A visibility assessment method and system based on rail transit vehicle-mounted PIS

Abstract

The present invention relates to the field of rail transit technology, and specifically to a visibility assessment method and system based on a rail transit vehicle-mounted PIS, comprising the following steps: S1, extracting and modeling multidimensional perception elements based on rail transit train operation data, carriage environment perception data, and passenger density data, and outputting a PIS visibility influencing factor map; S2, inputting the map into a structural perception fusion model, performing weighted processing in combination with the time axis and synchronized trajectory, and outputting a PIS visibility scoring curve for each operation stage; S3, setting a dynamic threshold based on the scoring curve to identify degraded sections, combining passenger behavior data classification, and outputting a visibility zoning assessment result with excellent, medium, and poor grades. The present invention, through multidimensional perception element modeling, structural perception fusion, and behavior data-driven classification, realizes dynamic assessment, scene recognition, and grading of vehicle-mounted PIS visibility, thereby improving the stability and accessibility of rail transit information services.

Description

Technical Field

[0001] The present invention relates to the field of rail transit technology, and in particular to a visibility assessment method and system based on a rail transit onboard PIS. Background Art

[0002] With the large-scale construction and advancement of smart operations of urban rail transit systems, on-board PIS, as the primary interface for passengers to obtain dynamic travel information, has become an important component for improving service quality, enhancing operational safety, and increasing passenger satisfaction. The PIS system provides passengers with station arrival information, transfer reminders, operating status, and other content through graphic displays and voice broadcasts, ensuring the immediacy and accuracy of information transmission. This is of key significance in alleviating anxiety in carriages and supporting emergency response.

[0003] However, existing technologies mainly focus on the accuracy and scheduling logic of PIS content, and rarely involve its visibility, a key dimension of passenger perception. In actual operation, PIS visibility is affected by multiple factors such as screen brightness, car illumination, vehicle posture, and passenger occlusion. As a result, even if the information is correctly released, it may not be effectively obtained by passengers due to visual occlusion, reflection, perspective offset and other problems, affecting the quality of information services and emergency response efficiency. Existing systems generally lack quantitative modeling and dynamic evaluation mechanisms for the visibility status of PIS, making it difficult to identify visibility degradation scenarios, and even more impossible to achieve refined regulation and hierarchical operation and maintenance. Summary of the Invention

[0004] The present invention provides a visibility assessment method and system based on rail transit on-board PIS. By integrating multi-source perception elements to construct a visibility influencing factor map, structural perception weighted modeling is performed in combination with the operating status, and a visibility scoring curve is output. Passenger behavior data is introduced to realize degradation scenario classification and zoning assessment. The method has strong dynamic adaptability and assessment accuracy, and significantly improves the information service accessibility, operational stability and intelligent operation and maintenance level of the rail transit PIS system.

[0005] A visibility assessment method based on a rail transit vehicle-mounted PIS comprises the following steps:

[0006] S1, based on rail transit train operation data, carriage interior environment perception data and passenger density distribution data, extracts multi-dimensional perception factors, including on-board display screen illumination, screen orientation angle, vehicle body pitch posture, and passenger occlusion area, and then performs correlation modeling on the multi-dimensional perception factors in time and space dimensions to output a map of on-board PIS visibility influencing factors;

[0007] S2: Input the PIS visibility influencing factor map into the structural perception fusion model. Combined with the train operation timeline and the synchronous trajectory of the sensor data in the carriage, the visibility impact intensity of each PIS visibility influencing factor is weighted and processed. The visibility feature mapping model under different train operation states is constructed, and the PIS visibility score curve under each operation stage is output.

[0008] S3, based on the PIS visibility scoring curve, sets a dynamic visibility evaluation threshold, identifies visibility-degraded sections, and classifies visibility-degraded scenarios based on passenger complaint records and emergency broadcast trigger frequencies in historical operating data. It then outputs visibility zoning assessment results divided into excellent, medium, and poor grades.

