this application, each confidence value is weighted and compared according to its corresponding target weight to obtain a target confidence value, including: Based on the corresponding environmental sensors, it is determined whether the current scene falls within a preset scene category; the preset scene categories include electromagnetic interference scene, acceleration interference scene, and visual interference scene. If it is any of the preset scenario categories, then based on the preset scenario category, determine the corresponding scenario confidence value and other standard confidence values, and determine the preset weight corresponding to the scenario confidence value as the weight to be reduced and the preset weight corresponding to the standard confidence value as the weight to be increased. Calculate the product of the weight to be reduced and the preset reduction coefficient, and the product of the weight to be increased and the preset increase coefficient, respectively, to obtain the target reduction weight corresponding to the scenario confidence value and the target increase weight corresponding to the standard confidence value. Based on the target reduction weight and the target increase weight, weight each confidence value to obtain multiple weighted confidence values. If it is not a preset scenario category, then the preset weight corresponding to each confidence value is determined as the target weight, and based on the target weight, each confidence value is weighted to obtain multiple weighted confidence values; By comparing the values ​​of each weighted confidence value, the target confidence value with the largest value is obtained.

[0009] In some embodiments of this application, the corresponding target state value is determined based on the target sensor corresponding to the target confidence value, including: Determine the type of the target sensor; If the target sensor is an electrical sensor, the target state value is determined based on the contact level corresponding to the gate arm; If the target sensor is an accelerometer, the corresponding gate tilt angle is calculated based on the real-time collected triaxial acceleration components to obtain the target state value. If the target sensor is a vision sensor, the corresponding target state value is determined based on the target detection angle output by the target vision model.

[0010] In some embodiments of this application, the corresponding real-time gate state is determined based on the target state value and the corresponding preset state threshold, including: When the target state value is the contact level, determine the level category corresponding to the contact level; If the level category is preset high level, then the real-time gate arm status is determined to be gate arm closed; If the level category is preset low level, then the real-time gate arm status is determined to be gate arm raised; When the target state value is the gate arm tilt angle or the target detection angle, the preset state threshold is determined to be the preset angle threshold, and it is determined whether the corresponding angle is higher than the corresponding preset angle threshold. If it is higher, then the real-time gate arm status is determined to be gate arm raised; If it is not higher than the value, then the real-time gate arm status is determined to be gate arm closed.

[0011] In some embodiments of this application, the corresponding gate arm operating state is determined based on the trajectory deviation and a preset deviation, including: For the trajectory deviation corresponding to each collection cycle, determine whether the trajectory deviation is greater than the preset deviation. If the deviation of all trajectories is greater than the preset deviation, the corresponding gate arm operation state is determined to be that the gate arm has sensor zero-point drift or mechanical micro-deformation; If the trajectory deviation is greater than the preset deviation only for the last collection cycle, then the corresponding gate arm operation status is determined to be that the gate arm has a transmission mechanism jamming or insufficient motor power fault. If the deviation of some continuous trajectories is greater than the preset deviation, the corresponding gate arm operating state is determined to be non-command swing caused by external disturbance. If the deviation of all trajectories is not greater than the preset deviation, then the corresponding gate arm operation status is determined to be that the gate arm does not have any operational abnormalities.

[0012] In some embodiments of this application, after determining the target gate arm state corresponding to the gate arm based on the real-time gate arm state and the gate arm operating state, the method further includes: Based on the preset detection cycle, the opening and closing time deviation, angular velocity attenuation, jitter frequency anomaly and trajectory deviation of the gate arm are determined. According to their respective weights, the opening and closing time deviation, angular velocity attenuation, jitter frequency anomaly and trajectory deviation are weighted and summed to obtain the corresponding sum value. Calculate the difference between the standard value and the sum value to obtain the corresponding health value, and compare the health value with multiple preset health thresholds to obtain the corresponding health status; If the health status is slightly deteriorated, then the corresponding periodic test result of the gate arm is determined to be that there may be slight bearing wear or increased transmission clearance. If the health status is severely degraded, the corresponding periodic test result of the gate arm is determined to be that there may be motor aging, mechanism jamming, or component corrosion failure. Based on the periodic detection results, the corresponding prompt message is output.

[0013] Secondly, this application provides a status identification device for a port railway gate, the device comprising: The comparison module is used to calculate the corresponding electrical confidence value, acceleration confidence value and visual confidence value based on the number of signal transitions and level noise amplitude, the triaxial acceleration and resting offset of the gate arm, and the edge pixels and adjustment coefficients, respectively. Then, according to their respective target weights, the confidence values ​​are weighted and compared to obtain the target confidence value. The target state value determination module is used to determine the corresponding target state value based on the target sensor corresponding to the target confidence value, and to determine the corresponding real-time gate state based on the target state value and the corresponding preset state threshold. The calculation module is used to calculate the corresponding trajectory deviation based on the corresponding real-time running trajectory, and determine the corresponding gate arm operating status based on the trajectory deviation and the preset deviation. The gate arm status determination module is used to determine the target gate arm status corresponding to the gate arm based on the real-time gate arm status and gate arm operation status.

[0014] Thirdly, this application provides a computer device, including: a processor, and a memory communicatively connected to the processor; The memory stores instructions that the computer executes; The processor executes computer execution instructions stored in memory to implement the method of this application.

[0015] Fourthly, this application provides a computer-readable storage medium storing program code, which, when executed by a processor, is used to implement the method of this application.

[0016] Compared with existing technologies, the method in this application solves the problem of low accuracy of traditional fixed weight and simple comparison by fusing dynamic credibility from three sources. It realizes the real-time adjustment of the credibility weight of each dimension of data based on environmental parameters such as illumination, rain, fog, vibration, and electromagnetic fields, which improves the accuracy of state recognition under complex working conditions and solves the defects of weak anti-interference and high false judgment rate. By comparing the deviation characteristics of the actual trajectory with the standard trajectory in real time, it can accurately distinguish subtle hidden faults such as external disturbances, mechanical deformation, transmission jamming, and sensor drift, which solves the problem of traditional single-point parameter detection and recognition having a single dimension and being unable to capture progressive anomalies. Based on multi-dimensional parameters such as long-term opening and closing time, operating speed, and vibration frequency, the health of the gate arm can be evaluated, which can predict the mechanical deterioration trend such as motor aging, bearing wear, and mechanism jamming in advance, realizing the transformation from fault alarm to fault prediction and improving the intelligent operation and maintenance level of port unmanned equipment. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] Figure 1 A flowchart illustrating a method for identifying the state of a port railway gate arm provided in this application embodiment; Figure 2 A schematic diagram of a method for identifying the state of a port railway gate arm provided in this application embodiment; Figure 3 A schematic diagram of the framework for a port railway gate state recognition method provided in this application embodiment; Figure 4 A schematic diagram of the structure of a port railway gate status identification device provided in this application embodiment; Figure 5 This is a structural block diagram of an apparatus for performing a state identification method for a port railway gate according to an embodiment of this application. Detailed Implementation

