Railway passenger station elevator safety prediction method based on multi-index multi-branch fusion algorithm

Abstract

The invention discloses a railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm, and belongs to the technical field of elevator safety prediction. The railway passenger station elevator safety prediction method based on the multi-index multi-branch fusion algorithm comprises the steps that operation data, component state data, passenger behavior data and environment monitoring data of an elevator are collected, basic information of the elevator is summarized to construct an index system, and the safety of the elevator is predicted based on the constructed index system and the collected data. The method comprises the steps of classifying the safety state of an elevator, analyzing evaluation indexes and monitoring parameters of each category, generating and selecting new features based on the detection parameters of each category, constructing an operation safety branch model, a part safety branch model, a passenger safety branch model and an environment safety branch model, processing the constructed branch models through a multi-branch fusion algorithm, and obtaining the safety state of the elevator. According to the method, more reliable safety prediction is provided for the elevator under the high-frequency use condition by aiming at the use characteristics of the railway passenger station elevator and particularly considering the peak period.

Description

Technical Field

[0001] The present invention relates to the technical field of elevator safety prediction, and in particular to a railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm. Background Art

[0002] Elevator safety prediction technology is an important research field in modern building intelligence and urban management. With the acceleration of urbanization and the increase in the number of high-rise buildings, the safety and reliability of elevators as the main vertical transportation tool are particularly important. Traditional elevator maintenance mainly relies on regular inspections and repairs after sudden failures. This method is not only inefficient, but also unable to effectively prevent potential failures that may lead to serious accidents. Therefore, elevator safety prediction technology came into being.

[0003] At present, in the field of railway passenger stations, the performance indicators of elevators have not been studied in depth, and it is difficult to fully capture the complex conditions of elevators during operation. In addition, the existing elevator safety prediction methods are mostly based on single parameters or a small number of parameters for analysis, resulting in insufficient prediction accuracy and reliability. Especially for railway passenger stations, which carry a large number of passengers every day and are used frequently, safety prediction is particularly critical. The passenger flow of railway passenger stations has obvious peak period characteristics. During peak periods, the passenger flow is dense, the frequency of elevator use increases significantly, and the load also increases accordingly. In addition, sudden passenger flow is also relatively common, and elevators need to be able to cope with high-intensity and high-frequency usage needs. Summary of the invention

[0004] The object of the present invention is to provide a railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm.

[0006] A railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm comprises the following steps:

[0007] S1: Build an index system based on basic elevator information;

[0008] S2: Collect elevator operation data, component status data, passenger behavior data and environmental monitoring data;

[0009] S3: Based on the constructed indicator system and the collected elevator operation data, component status data, passenger behavior data and environmental monitoring data, the safety status of the elevator is classified and the evaluation indicators and monitoring parameters of each category are analyzed;

[0010] S4: Generate and select new features based on the detection parameters of each category;

[0011] S5: Construct operation safety branch, component safety branch, passenger safety branch and environmental safety branch models based on the generated new features;

[0012] S6: Train and evaluate the operation safety branch, component safety branch, passenger safety branch and environmental safety branch models, and combine the multi-branch fusion algorithm to predict the safety of railway passenger station elevators.

[0013] Further, in step S1: based on the basic information of the elevator, including operation information, component information, fault information, event information and statistical information, three groups of elevator indicator systems are constructed, namely, an operation indicator system, a management indicator system and a safety indicator system. The operation indicator system is constructed based on the operation information and component information, and monitors and analyzes the operation status, component operation status and passenger behavior pattern of the elevator in detail, and comprehensively evaluates the safety, efficiency and reliability of the elevator. The operation information refers to the real-time status, parameters and environmental data of the elevator, specifically including the following: operation status, service mode, operation speed, number of passengers, passenger behavior, machine room temperature and shaft temperature. The component information refers to the component and device data that constitutes and controls the operation and safety system of the elevator, specifically including the following: traction machine, car, control system, motor, power system, car door and unlocking area status;