[0009] Optionally, the S1 includes:

[0010] S11, based on rail transit train operation data, carriage interior environment perception data and passenger density distribution data, adopts a sliding time window The time synchronization method is used to perform time alignment processing to generate an aligned time series perception dataset ;

[0011] S12, from the aligned time series perception dataset Extract the perceptual factors related to PIS visibility, including the illumination of the vehicle display screen , screen orientation angle , vehicle pitch attitude, passenger occlusion area The vehicle body pitch posture includes the pitch angle and roll angle ;

[0012] S13, embed the perception elements into a unified space-time structure, and use the space grid mapping and time segment aggregation method to form a perception element sequence. The spatial position of 、 , then the perception factor vector is constructed as ,in, For location The proportion of occluded pixels at ;

[0013] S14, evaluate the visibility impact strength of the perception element vector at each position-time, and calculate the impact strength value ;

[0014] S15, based on the influence intensity value of each position-time in each frame of data , build a visual impact factor map .

[0015] Optionally, the S11 includes:

[0016] S111, using the master clock (GPS timing) in the rail transit train operation system as the reference time axis, defines the time sampling sequence as ,in, For the Sampling time, , is the total number of sampling points within the target evaluation period, ;

[0017] S112, constructing a structured data sequence with a time stamp for the rail transit train operation data, the carriage interior environment perception data, and the passenger density distribution data;

[0018] S113, for each main time point , select the nearest data points in the time window before and after the structured data sequence, and perform linear interpolation to generate the aligned train operation data feature vector, environmental perception data feature vector and passenger density data feature vector;

[0019] S114: Splice the aligned train operation data feature vector, environment perception data feature vector, and passenger density data feature vector into a time series perception data set .

[0020] Optionally, the S2 includes:

[0021] S21, based on the running tag signal output by the train main control system, divides the entire running cycle into several running state segments according to the time series, and extracts the synchronous trajectory data set in each running state segment ,in, For the moment the perceptual elements;

[0022] S22, the perception elements in each running state segment Input the structure-aware fusion model and calculate the fused visibility index ;

[0023] S23, in each running state segment Output continuous visibility index in time series , forming a scoring curve , and normalized.

[0024] Optionally, the S21 includes:

[0025] S211, obtain the operation tag sequence recorded by timestamp from the train main control system ,in, For the The timestamp of each tag record, For time point The train running status label at The total number of running tag sequences, and the running status fragment set is constructed based on the continuous time period of the same tag ;

[0026] S212, let the sequence of perception elements during train operation be , Record timestamps for perception elements, For the moment The perceptual element vector of is the total length of the perception element sequence, for each running state segment , extract the perception elements within its time interval and construct the synchronous trajectory dataset of the segment .

[0027] Optionally, the S22 includes:

[0028] S221, using minimum-maximum normalization to perceptual element vector All dimensions in the .

[0029] S222: Use the structure perception fusion model to perform weighted combination on the normalized perception factor vectors to obtain the visibility index at a single moment. .

[0030] Optionally, the S23 includes:

[0031] S231, from the running state fragment The calculated visibility index is sorted in chronological order from the perception elements within the image, and the visibility index time series is extracted. ;

[0032] S232, running status fragment All visibility indices within Perform minimum-maximum normalization to obtain a normalized visibility score , and all normalized visibility scores form a scoring curve .

[0033] Optionally, the S3 includes:

[0034] S31, based on the running status fragment Scoring curve within , set dynamic threshold , identify satisfaction The time interval is the visibility degradation section;

[0035] S32, in each visibility degradation section Extract passenger behavior data within the corresponding time period, including the frequency of passenger complaint records (such as PIS visibility-related complaints), the number of emergency broadcast triggers (such as "the next station announcement is repeated or failed"), the density of passenger re-inquiries (which can be counted through voice recognition), and record the behavior label feature set , expressed as:

[0036] ;

[0037] in, is the number of passenger complaints per unit time in the deteriorated section, is the number of emergency broadcast triggers per unit time, For the Behavioral characteristic vector of the degraded section;

[0038] S33, based on label feature set , each visibility degradation segment is divided into scene categories, including high-reflection occlusion type, interactive failure type, and systemic failure type, and the scene category and the mean score are combined to form a , output visibility level , including excellent, medium and poor.