[0019] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0020] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0021] Figure 1 This is a flowchart illustrating a method for identifying the state of a port railway gate as provided in an embodiment of this application. Figure 1 As shown, the state identification method for port railway gates may include the following steps: S110. Based on the number of signal transitions and the amplitude of level noise, the triaxial acceleration and resting offset of the gate arm, and the edge pixels and adjustment coefficient, calculate the corresponding electrical confidence value, acceleration confidence value and visual confidence value, respectively. Then, according to their respective target weights, perform weighted comparison of each confidence value to obtain the target confidence value.

[0022] The signal transition count refers to the number of unexpected transitions between high and low levels of the dry contact signal within a unit of time (usually 1 second). The larger the value, the more severe the impact of electromagnetic interference, vibration, or oxidation on the electrical contacts, resulting in frequent false signal flips and poor circuit stability. The smaller the value, the better the electrical circuit contact and the less significant the interference.

[0023] Level noise amplitude is the voltage fluctuation amplitude of a dry contact signal during a stable level period, which can be obtained through ADC sampling. The larger the value, the stronger the electromagnetic interference (such as the start-up and shutdown of a large port crane), or the more severe the signal distortion caused by aging lines or poor grounding. The smaller the value, the cleaner the electrical signal and the more stable the voltage.

[0024] Triaxial acceleration refers to the instantaneous acceleration components in the X, Y, and Z spatial directions collected in real time by the triaxial accelerometer in the middle section of the gate arm, with the unit being m / s².

[0025] The resting offset is the combined offset of the three-axis components of the accelerometer output from the factory-calibrated zero point when the gate arm is in a static state (0° locked position or 90° locked position), and the residual value after salt spray corrosion correction. The larger this value, the more significant the zero point of the accelerometer has shifted due to salt spray corrosion, temperature drift, or mechanical deformation, and the less reliable the static reference is; the smaller the value, the more stable the sensor zero point and the more accurate the reference is.

[0026] Edge pixel is the gradient variance of edge pixels, which is the gradient variance value of the gate arm edge pixels. It reflects the sharpness and clarity of the gate arm edge in the image. The larger the value, the sharper the gate arm edge, the clearer the image, and the higher the image quality. The smaller the value, the blurrier the image (caused by rain, fog, dirty lens, insufficient light, backlight at night, etc.) and the blurry edges.

[0027] The adjustment coefficient is a fixed empirical constant (taken as 0.02), which is a penalty factor used to map the gradient variance of edge pixels to the interval [0,1].

[0028] The electrical confidence value is a quantitative value used to quantify the reliability of the electrical dry contact acquisition channel at the current moment. The larger the electrical confidence value (closer to 1), the cleaner, more stable and reliable the electrical signal is, and the electrical decision result can be adopted with high weight. The smaller the electrical confidence value (closer to 0), the more severe the interference of the electrical signal is, and its weight in the fusion will be reduced.

[0029] Acceleration confidence value is a quantitative value used to characterize the real-time reliability of the three-axis acceleration sensing channel. The closer the acceleration confidence value is to 1, the smoother the gate arm operation, the no zero-point drift of the sensor, and the accurate attitude angle calculation results. The closer the acceleration confidence value is to 0, the more severe the vibration, the more serious the sensor drift, and the more distorted and unusable the acceleration angle data.

[0030] The visual confidence value is a quantitative value used to characterize the reliability of visual camera channel recognition. The closer the visual confidence value is to 1, the clearer the image, the unobstructed gate detection, and the accurate visual recognition angle. The closer the visual confidence value is to 0, the blurry image caused by heavy fog, rainstorm, or backlighting, and the unreliable visual recognition result.

[0031] The target weight refers to the weighting coefficients dynamically assigned to the electrical confidence value, acceleration confidence value, and visual confidence value based on real-time environmental parameters such as illumination, rainfall, vibration intensity, and electromagnetic intensity. In practical applications, the visual imaging quality deteriorates in rainy, foggy, or nighttime scenarios, so the system automatically reduces the visual weight. When strong vibrations are generated by dock cranes or vehicles passing by, the acceleration weight is reduced, and electrical and visual data are prioritized. When strong electromagnetic interference is generated by nearby crane operations, the probability of electrical signal distortion increases, so the electrical weight is reduced, and acceleration and visual data are used as the core judgment criteria. This ensures that the data with the best current quality has a higher weight, reflecting the system's adaptability to environmental changes.

[0032] The target confidence value is the highest value among the values ​​calculated by weighting each confidence value with its corresponding weight.

[0033] Based on this, by collecting the on / off signals of the electrical dry contacts of the gate arm, the acceleration data of the three-axis accelerometer installed on the gate arm in the XYZ directions, and the camera image data of the gate arm rotation angle, independent credibility evaluation factors are established for the three types of perception data. The weights of each dimension are adjusted in combination with the real-time environmental conditions of the port to achieve the optimal data fusion effect under different scenarios and avoid the defects of insufficient detection accuracy of fixed weight model.

[0034] S120. Determine the corresponding target state value based on the target sensor corresponding to the target confidence value, and determine the corresponding real-time gate state based on the target state value and the corresponding preset state threshold.

[0035] In this context, the target state value refers to the original state observation value submitted by the sensor with the largest value during the multi-source fusion voting process. In practical applications, the electrical channel submits the dry contact level (high or low level), the acceleration channel submits the calculated attitude angle (continuous value, range 0°~90°), and the vision channel submits the regressed visual angle (continuous value, range 0°~90°). In the scenarios of the acceleration and vision channels, the larger the angle-type target state value, the larger the current tilt angle of the gate, the closer it is to being fully raised (90°), and the more the gate tends to be in the "open" state. The smaller the target state value, the smaller the tilt angle of the gate, the closer it is to being horizontal (0°), and the more the gate tends to be in the "closed" state. In the scenario of the electrical channel, the target state value is a discrete level category (high or low level), which is directly mapped to a specific physical switch state.