[0014] The safety index system is built based on fault information and event information, recording and analyzing the frequency of abnormal operation, component failure, safety system failure and safety events, and comprehensively identifying various abnormalities and safety hazards in elevator operation. Fault information is to describe and record the faults and abnormal conditions in elevator operation, including the following: safety circuit break, door opening and closing failure, overspeed, power failure and high temperature alarm. Elevator event information is to describe and record the events and abnormal conditions in elevator operation, ensuring the safety and smoothness of elevator operation, including the following: abnormal behavior in the car, water ingress into the machine room and shaft, and abnormal temperature events;

[0015] The management indicator system is constructed based on basic information and statistical information. It collects and counts basic properties, operating time, energy consumption, maintenance cycle and number of switches to comprehensively evaluate the operation quality and maintenance of elevators. Basic information refers to the basic data and identification information of the elevator, including the following: registration code, category, model, factory and installation date, manufacturer, installation unit, factory number, rated speed and load capacity. Statistical information refers to the summary and analysis of data collected during the operation of the elevator in order to evaluate the performance and maintenance of the elevator, including the following: cumulative operating time and number of times, cumulative number of door openings and closings, power consumption statistics and maintenance cycle.

[0016] Furthermore, in step S2: the operation data, component status data, passenger behavior data and environmental monitoring data of the elevator are obtained from the elevator control system, monitoring equipment, sensors and event records; the operation data includes: operation status data, speed data, operation mode data and time data; the component safety data includes: operation time data, fault data and operation time data; the passenger behavior data includes: number of passengers, passenger flow and event alarm data; the environmental monitoring data includes: temperature and humidity data and environmental abnormality data.

[0017] Furthermore, in step S2: the collected data is normalized to ensure the quality and consistency of the data. Z-Score Normalization is used to normalize the data to a mean of 0 and a standard deviation of 1. The formula is as follows:

[0018]

[0019] Among them, X is the elevator running speed data set, X' is the normalized elevator running speed data, μ is the mean of the elevator running speed data set, and σ is the overall standard deviation of the elevator running speed data set;

[0020] The mean calculation formula is as follows:

[0021]

[0022] Among them, μ is the mean of the elevator speed data set, N is the number of elevator speed data sets, and X i Indicates the data value of the running speed of the i-th elevator;

[0023] The formula for calculating the population standard deviation of a data set is as follows:

[0024]

[0025] Among them, σ is the overall standard deviation of the elevator running speed data set, N represents the number of elevator running speed data sets, and X i represents the i-th elevator running speed data value, μ is the mean value of the elevator running speed data set;

[0026] Use the Min-Max Scaling method to scale the running speed data to the range of 0 to 1 to achieve normalization. The formula is as follows:

[0027]

[0028] Among them, X is the elevator running speed data set, X' is the normalized elevator running speed data set, and X min is the minimum value in the elevator speed data set, X maxIt is the maximum value in the elevator running speed data set.

[0029] Further, in step S3: based on the constructed indicator system and the collected elevator operation data, component status data, passenger behavior data and environmental monitoring data, the safety status of the elevator is classified, and the evaluation indicators and monitoring parameters of each category are analyzed;

[0030] The operational safety assessment indicators and specific monitoring parameters are as follows:

[0031] Evaluation indicators include: actual operating state change frequency and duration, actual operating speed deviation rate, operating mode switching and operating time information collection; specific monitoring parameters include: operating speed, actual operating speed, operating mode and cumulative operating time;

[0032] The component safety assessment indicators and specific monitoring parameters are as follows:

[0033] Evaluation indicators include: component wear degree and component failure rate; specific monitoring parameters include: component operation data and component failure data;

[0034] Passenger safety assessment indicators and specific monitoring parameters are as follows:

[0035] Evaluation indicators include: the impact of the number of passengers on operating efficiency and safety; specific monitoring parameters include: the number of passengers, real-time passenger flow, expected passenger flow and passenger event alarm;

[0036] The environmental safety assessment indicators and specific monitoring parameters are as follows:

[0037] Evaluation indicators include: analysis of the relationship between internal and external environmental temperature and humidity; specific monitoring parameters include: ambient temperature, humidity and environmental event alarms.