[0039] Optionally, the classification of scenes includes:

[0040] Highly reflective blocking type: 、 、 ,in, A high threshold for complaints, is the broadcast high threshold;

[0041] Interaction failure type: 、 ,in, is the score fluctuation threshold, For the The standard deviation of the visibility score curve in the visibility degradation section;

[0042] Systemic failure type: 、 ;

[0043] The visibility level is expressed as:

[0044] .

[0045] A visibility assessment system based on a rail transit vehicle-mounted PIS is used to implement the above-mentioned visibility assessment method based on a rail transit vehicle-mounted PIS, and includes the following modules:

[0046] Perception factor extraction module: Based on rail transit train operation data, carriage interior environment perception data, and passenger density distribution data, it extracts multi-dimensional perception factors including on-board display screen illumination, screen orientation angle, vehicle body pitch posture, and passenger occlusion area, and performs correlation modeling in the time and space dimensions to generate a PIS visibility influencing factor map;

[0047] Structural Perception Fusion Module: This module inputs the PIS visibility influencing factor map into the structural perception fusion model. Combining the train's operating timeline with the synchronized trajectory of the sensor data within the carriage, it performs weighted processing on the perception factors, constructs a visibility feature mapping model under different operating states, and outputs the PIS visibility score curve for each operating stage.

[0048] Visibility Assessment and Classification Module: This module sets dynamic visibility evaluation thresholds based on the PIS visibility scoring curve, identifies sections with degraded visibility, and classifies these sections based on passenger complaint records and emergency broadcast trigger frequencies in historical operating data. The module then outputs visibility zoning assessment results divided into excellent, medium, and poor grades.

[0049] Beneficial effects of the present invention:

[0050] The present invention constructs a multidimensional perception factor system based on vehicle illumination, posture perception, and passenger distribution, and performs sliding time window alignment and space-time fusion modeling under a unified time axis. This realizes the dynamic perception and structured expression of the complex factors affecting the visibility of the vehicle-mounted PIS, effectively improving the spatiotemporal resolution and robustness of visibility modeling.

[0051] The present invention introduces a structural perception fusion model combined with operating status fragments and adopts normalization processing and weighted aggregation strategy to establish the PIS visibility feature mapping and scoring curve under the operating state. It can dynamically adapt to the weight changes of visibility factors under different operating conditions, and realize a visibility scoring mechanism with scene adaptation and strong state correlation, breaking through the traditional method's reliance on fixed parameter models.

[0052] The present invention sets dynamic thresholds based on scoring curves and integrates behavioral data to perform scene classification and grading. The present invention realizes zoning identification and graded evaluation of visibility degradation situations, and can accurately distinguish typical degradation scenarios such as high-reflective occlusion, interactive failure, and systemic failure. It helps operating units to carry out operation and maintenance scheduling and system parameter adjustment according to regional priority, and significantly improves the stability, safety and passenger satisfaction of the PIS information service system. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0054] Figure 1 Schematic diagram of the evaluation method flow in an embodiment of the present invention;

[0055] Figure 2 Schematic diagram of system function modules according to an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. Those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0057] like Figure 1 As shown, a visibility assessment method based on rail transit vehicle-mounted PIS includes the following steps:

[0058] S1, based on rail transit train operation data, carriage interior environment perception data and passenger density distribution data, extracts multi-dimensional perception factors, including on-board display screen illumination, screen orientation angle, vehicle body pitch posture, and passenger occlusion area, and then performs correlation modeling on the multi-dimensional perception factors in time and space dimensions to output a map of on-board PIS visibility influencing factors;

[0059] S2: Input the PIS visibility influencing factor map into the structural perception fusion model. Combined with the train operation timeline and the synchronous trajectory of the sensor data in the carriage, the visibility impact intensity of each PIS visibility influencing factor is weighted and processed. The visibility feature mapping model under different train operation states is constructed, and the PIS visibility score curve under each operation stage is output.