[0036] The preset state threshold is a pre-set boundary reference value used to convert continuous angle values ​​into discrete switch states. For example, in the application scenario of 0°~90° unidirectional flipping of port gates, this threshold can be set to 45°. When the target state value (i.e., the gate arm tilt angle) is greater than 45°, it indicates that the gate arm has exceeded the midpoint of its travel and is closer to the vertical lifting position, and the system determines it as "lifted". When the target state value is less than 45°, it indicates that the gate arm has not yet exceeded the midpoint of its travel and is closer to the horizontal closed position, and the system determines it as "lowered".

[0037] The real-time gate arm status is the discrete state result of the gate arm at the current moment after multi-source fusion and threshold determination; including "raised" (gate arm is in the open and allowing passage position) and "lowered" (gate arm is in the closed and blocking position).

[0038] Based on this, the state with the highest number of votes after the three-way weighted voting is the final result of the gate opening / closing quantity.

[0039] S130. Calculate the corresponding trajectory deviation based on the corresponding real-time running trajectory, and determine the corresponding gate arm running status based on the trajectory deviation and the preset deviation.

[0040] The real-time running trajectory refers to the angle-time series curve formed by the actual tilt angle value at each moment during the complete movement of the gate arm from the 0° lowering and locking position to the 90° raising and releasing position, or vice versa, from 90° to 0°. This trajectory is not an isolated angle value at a certain moment, but a digital record of the entire process of the gate arm's continuous movement throughout the entire opening and closing cycle. It can be obtained by real-time sampling at a sampling frequency of 20Hz.

[0041] The trajectory deviation refers to the absolute value of the angle difference between the real-time running trajectory P(t) and the standard trajectory P0(t) at the same point in time, in degrees (°). P0(t) is a standard angle-time trajectory library established by averaging multiple consecutive calibration runs when the barrier gate is in a brand new, fault-free state.

[0042] The preset deviation is a pre-set allowable threshold for angle deviation, used to determine whether the real-time trajectory of a certain frame deviates from the standard trajectory to a degree that needs to be monitored or recorded; if the trajectory deviation of all frames is less than the threshold, it is determined to be normal operation; if the trajectory deviation of any frame exceeds the threshold, it is determined to be an operational anomaly.

[0043] The gate arm operating status refers to the determination of whether there are any abnormalities in the dynamic operation of the gate arm. This status can take four possible values: non-commanded oscillation caused by external disturbances, sensor zero-point drift or minor mechanical deformation, transmission mechanism jamming or insufficient motor power, and normal operation without abnormalities.

[0044] Based on this, traditional railway barrier gate monitoring equipment only collects the gate arm status at a certain angle or periodically at a certain moment to determine whether the gate arm equipment is operating normally. This acquisition method cannot identify subtle mechanical anomalies, instantaneous swaying, and defects in the mechanism deformation of the gate arm. However, by introducing digital twin technology, a standard motion trajectory model of the barrier gate is constructed, enabling refined comparison and detection of the entire time-series trajectory. The system is calibrated under standard conditions, and the complete time-series motion curve of the gate arm from 0° lowering and locking to 90° raising and releasing is plotted, establishing a digital twin standard trajectory library. .

[0045] During equipment operation, the system collects the actual movement trajectory of the gate arm in real time at a frequency of 20Hz. Calculate trajectory deviation frame by frame By classifying the temporal characteristics of trajectory deviation, various anomaly types can be accurately distinguished: short-term small deviations are non-commanded oscillations caused by external disturbances; continuous deviations throughout the entire stroke are sensor zero-point drift or mechanical micro-deformation; sudden deviations at the end of the stroke are transmission mechanism jamming or insufficient motor power faults, achieving the ability to identify subtle faults that traditional detection methods cannot achieve.

[0046] S140. Determine the target gate arm status corresponding to the gate arm based on the real-time gate arm status and gate arm operation status.

[0047] The target gate arm status is an integrated and standardized output result generated by fusing two types of independent output data (real-time gate arm status and gate arm operation status). It can include real-time gate arm status (binary passage permission: gate arm raised / gate arm closed, which only represents whether the level crossing allows train passage and only describes the steady-state position of the gate arm's endpoint) and gate arm operation status (multi-class operating conditions: normal and stable / minor external disturbance / hidden equipment abnormality / serious mechanical failure, which describes whether there are faults or interferences in the dynamic operation of the entire gate arm opening and closing process).

[0048] Based on the feasible implementation of S110 described above, this application further provides calculations of corresponding electrical confidence values, acceleration confidence values, and visual confidence values ​​based on the number of signal transitions and level noise amplitude, the triaxial acceleration and resting offset corresponding to the gate arm, and edge pixels and adjustment coefficients, including: Based on their respective preset weights, the number of signal transitions and the amplitude of level noise are weighted and summed to obtain the first sum, and the variance of the triaxial acceleration and the resting offset are weighted and summed to obtain the second sum. The difference between the standard value and the first sum, as well as the difference between the standard value and the first sum, is calculated to obtain the electrical confidence value and the acceleration confidence value. The difference between the standard value and the multiplication value is calculated based on the gradient variance of the edge pixel and the adjustment coefficient. The visual confidence value is obtained by multiplying the model confidence value output by the difference in the target visual model output.

[0049] Among them, the preset weight refers to the fixed weighting coefficient set in advance for each deduction item, which is used to adjust the influence of different deduction factors on the final confidence score.

[0050] The standard value is 1, representing the ideal state of "100% credibility".

[0051] The target vision model refers to a lightweight deep learning target detection network deployed in an edge computing unit, used to identify and locate gate arm targets from images acquired by industrial cameras.

[0052] Model confidence value refers to the degree of certainty that the target visual model is about its own detection results. The larger the model confidence value (closer to 1), the more confident the target visual model is about the detection results. That is, the gate arm is clearly imaged in the image, has obvious features, a good viewing angle, and is not occluded. The model can accurately identify and locate the gate arm target, and the detection results are highly reliable. The smaller the model confidence value (closer to 0), the more uncertain the model is about the detection results. That is, the gate arm may be partially occluded, located at the edge of the image, or have extremely low image contrast, which makes it impossible for the model to make a clear detection judgment. The usability of the visual data in this frame is poor.

[0053] Based on this, the quantification value of electrical signal reliability It can be calculated based on the switching frequency and level stability of the dry contact signal. The calculation formula is as follows: ,in This represents the number of abnormal signal transitions per unit time. The level noise amplitude can be obtained through ADC sampling; , These are empirical weighting coefficients, taken as 0.2 and 0.5 respectively. When the signal is stable without abrupt changes... When the value approaches 1, electromagnetic interference causes frequent signal jitter. It will decrease accordingly.

[0054] Accelerometer reliability measurement It can be calculated based on the variance of the triaxial acceleration data and the resting zero-point drift. The calculation formula is as follows: ,in The sum of the variances of the X, Y, and Z axis accelerations reflects the intensity of the vibration disturbance. This is the zero-point offset of the gate arm when it is at rest; , These are empirical weighting coefficients, taken as 0.15 and 0.3 respectively. The stronger the mechanical vibration, The lower the value, the better.

[0055] Reliable quantification of visual recognition It can be calculated based on the confidence of the target detection box and the sharpness of the gate arm edge. The calculation formula is as follows: ,in The gate arm detection confidence score is output by the target detection model. The pixel gradient variance at the gate arm edge reflects the degree of image blur. The adjustment factor is set to 0.02. The value will decrease accordingly when insufficient light or rain / fog causes blurred images.

[0056] Furthermore, when it is necessary to output continuous gate arm angle values, the following fusion model is used: ; in To calculate the angle using the accelerometer, For visual recognition angle.

[0057] Based on the feasible implementation of S110 described above, this application further provides a method for weighted comparison of each confidence value according to its corresponding target weight to obtain a target confidence value, including: Based on the corresponding environmental sensors, it is determined whether the current scene falls within a preset scene category; the preset scene categories include electromagnetic interference scene, acceleration interference scene, and visual interference scene. If it is any of the preset scenario categories, then based on the preset scenario category, determine the corresponding scenario confidence value and other standard confidence values, and determine the preset weight corresponding to the scenario confidence value as the weight to be reduced and the preset weight corresponding to the standard confidence value as the weight to be increased. Calculate the product of the weight to be reduced and the preset reduction coefficient, and the product of the weight to be increased and the preset increase coefficient, respectively, to obtain the target reduction weight corresponding to the scenario confidence value and the target increase weight corresponding to the standard confidence value. Based on the target reduction weight and the target increase weight, weight each confidence value to obtain multiple weighted confidence values. If it is not a preset scenario category, then the preset weight corresponding to each confidence value is determined as the target weight, and based on the target weight, each confidence value is weighted to obtain multiple weighted confidence values; By comparing the values ​​of each weighted confidence value, the target confidence value with the largest value is obtained.

[0058] Among them, electromagnetic interference scenarios refer to scenarios where strong electromagnetic interference is generated when a crane is operating nearby; acceleration interference scenarios refer to scenarios where accelerometer sensing data is easily affected by vibration when a dock crane or vehicle passes by, in which case the acceleration weight needs to be reduced and electrical and visual data should be collected first; visual interference scenarios refer to scenarios such as rainy days, heavy fog, and nighttime, where visual imaging quality is reduced, in which case the visual weight needs to be reduced.

[0059] Scene confidence value refers to the confidence value that needs to be reduced in weight after a certain preset interference scene is identified. In electromagnetic interference scene, scene confidence value is the electrical confidence value E, because electrical signals are most severely interfered with in this scene, and its confidence is the lowest. In acceleration interference scene, scene confidence value is the acceleration confidence value A. In visual interference scene, scene confidence value is the visual confidence value V.

[0060] The standard confidence value refers to the confidence value of one or more channels that needs to be weighted higher after being identified as a certain preset interference scenario, that is, other confidence values ​​other than the scenario confidence value. In the electromagnetic interference scenario, the standard confidence value includes the acceleration confidence value A and the visual confidence value V; in the acceleration interference scenario, the standard confidence value includes the electrical confidence value E and the visual confidence value V; in the visual interference scenario, the standard confidence value includes the electrical confidence value E and the acceleration confidence value A.

[0061] The weight to be reduced refers to the original weight coefficient of the scene confidence value (i.e., the sensor channel most severely affected by interference) in the preset state; the weight to be increased refers to the original weight coefficient of the standard confidence value (i.e., the sensor channel that is not affected by interference or is less affected by interference) in the preset state.

[0062] The preset reduction coefficient refers to the multiplication coefficient used to actively reduce the weight corresponding to the scene confidence value after the scene is identified as an interference scene. Its value range is between (0,1), that is, the weight to be reduced is multiplied by a positive number less than 1 to obtain a smaller target reduction weight.

[0063] The preset enhancement coefficient refers to the multiplication coefficient used to actively enhance the weight corresponding to the standard confidence value after the scene is identified as an interference scenario. Its value is greater than 1, that is, the weight to be enhanced is multiplied by a positive number greater than 1 to obtain a larger target enhancement weight.

[0064] Based on this, the core data fusion model is: ,in , , For the corresponding dynamic weight coefficients, satisfying , For the quantification of electrical signal reliability, This is a quantified value of the reliability of the triaxial accelerometer. is the quantified value of visual recognition confidence; S is the target confidence value with the largest value after comparison.

[0065] , , The system dynamically adjusts its weights based on environmental variables such as lighting, rainfall, vibration intensity, and electromagnetic field strength in port scenarios to improve data fusion accuracy in complex environments. The weight adjustment follows this logic: in rainy, foggy, or nighttime scenarios, visual imaging quality deteriorates, and the system automatically reduces the visual weights. , Increase the weighting of electrical and acceleration parameters; when dock cranes and vehicles pass by, accelerometer data is easily affected by vibration, so the system automatically reduces the weighting of acceleration. Prioritize the collection of electrical and visual data; When a crane is operating nearby, it generates strong electromagnetic interference, increasing the probability of electrical signal distortion. The system then automatically reduces the electrical weight. Acceleration and visual data are used as the core criteria for judgment.

[0066] Based on the feasible implementation of S120 described above, this application further provides a method for determining a corresponding target state value based on a target sensor corresponding to a target confidence value, including: Determine the type of the target sensor; If the target sensor is an electrical sensor, the target state value is determined based on the contact level corresponding to the gate arm; If the target sensor is an accelerometer, the corresponding gate tilt angle is calculated based on the real-time collected triaxial acceleration components to obtain the target state value. If the target sensor is a vision sensor, the corresponding target state value is determined based on the target detection angle output by the target vision model.