[0060] S3, based on the PIS visibility scoring curve, sets a dynamic visibility evaluation threshold, identifies visibility-degraded sections, and classifies visibility-degraded scenarios based on passenger complaint records and emergency broadcast trigger frequencies in historical operating data. It then outputs visibility zoning assessment results divided into excellent, medium, and poor grades.

[0061] S1 includes:

[0062] S11, based on rail transit train operation data, carriage interior environment perception data and passenger density distribution data, adopts a sliding time window The time synchronization method is used to perform time alignment processing to generate an aligned time series perception dataset ;

[0063] S12, from the aligned time series perception dataset Extract the perceptual factors related to PIS visibility, including the illumination of the vehicle display screen , screen orientation angle , vehicle pitch attitude, passenger occlusion area , the vehicle pitch attitude includes the pitch angle and roll angle ;

[0064] S13, embed the perception elements into a unified space-time structure, and use the space grid mapping and time segment aggregation method to form a perception element sequence. The spatial position of 、 , then the perception factor vector is constructed as ,in, For location The proportion of occluded pixels at ;

[0065] S14, evaluate the visibility impact strength of the perception element vector at each position-time, and calculate the impact strength value , expressed as: ;

[0066] in, is the maximum acceptable illumination threshold for screen brightness, 、 、 、 、 is the normalized weight coefficient of each impact factor;

[0067] S15, based on the influence intensity value of each position-time in each frame of data , build a visual impact factor map , expressed as:

[0068] ;

[0069] in, For the selected evaluation period, It is the spatial projection range of the visible area in front of the car display screen.

[0070] The S11 includes:

[0071] S111, using the master clock (GPS timing) in the rail transit train operation system as the reference time axis, defines the time sampling sequence as ,in, For the Sampling time, , is the total number of sampling points within the target evaluation period, ;

[0072] S112: construct a structured data sequence with a timestamp for the rail transit train operation data, the carriage interior environment perception data, and the passenger density distribution data, respectively, and express it as:

[0073] ;

[0074] ;

[0075] ;

[0076] in, is the train operation data sequence, is the environmental perception data sequence, is the passenger density data series, 、 、 Respectively represent the original timestamps in each data source, 、 、 are the feature vectors collected at this timestamp (such as speed, illumination image, occlusion rate map, etc.);

[0077] S113, for each main time point , select the nearest data points in the time window before and after the structured data sequence, and perform linear interpolation to generate the aligned train operation data feature vector, environmental perception data feature vector and passenger density data feature vector, which are expressed as:

[0078] ;

[0079] ;

[0080] ;

[0081] in, 、 、 At time points Linear interpolation is used to obtain the aligned train operation data feature vector, the aligned environment perception data feature vector, and the aligned passenger density data feature vector. 、 Timestamp 、 The train operation data feature vector, 、 Timestamp 、 The cabin environment perception data feature vector, 、 Timestamp 、 The passenger density distribution data feature vector;

[0082] S114: Splice the aligned train operation data feature vector, environment perception data feature vector, and passenger density data feature vector into a time series perception data set , expressed as:

[0083] ;

[0084] ;

[0085] in, The total number of sampling points in the target evaluation period.

[0086] S2 includes:

[0087] S21, based on the running tag signal output by the train main control system, divides the entire running cycle into several running state segments according to the time series, and extracts the synchronous trajectory data set in each running state segment ,in, For the moment the perceptual elements;

[0088] S22, the perception elements in each running state segment Input the structure-aware fusion model and calculate the fused visibility index ;

[0089] S23, in each running state segment Output continuous visibility index in time series , forming a scoring curve , and normalized.