[0067] The contact level refers to the electrical signal state output by the dry contact limit switch inside the gate control cabinet, which is expressed as either a discrete high level or a low level voltage state. In the typical electrical design of a port gate, the limit switch is installed near the gate arm drive motor. When the gate arm moves to the fully lowered (0°) position, the gate arm base or transmission mechanism will physically contact the mechanical contacts of the limit switch, triggering the switch to act and output a definite level signal; when the gate arm leaves the lowered position, the limit switch resets and outputs the opposite level signal.

[0068] The gate arm tilt angle refers to the real-time tilt angle of the gate arm relative to the horizontal plane, calculated based on the X, Y, and Z axis acceleration components αx, αy, and αz collected in real time by a triaxial MEMS accelerometer installed in the middle section of the gate arm. The unit is degrees (°), and the value ranges from 0° to 90°. 0° represents that the gate arm is completely horizontal and in the lowered closed position; 90° represents that the gate arm is completely vertical and in the raised open position; the intermediate value represents the gate arm in the transition process of raising or lowering.

[0069] Furthermore, the gate arm tilt angle can be used express: ; in, , , These represent the three-axis acceleration components collected in real time by the accelerometer, with units of m / s². The X-axis points along the gate arm towards the tip; the Y-axis is perpendicular to the arm and points directly upward when the gate arm is horizontal; and the Z-axis is perpendicular to the XY plane and points to the side of the gate arm, conforming to a right-hand screw.

[0070] In response, further determination can be made using the relative deviation rate for abnormal operating conditions: ; in This represents the baseline values ​​of the barrier gate parameters (such as standard angular velocity, standard electrical parameters, etc.), and S represents the real-time detection value. The relative deviation rate is used to classify faults into three levels: normal, minor warning, and serious fault, enabling accurate identification of instantaneous faults.

[0071] The target detection angle refers to the angle between the target vision model (based on the YOLOv8n architecture and fine-tuned for port gate scenarios) and the horizontal direction of the image. After reasoning about the image captured by the industrial camera, the target model extracts the pixel-level mask of the gate arm through a semantic segmentation network within the detected gate arm target area, and fits the center axis of the gate arm using the least squares method.

[0072] Based on the feasible implementation of S120 described above, this application further provides a method for determining the corresponding real-time gate arm state based on the target state value and the corresponding preset state threshold, including: When the target state value is the contact level, determine the level category corresponding to the contact level; If the level category is preset high level, then the real-time gate arm status is determined to be gate arm closed; If the level category is preset low level, then the real-time gate arm status is determined to be gate arm raised; When the target state value is the gate arm tilt angle or the target detection angle, the preset state threshold is determined to be the preset angle threshold, and it is determined whether the corresponding angle is higher than the corresponding preset angle threshold. If it is higher, then the real-time gate arm status is determined to be gate arm raised; If it is not higher than the value, then the real-time gate arm status is determined to be gate arm closed.

[0073] The preset high level refers to a preset voltage reference value used to characterize the dry contact signal as being in a "high" state. It is typically the effective high-level voltage range in digital logic circuits. In a typical 3.3V or 5V digital system, the preset high level is a voltage range. For example, in a 3.3V system, a voltage greater than 2.0V is typically considered high; in a 5V system, a voltage greater than 3.5V is typically considered high. The input analog voltage signal is shaped and judged by the Schmitt trigger built into the GPIO interface of the edge computing unit. When the input voltage is higher than the preset high-level threshold voltage, the Schmitt trigger outputs logic "1", and the system determines that the current state is high; when the input voltage is lower than the preset low-level threshold voltage, it outputs logic "0", and the system determines that the state is low.

[0074] The preset low level refers to a preset voltage reference value used to characterize the dry contact signal as being in a "low" state. It is usually the effective low-level voltage range in digital logic circuits.

[0075] The preset angle threshold refers to the boundary reference angle value that the system pre-sets to convert the continuously changing gate tilt angle into discrete "raised" or "closed" switching states.

[0076] Based on the feasible implementation of S130 described above, this application further provides a method for determining the corresponding gate arm operating state according to the trajectory deviation and a preset deviation, including: For the trajectory deviation corresponding to each collection cycle, determine whether the trajectory deviation is greater than the preset deviation. If the deviation of all trajectories is greater than the preset deviation, the corresponding gate arm operation state is determined to be that the gate arm has sensor zero-point drift or mechanical micro-deformation; If the trajectory deviation is greater than the preset deviation only for the last collection cycle, then the corresponding gate arm operation status is determined to be that the gate arm has a transmission mechanism jamming or insufficient motor power fault. If the deviation of some continuous trajectories is greater than the preset deviation, the corresponding gate arm operating state is determined to be non-command swing caused by external disturbance. If the deviation of all trajectories is not greater than the preset deviation, then the corresponding gate arm operation status is determined to be that the gate arm does not have any operational abnormalities.

[0077] The acquisition cycle refers to the basic time unit interval for continuously sampling the actual running trajectory of the gate arm, that is, the time difference between two adjacent trajectory data acquisition times; for example, the sampling frequency is set to 20Hz, so the time length corresponding to a single acquisition cycle is 50 milliseconds.

[0078] Partially continuous trajectory deviation refers to a sequence of trajectory deviations collected within a single gate opening and closing cycle where one or more consecutive acquisition cycles exceed a preset deviation threshold, but this excess range does not cover the entire travel (not all acquisition cycles exceed the threshold) nor is it limited to a specific deviation pattern in the last acquisition cycle.

[0079] Based on this, by classifying the temporal characteristics of trajectory deviation, various types of anomalies can be accurately distinguished: short-term small deviations are non-commanded oscillations caused by external disturbances; continuous deviations throughout the entire stroke are sensor zero-point drift or mechanical micro-deformation; sudden deviations at the end of the stroke are transmission mechanism jamming or insufficient motor power failures, thus achieving the ability to identify subtle faults that traditional detection methods cannot achieve.

[0080] Based on the feasible implementation of S140 described above, this application further provides a method that, after determining the target gate arm state corresponding to the gate arm based on the real-time gate arm state and the gate arm operating state, also includes: Based on the preset detection cycle, the opening and closing time deviation, angular velocity attenuation, jitter frequency anomaly and trajectory deviation of the gate arm are determined. According to their respective weights, the opening and closing time deviation, angular velocity attenuation, jitter frequency anomaly and trajectory deviation are weighted and summed to obtain the corresponding sum value. Calculate the difference between the standard value and the sum value to obtain the corresponding health value, and compare the health value with multiple preset health thresholds to obtain the corresponding health status; If the health status is slightly deteriorated, then the corresponding periodic test result of the gate arm is determined to be that there may be slight bearing wear or increased transmission clearance. If the health status is severely degraded, the corresponding periodic test result of the gate arm is determined to be that there may be motor aging, mechanism jamming, or component corrosion failure. Based on the periodic detection results, the corresponding prompt message is output.

[0081] The preset testing cycle refers to the length of the time window selected for statistically calculating various deterioration indicators when assessing the health of the gate arm equipment. For example, the testing cycle can be set to 15 days, which means continuously recording the core operating parameters of each gate arm opening and closing action in the last 15 days (including single opening and closing duration T, average peak angular velocity ω, operating jitter frequency τ, lateral swing angle θ, etc.), and using the accumulated data of these 15 days as a basis, calculating various deviation and abnormality indicators, and then obtaining the current equipment health score.

[0082] Opening and closing time deviation : ,in This represents the average duration of a single opening and closing operation over the past 7 days. This refers to the standard opening and closing time of the equipment at the factory.

[0083] angular velocity decay : ,in This represents the average peak angular velocity over the past 7 days. This refers to the factory-standard peak angular velocity of the equipment.

[0084] jitter frequency anomaly : ,in This represents the number of sudden changes in angular velocity during a single opening and closing process. The normal threshold is used, and the test is performed 5 times.

[0085] Trajectory deviation : ,in The average deviation angle between the actual trajectory and the standard digital twin trajectory. The allowable deviation threshold is set to 3°.

[0086] The Health Value (HI) is a normalized score calculated by the system after considering four degradation indicators. It is used to quantitatively characterize the overall mechanical health of the barrier gate equipment. The Health Value ranges from 0 to 1. The closer the value is to 0, the higher the degree of mechanical degradation of the equipment. The weighting coefficients w1=0.3, w2=0.3, w3=0.2, and w4=0.2 were determined by combining the Analytic Hierarchy Process (AHP) with port gate operation and maintenance experience.

[0087] The preset health threshold refers to a pre-set discrete boundary reference value used to divide the continuously changing health value HI into different health levels, including two levels: mild decay threshold and severe decay threshold. For example, the mild decay threshold is set to 0.8 (i.e., when HI<0.8, it enters the mild decay range), and the severe decay threshold is set to 0.6 (i.e., when HI<0.6, it enters the severe decay range).

[0088] Based on this, the complete three-level health status classification logic is as follows: when HI ≥ 0.8, it is judged as "healthy", the equipment is in good condition and no maintenance measures are required; when 0.6 ≤ HI<0.8, it is judged as "mild degradation", the equipment has shown a detectable trend of mechanical deterioration (such as slight bearing wear, increased transmission clearance), triggering a maintenance warning to prompt maintenance personnel to perform inspection and maintenance within the planned shutdown window; when HI<0.6, it is judged as "severe degradation", the equipment has shown significant mechanical performance degradation (such as motor aging, mechanism jamming, component corrosion failure), triggering an emergency maintenance prompt, recommending immediate repair to avoid sudden failures leading to traffic safety accidents.

[0089] Please refer to Figure 2 , Figure 2 This is a schematic diagram of a method for identifying the state of a port railway gate arm provided in an embodiment of this application; as shown. Figure 2 As shown, the system uses a multi-source sensing module to collect electrical dry contact signals, triaxial acceleration data, and gate arm visual image data in parallel, achieving time synchronization of sensor data. The synchronized data enters the multi-source data fusion module, where attitude and angular velocity calculations, target detection, and angle regression are performed. After anti-interference correction, a high-confidence result is obtained through dynamic confidence weight fusion. The high-confidence result is then input into the fault prediction module, which sequentially performs digital twin trajectory matching and temporal feature extraction to carry out fault prediction and health monitoring, thereby achieving gate arm status identification and equipment maintenance early warning.

[0090] Please refer to Figure 3 , Figure 3A schematic diagram of the framework for a port railway gate state recognition method provided in this application embodiment; as shown Figure 3 As shown, a three-axis accelerometer is installed on the gate arm, and a visual camera is deployed above the control cabinet. It is equipped with an external environmental information acquisition device. The collected data is sent to the multi-source data fusion and fault prediction module for processing. The processed data is uploaded to the cloud platform to realize remote status monitoring, data visualization and alarm, and complete the full-process status perception and operation and maintenance management of the gate.

[0091] Based on the above steps, it can be seen that this application solves the problem of low accuracy of traditional fixed weight and simple comparison by integrating dynamic credibility from three sources. It realizes the real-time adjustment of the credibility weight of each dimension of data according to environmental parameters such as light, rain, fog, vibration, and electromagnetic fields, which improves the accuracy of state recognition under complex working conditions and solves the defects of weak anti-interference and high false judgment rate. By comparing the deviation characteristics of the actual trajectory with the standard trajectory in real time, it can accurately distinguish subtle hidden faults such as external disturbances, mechanical deformation, transmission jamming, and sensor drift, which solves the problem of traditional single-point parameter detection and recognition having a single dimension and being unable to capture progressive anomalies. Based on multi-dimensional parameters such as long-term opening and closing time, running speed, and vibration frequency, the health of the gate arm can be evaluated, and the mechanical deterioration trend such as motor aging, bearing wear, and mechanism jamming can be predicted in advance, realizing the transformation from fault alarm to fault prediction and improving the intelligent operation and maintenance level of port unmanned equipment.

[0092] Figure 4 This is a schematic diagram of the structure of a port railway gate status identification device provided in an embodiment of this application. Figure 4 As shown, this port railway gate arm status identification device includes: a comparison module, a target status value determination module, a calculation module, and a gate arm status determination module; wherein: The comparison module is used to calculate the corresponding electrical confidence value, acceleration confidence value and visual confidence value based on the number of signal transitions and level noise amplitude, the triaxial acceleration and resting offset of the gate arm, and the edge pixels and adjustment coefficients, respectively. Then, according to their respective target weights, the confidence values ​​are weighted and compared to obtain the target confidence value. The target state value determination module is used to determine the corresponding target state value based on the target sensor corresponding to the target confidence value, and to determine the corresponding real-time gate state based on the target state value and the corresponding preset state threshold. The calculation module is used to calculate the corresponding trajectory deviation based on the corresponding real-time running trajectory, and determine the corresponding gate arm operating status based on the trajectory deviation and the preset deviation. The gate arm status determination module is used to determine the target gate arm status corresponding to the gate arm based on the real-time gate arm status and gate arm operation status.