[0090] S21 includes:

[0091] S211, obtain the operation tag sequence recorded by timestamp from the train main control system ,in, For the The timestamp of each tag record, For time point The train running status label at the position includes ACC (acceleration), CRU (constant speed cruise), DEC (deceleration), STP (stop), The total number of running tag sequences, and the running status fragment set is constructed based on the continuous time period of the same tag , where each fragment is defined as , corresponding to The start and end times of the segment state, is the total number of identified running status fragments;

[0092] S212, let the sequence of perception elements during train operation be , Record timestamps for perception elements, For the moment The perceptual element vector of is the total length of the perception element sequence, for each running state segment , extract the perception elements within its time interval and construct the synchronous trajectory dataset of the segment , expressed as:

[0093] ;

[0094] in, 、 Respectively The start and end times of the segment.

[0095] S22 includes:

[0096] S221, using minimum-maximum normalization to perceptual element vector Each dimension in the equation is normalized to make different physical quantities comparable, which can be expressed as:

[0097] ;

[0098] ;

[0099] in, is the perception element vector in the The original value of the dimension, 、 They are the perceptual element vectors The minimum and maximum values ​​of the dimension within the state segment, is the normalized perception factor vector, is the vector dimension;

[0100] S222: Use the structure perception fusion model to perform weighted combination on the normalized perception factor vectors to obtain the visibility index at a single moment. , expressed as:

[0101] ;

[0102] in, For the Perception elements in the running state segment The structure-aware weights in

[0103] ;

[0104] in, For the Perception elements in the running state segment The average value in is the temperature parameter, For the Perception elements in the running state segment The average value in .

[0105] S23 includes:

[0106] S231, from the running state fragment The calculated visibility index is sorted in chronological order from the perception elements within the image, and the visibility index time series is extracted. , expressed as:

[0107] ;

[0108] in, Running state fragment Neidi A point in time, is the visibility index at the corresponding time point, is the number of available time points in this state segment;

[0109] S232, running status fragment All visibility indices within Perform minimum-maximum normalization to obtain a normalized visibility score , and all normalized visibility scores form a scoring curve , expressed as:

[0110] ;

[0111] in, is the normalized visibility score, To prevent division by zero for small positive values, 、 are the extreme values ​​of visibility index in the segment respectively;

[0112] .

[0113] S3 includes:

[0114] S31, based on the running status fragment Scoring curve within , set dynamic threshold , identify satisfaction The time interval is the visibility degradation period, which is expressed as:

[0115] ;

[0116] in, is the mean of the normalized scores within the state segment, is the standard deviation of the normalized scores, is the regulating factor;

[0117] S32, in each visibility degradation section Extract passenger behavior data within the corresponding time period, including the frequency of passenger complaint records (such as PIS visibility-related complaints), the number of emergency broadcast triggers (such as "the next station announcement is repeated or failed"), the density of passenger re-inquiries (which can be counted through voice recognition), and record the behavior label feature set , expressed as:

[0118] ;

[0119] in, is the number of passenger complaints per unit time in the deteriorated section, is the number of emergency broadcast triggers per unit time, For the Behavioral characteristic vector of the degraded section;

[0120] S33, based on label feature set , each visibility degradation segment is divided into scene categories, including high-reflection occlusion type, interactive failure type, and systemic failure type, and the scene category and the mean score are combined to form a , output visibility level , including excellent, medium and poor.

[0121] The scene categories include:

[0122] Highly reflective blocking type: 、 、 ,in, A high threshold for complaints, is the broadcast high threshold;

[0123] Interaction failure type: 、 ,in, is the score fluctuation threshold, For the The standard deviation of the visibility score curve in the visibility degradation section;

[0124] Systemic failure type: 、 ;

[0125] The visibility level is expressed as:

[0126] .

[0127] like Figure 2 As shown, a visibility assessment system based on a rail transit vehicle-mounted PIS is used to implement the above-mentioned visibility assessment method based on a rail transit vehicle-mounted PIS, and includes the following modules:

[0128] Perception factor extraction module: Based on rail transit train operation data, carriage interior environment perception data, and passenger density distribution data, it extracts multi-dimensional perception factors including on-board display screen illumination, screen orientation angle, vehicle body pitch posture, and passenger occlusion area, and performs correlation modeling in the time and space dimensions to generate a PIS visibility influencing factor map;

[0129] Structural Perception Fusion Module: This module inputs the PIS visibility influencing factor map into the structural perception fusion model. Combining the train's operating timeline with the synchronized trajectory of the sensor data within the carriage, it performs weighted processing on the perception factors, constructs a visibility feature mapping model under different operating states, and outputs the PIS visibility score curve for each operating stage.