[0093] In this embodiment of the application, the comparison module can also be specifically used for: Based on their respective preset weights, the number of signal transitions and the amplitude of level noise are weighted and summed to obtain the first sum, and the variance of the triaxial acceleration and the resting offset are weighted and summed to obtain the second sum. The difference between the standard value and the first sum, as well as the difference between the standard value and the first sum, is calculated to obtain the electrical confidence value and the acceleration confidence value. The difference between the standard value and the multiplication value is calculated based on the gradient variance of the edge pixel and the adjustment coefficient. The visual confidence value is obtained by multiplying the model confidence value output by the difference in the target visual model output.

[0094] In this embodiment of the application, the comparison module can also be specifically used for: Based on the corresponding environmental sensors, it is determined whether the current scene falls within a preset scene category; the preset scene categories include electromagnetic interference scene, acceleration interference scene, and visual interference scene. If it is any of the preset scenario categories, then based on the preset scenario category, determine the corresponding scenario confidence value and other standard confidence values, and determine the preset weight corresponding to the scenario confidence value as the weight to be reduced and the preset weight corresponding to the standard confidence value as the weight to be increased. Calculate the product of the weight to be reduced and the preset reduction coefficient, and the product of the weight to be increased and the preset increase coefficient, respectively, to obtain the target reduction weight corresponding to the scenario confidence value and the target increase weight corresponding to the standard confidence value. Based on the target reduction weight and the target increase weight, weight each confidence value to obtain multiple weighted confidence values. If it is not a preset scenario category, then the preset weight corresponding to each confidence value is determined as the target weight, and based on the target weight, each confidence value is weighted to obtain multiple weighted confidence values; By comparing the values ​​of each weighted confidence value, the target confidence value with the largest value is obtained.

[0095] In this embodiment of the application, the target state value determination module can also be specifically used for: Determine the type of the target sensor; If the target sensor is an electrical sensor, the target state value is determined based on the contact level corresponding to the gate arm; If the target sensor is an accelerometer, the corresponding gate tilt angle is calculated based on the real-time collected triaxial acceleration components to obtain the target state value. If the target sensor is a vision sensor, the corresponding target state value is determined based on the target detection angle output by the target vision model.

[0096] In this embodiment of the application, the target state value determination module can also be specifically used for: When the target state value is the contact level, determine the level category corresponding to the contact level; If the level category is preset high level, then the real-time gate arm status is determined to be gate arm closed; If the level category is preset low level, then the real-time gate arm status is determined to be gate arm raised; When the target state value is the gate arm tilt angle or the target detection angle, the preset state threshold is determined to be the preset angle threshold, and it is determined whether the corresponding angle is higher than the corresponding preset angle threshold. If it is higher, then the real-time gate arm status is determined to be gate arm raised; If it is not higher than the value, then the real-time gate arm status is determined to be gate arm closed.

[0097] In this embodiment of the application, the calculation module can also be specifically used for: For the trajectory deviation corresponding to each collection cycle, determine whether the trajectory deviation is greater than the preset deviation. If the deviation of all trajectories is greater than the preset deviation, the corresponding gate arm operation state is determined to be that the gate arm has sensor zero-point drift or mechanical micro-deformation; If the trajectory deviation is greater than the preset deviation only for the last collection cycle, then the corresponding gate arm operation status is determined to be that the gate arm has a transmission mechanism jamming or insufficient motor power fault. If the deviation of some continuous trajectories is greater than the preset deviation, the corresponding gate arm operating state is determined to be non-command swing caused by external disturbance. If the deviation of all trajectories is not greater than the preset deviation, then the corresponding gate arm operation status is determined to be that the gate arm does not have any operational abnormalities.

[0098] In this embodiment of the application, the other modules included in the port railway gate status identification device can also be specifically used for: Based on the preset detection cycle, the opening and closing time deviation, angular velocity attenuation, jitter frequency anomaly and trajectory deviation of the gate arm are determined. According to their respective weights, the opening and closing time deviation, angular velocity attenuation, jitter frequency anomaly and trajectory deviation are weighted and summed to obtain the corresponding sum value. Calculate the difference between the standard value and the sum value to obtain the corresponding health value, and compare the health value with multiple preset health thresholds to obtain the corresponding health status; If the health status is slightly deteriorated, then the corresponding periodic test result of the gate arm is determined to be that there may be slight bearing wear or increased transmission clearance. If the health status is severely degraded, the corresponding periodic test result of the gate arm is determined to be that there may be motor aging, mechanism jamming, or component corrosion failure. Based on the periodic detection results, the corresponding prompt message is output.

[0099] Figure 5This is a schematic diagram of the structure of a device for implementing a port railway gate state identification method according to an embodiment of this application. Figure 5 As shown, the device includes: The device may include one or more processors with processing cores, one or more computer-readable storage media such as memory, communication components, etc. The processor, memory, and communication components are connected via a bus.

[0100] In the specific implementation process, at least one processor executes computer execution instructions stored in memory, causing at least one processor to execute the above-described port railway gate state recognition method.

[0101] The specific implementation process of the processor can be found in the above method embodiments, and its implementation principle and technical effect are similar, so it will not be repeated here.

[0102] Furthermore, the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0103] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0104] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0105] In some embodiments, a computer program product is also provided, including a computer program or instructions that, when executed by a processor, implement the steps in any of the above-described port railway gate state recognition methods.

[0106] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0107] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0108] Therefore, embodiments of this application provide a computer-readable storage medium storing multiple lines of program code that can be loaded by a processor to execute the steps in any of the port railway gate state recognition methods provided in embodiments of this application.

[0109] The storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0110] According to one aspect of this application, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium.

[0111] Since the instructions stored in the storage medium can execute the steps in any of the port railway gate state identification methods provided in the embodiments of this application, the beneficial effects that any of the port railway gate state identification methods provided in the embodiments of this application can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0112] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope of this application is indicated by the appended claims.

[0113] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A method for identifying the state of a port railway gate arm, characterized in that, The method includes: Based on the number of signal transitions and the amplitude of level noise, the triaxial acceleration and resting offset of the gate arm, and the edge pixels and adjustment coefficient, the corresponding electrical confidence value, acceleration confidence value and visual confidence value are calculated respectively. Then, according to their respective target weights, the confidence values ​​are weighted and compared to obtain the target confidence value. Based on the target sensor corresponding to the target confidence value, determine the corresponding target state value, and based on the target state value and the corresponding preset state threshold, determine the corresponding real-time gate state. Based on the corresponding real-time running trajectory, calculate the corresponding trajectory deviation, and determine the corresponding gate arm running state based on the trajectory deviation and the preset deviation. Based on the real-time gate arm status and the gate arm operating status, the target gate arm status corresponding to the gate arm is determined.

2. The method according to claim 1, characterized in that, The calculation of corresponding electrical confidence values, acceleration confidence values, and visual confidence values ​​based on the number of signal transitions and level noise amplitude, the triaxial acceleration and resting offset of the gate arm, and edge pixels and adjustment coefficients includes: Based on their respective preset weights, the number of signal transitions and the amplitude of the level noise are weighted and summed to obtain a first sum, and the variance of the triaxial acceleration and the resting offset are weighted and summed to obtain a second sum. The difference between the standard value and the first sum, and the difference between the standard value and the first sum, are calculated to obtain the electrical confidence value and the acceleration confidence value. The difference between the standard value and the multiplication value is calculated based on the gradient variance of the edge pixel and the adjustment coefficient. The visual confidence value is obtained by multiplying the model confidence value output by the target visual model by the difference.

3. The method according to claim 1, characterized in that, The step of weighting and comparing each confidence value according to its corresponding target weight to obtain the target confidence value includes: Based on the corresponding environmental sensors, it is determined whether the current scene belongs to a preset scene category; wherein, the preset scene categories include electromagnetic interference scene, acceleration interference scene and visual interference scene; If it is any of the preset scenario categories, then based on the preset scenario category, determine the corresponding scenario confidence value and other standard confidence values, and determine the preset weight corresponding to the scenario confidence value as the weight to be reduced and the preset weight corresponding to the standard confidence value as the weight to be increased. Calculate the product of the weight to be reduced and the preset reduction coefficient, and the product of the weight to be increased and the preset increase coefficient, respectively, to obtain the target reduction weight corresponding to the scenario confidence value and the target increase weight corresponding to the standard confidence value. Based on the target reduction weight and the target increase weight, weight each confidence value is calculated to obtain multiple weighted confidence values. If it is not the preset scenario category, then the preset weight corresponding to each confidence value is determined as the target weight, and based on the target weight, each confidence value is weighted to obtain multiple weighted confidence values; By comparing the values ​​of each weighted confidence value, the target confidence value with the largest value is obtained.

4. The method according to claim 1, characterized in that, The step of determining the corresponding target state value based on the target sensor corresponding to the target confidence value includes: Determine the category of the target sensor; If the target sensor is an electrical sensor, the target state value is determined based on the contact level corresponding to the gate arm; If the target sensor is an accelerometer, the corresponding gate tilt angle is calculated based on the real-time collected triaxial acceleration components to obtain the target state value. If the target sensor is a vision sensor, the corresponding target state value is determined based on the target detection angle output by the target vision model.

5. The method according to claim 4, characterized in that, The step of determining the corresponding real-time gate state based on the target state value and the corresponding preset state threshold includes: When the target state value is the contact level, determine the level category corresponding to the contact level; If the level category is a preset high level, then the real-time gate state is determined to be gate closed; If the level category is a preset low level, then the real-time gate arm status is determined to be gate arm raised; When the target state value is the gate arm tilt angle or the target detection angle, the preset state threshold is determined to be the preset angle threshold, and it is determined whether the corresponding angle is higher than the corresponding preset angle threshold. If it is higher, then the real-time gate arm status is determined to be gate arm raised; If it is not higher than the specified value, then the real-time gate state is determined to be gate closed.

6. The method according to claim 1, characterized in that, The step of determining the corresponding gate arm operating state based on the trajectory deviation and the preset deviation includes: For each acquisition cycle, determine whether the trajectory deviation is greater than the preset deviation. If all of the trajectory deviations are greater than the preset deviation, then the corresponding gate arm operating state is determined to be that the gate arm has sensor zero-point drift or mechanical micro-deformation; If the trajectory deviation is greater than the preset deviation only for the last collection cycle, then the corresponding gate arm operating status is determined to be that the gate arm has a transmission mechanism jamming or insufficient motor power fault. If the deviation of some consecutive trajectories is greater than the preset deviation, then the corresponding gate arm operating state is determined to be non-commanded swing caused by external disturbance. If all the trajectory deviations are not greater than the preset deviation, then the corresponding gate arm operating status is determined to be that the gate arm does not have any operating abnormality.

7. The method according to claim 1, characterized in that, After determining the target gate arm state corresponding to the gate arm based on the real-time gate arm state and the gate arm operating state, the method further includes: Based on a preset detection cycle, the opening and closing time deviation, angular velocity attenuation, jitter frequency anomaly, and trajectory deviation of the gate arm are determined. Then, according to their respective weights, the opening and closing time deviation, angular velocity attenuation, jitter frequency anomaly, and trajectory deviation are weighted and summed to obtain the corresponding sum value. The difference between the standard value and the sum is calculated to obtain the corresponding health value, and the health value is compared with multiple preset health thresholds to obtain the corresponding health status. If the health status is slightly deteriorated, then the periodic detection result corresponding to the gate arm is determined to be that there may be slight bearing wear or increased transmission clearance. If the health status is severely degraded, then the periodic test result corresponding to the gate arm is determined to be that there may be motor aging, mechanism jamming, or component corrosion failure. Based on the periodic detection results, the corresponding prompt message is output.

8. A status identification device for a port railway gate arm, characterized in that, The device includes: The comparison module is used to calculate the corresponding electrical confidence value, acceleration confidence value and visual confidence value based on the number of signal transitions and level noise amplitude, the triaxial acceleration and resting offset of the gate arm, and the edge pixels and adjustment coefficients, respectively. Then, according to their respective target weights, the confidence values ​​are weighted and compared to obtain the target confidence value. The target state value determination module is used to determine the corresponding target state value based on the target sensor corresponding to the target confidence value, and to determine the corresponding real-time gate state based on the target state value and the corresponding preset state threshold. The calculation module is used to calculate the corresponding trajectory deviation based on the corresponding real-time running trajectory, and determine the corresponding gate arm running state based on the trajectory deviation and the preset deviation. The gate arm status determination module is used to determine the target gate arm status corresponding to the gate arm based on the real-time gate arm status and the gate arm operating status.

9. A computer device, characterized in that, include: One or more processors; Memory; One or more programs, wherein the one or more programs are stored in memory and configured to be executed by one or more processors, the one or more programs being configured to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program code that can be called by a processor to perform the method as described in any one of claims 1 to 7.

Images