[0130] Visibility Assessment and Classification Module: This module sets dynamic visibility evaluation thresholds based on the PIS visibility scoring curve, identifies sections with degraded visibility, and classifies these sections based on passenger complaint records and emergency broadcast trigger frequencies in historical operating data. The module then outputs visibility zoning assessment results divided into excellent, medium, and poor grades.

[0131] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. To provide a thorough understanding of the present invention, specific details are described in detail below in connection with the preferred embodiments of the present invention, but those skilled in the art will be able to fully understand the present invention without these detailed descriptions. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0132] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A visibility assessment method based on rail transit vehicle-mounted PIS, characterized in that: The following steps are involved: S1, based on rail transit train operation data, carriage interior environment perception data and passenger density distribution data, extracts multi-dimensional perception factors, including on-board display screen illumination, screen orientation angle, vehicle body pitch posture, and passenger occlusion area, and then performs correlation modeling on the multi-dimensional perception factors in time and space dimensions to output a map of on-board PIS visibility influencing factors; S2: Input the PIS visibility influencing factor map into the structural perception fusion model. Combined with the train operation timeline and the synchronous trajectory of the sensor data in the carriage, the visibility impact intensity of each PIS visibility influencing factor is weighted and processed. The visibility feature mapping model under different train operation states is constructed, and the PIS visibility score curve under each operation stage is output. S3, based on the PIS visibility scoring curve, sets dynamic visibility evaluation thresholds, identifies visibility-degraded sections, and classifies visibility-degraded scenarios based on passenger complaint records and emergency broadcast trigger frequencies in historical operational data. It then outputs visibility zoning assessment results divided into excellent, medium, and poor grades. Said S1 comprises: S11, based on rail transit train operation data, carriage interior environment perception data and passenger density distribution data, adopts a sliding time window The time synchronization method is used to perform time alignment processing to generate an aligned time series perception dataset ; S12, from the aligned time series perception dataset Extract the perceptual factors related to PIS visibility, including the illumination of the vehicle display screen , screen orientation angle , vehicle pitch attitude, passenger occlusion area The vehicle body pitch posture includes the pitch angle and roll angle ; S13, embed the perception elements into a unified space-time structure, and use the space grid mapping and time segment aggregation method to form a perception element sequence. The spatial position of 、 , then the perception factor vector is constructed as ,in, For location The proportion of occluded pixels at ; S14, evaluate the visibility impact strength of the perception element vector at each position-time, and calculate the impact strength value ; S15, based on the influence intensity value of each position-time in each frame of data , build a visual impact factor map .

2. A visibility assessment method based on rail transit vehicle-mounted PIS according to claim 1, characterized in that: The S11 includes: S111, using the master clock in the rail transit train operation system as the reference time axis, defines the time sampling sequence as ,in, For the Sampling time, , is the total number of sampling points within the target evaluation period, ; S112, constructing a structured data sequence with a time stamp for the rail transit train operation data, the carriage interior environment perception data, and the passenger density distribution data; S113, for each main time point , select the nearest data points in the time window before and after the structured data sequence, and perform linear interpolation to generate the aligned train operation data feature vector, environmental perception data feature vector and passenger density data feature vector; S114: Splice the aligned train operation data feature vector, environment perception data feature vector, and passenger density data feature vector into a time series perception data set .

3. The visibility assessment method based on rail transit vehicle-mounted PIS according to claim 2 is characterized in that: The S2 includes: S21, based on the running tag signal output by the train main control system, divides the entire running cycle into several running state segments according to the time series, and extracts the synchronous trajectory data set in each running state segment ,in, For the moment the perceptual elements; S22, the perception elements in each running state segment Input the structure-aware fusion model and calculate the fused visibility index ; S23, in each running state segment Output continuous visibility index in time series , forming a scoring curve , and normalized.

4. A visibility assessment method based on rail transit vehicle-mounted PIS according to claim 3, characterized in that: The S21 includes: S211, obtain the operation tag sequence recorded by timestamp from the train main control system ,in, For the The timestamp of each tag record, For time point The train running status label at The total number of running tag sequences, and the running status fragment set is constructed based on the continuous time period of the same tag ; S212, let the sequence of perception elements during train operation be , Record timestamps for perception elements, For the moment The perceptual element vector of is the total length of the perception element sequence, for each running state segment , extract the perception elements within its time interval and construct the synchronous trajectory dataset of the segment .

5. The visibility assessment method based on rail transit vehicle-mounted PIS according to claim 4 is characterized in that: The S22 includes: S221, using minimum-maximum normalization to perceptual element vector All dimensions in the . S222: Use the structure perception fusion model to perform weighted combination on the normalized perception factor vectors to obtain the visibility index at a single moment. .

6. The visibility assessment method based on rail transit vehicle-mounted PIS according to claim 5 is characterized in that: The S23 includes: S231, from the running state fragment The calculated visibility index is sorted in chronological order from the perception elements within the image, and the visibility index time series is extracted. ; S232, running status fragment All visibility indices within Perform minimum-maximum normalization to obtain a normalized visibility score , and all normalized visibility scores form a scoring curve .

7. The visibility assessment method based on rail transit vehicle-mounted PIS according to claim 6 is characterized in that: The S3 includes: S31, based on the running status fragment Scoring curve within , set dynamic threshold , identify satisfaction The time interval is the visibility degradation section; S32, in each visibility degradation section Extract passenger behavior data within the corresponding time period, including the frequency of passenger complaint records, the number of emergency broadcast triggers, and the density of passenger re-inquiries, and record the behavior label feature set , expressed as: ； in, is the number of passenger complaints per unit time in the deteriorated section, is the number of emergency broadcast triggers per unit time, For the Behavioral characteristic vector of the degraded section; S33, based on label feature set , each visibility degradation segment is divided into scene categories, including high-reflection occlusion type, interactive failure type, and systemic failure type, and the scene category and the mean score are combined to form a , output visibility level , including excellent, medium and poor.

8. The visibility assessment method based on rail transit vehicle-mounted PIS according to claim 7 is characterized in that: The classification of scenes includes: Highly reflective blocking type: 、 、 ,in, A high threshold for complaints, is the broadcast high threshold; Interaction failure type: 、 ,in, is the score fluctuation threshold, For the The standard deviation of the visibility score curve in the visibility degradation section; Systemic failure type: 、 ; The visibility level is expressed as: 。 9. A visibility assessment system based on a rail transit vehicle-mounted PIS, for implementing a visibility assessment method based on a rail transit vehicle-mounted PIS according to any one of claims 1 to 8, characterized in that: Includes the following modules: Perception factor extraction module: Based on rail transit train operation data, carriage interior environment perception data, and passenger density distribution data, it extracts multi-dimensional perception factors including on-board display screen illumination, screen orientation angle, vehicle body pitch posture, and passenger occlusion area, and performs correlation modeling in the time and space dimensions to generate a PIS visibility influencing factor map; Structural Perception Fusion Module: This module inputs the PIS visibility influencing factor map into the structural perception fusion model. Combining the train's operating timeline with the synchronized trajectory of the sensor data within the carriage, it performs weighted processing on the perception factors, constructs a visibility feature mapping model under different operating states, and outputs the PIS visibility score curve for each operating stage. Visibility Assessment and Classification Module: This module sets dynamic visibility evaluation thresholds based on the PIS visibility scoring curve, identifies sections with degraded visibility, and classifies these sections based on passenger complaint records and emergency broadcast trigger frequencies in historical operating data. The module then outputs visibility zoning assessment results divided into excellent, medium, and poor grades.

Images