[0038] Further, in step S4: new features are generated and selected based on the detection parameters of each category:

[0039] The feature generation includes: operating state change rate, speed deviation rate, mode switching rate and accumulated operating information; component operating information and failure frequency; number of passengers and passenger flow; temperature and humidity change rate, temperature and humidity abnormalities and environmental abnormalities;

[0040] The feature selection includes: using the RFE recursive feature elimination method to select the operating state change rate, fault frequency, passenger density and temperature and humidity change rate from the generated multiple features.

[0041] Furthermore, in step S5: based on the generated new features, operation safety branch, component safety branch, passenger safety branch and environmental safety branch models are constructed, wherein the operation safety branch model includes processing and analyzing the operation status data of the elevator: the frequency and duration of operation status changes, the operation speed deviation rate and the operation mode switching status; the component safety branch includes processing and analyzing the status data of the elevator components: component operation time and fault data; the passenger safety branch includes processing and analyzing the passenger data: the number of passengers, passenger flow, safety incidents and abnormal behavior; the environmental safety branch includes processing and analyzing the data of the elevator operation environment: temperature and humidity and environmental anomaly monitoring.

[0042] Furthermore, in step S6: during the training of the multi-branch model, a method combining supervised learning and unsupervised learning is adopted, and historical data and real-time data are comprehensively used for training. The specific steps are as follows:

[0043] Divide the data set into training set, validation set and test set to ensure the generalization ability and prediction accuracy of the model;

[0044] According to different security assessment categories, LSTM, SVM, K-Means and ARIMA algorithms are used for model training:

[0045] The accuracy, stability, and interpretability of the model are evaluated using precision, recall, and F1-score indicators. When evaluating the elevator operation safety model, precision and recall are calculated to ensure that the model can accurately predict changes in the elevator's operating status.

[0046] The formula for calculating accuracy is as follows:

[0047]

[0048] Among them, P represents precision, TP represents the number of elevators predicted as "in operation", and FP represents the number of elevators incorrectly predicted as "in operation".

[0049] The formula for calculating recall is as follows:

[0050]

[0051] Where R represents the recall rate, TP represents the number of elevators predicted to be in the "running" state, and FN represents the number of elevators incorrectly predicted to be in the "stopped" state.

[0052] F1-score comprehensively measures the precision and recall of the model. The formula is as follows:

[0053]

[0054] Among them, F1-score represents the harmonic mean of precision and recall, P represents precision, and R represents recall.

[0055] Furthermore, in step S6: each branch model is fused by a weighted average method in the prediction stage, and the prediction results of each branch are combined to obtain the final elevator safety assessment result, wherein the prediction results of each branch are as follows:

[0056] The operation safety branch prediction results include: operation status prediction risk score, speed deviation prediction risk score, mode switching prediction risk score and maintenance prediction risk score;

[0057] The component safety branch prediction results include: component status prediction risk score and failure probability prediction risk score;

[0058] The passenger safety branch prediction results include: peak-hour operation efficiency prediction risk score, sudden passenger flow operation efficiency prediction risk score, and passenger behavior pattern prediction risk score;

[0059] The prediction results of the environmental safety branch include: environmental temperature and humidity prediction risk score and environmental anomaly prediction risk score;

[0060] The risk score of the running security branch is calculated by weighted average, and the formula is as follows:

[0061] R1=w1*ss1+w2*ss2+W3*ss3+w4*ss4.

[0062] Where R1 is the risk score of the running security branch, wi is the weight of the ith risk score, si is the ith risk score,

[0063] The risk scores of component safety branches are calculated by weighted average, and the formula is as follows:

[0064] R2=w1*s1+w2*s2;

[0065] Where R2 is the risk score of the component safety branch, wi is the weight of the ith risk score, si is the ith risk score,

[0066] The risk score of the passenger safety branch is calculated by taking a weighted average, as follows:

[0067] R3=w1*s1+w2*s2+w3*s3;

[0068] Where R3 is the risk score of the passenger safety branch, wi is the weight of the ith risk score, si is the ith risk score,

[0069] The risk scores of the environmental security branches are calculated by weighted average, and the formula is as follows:

[0070] R4=w1*s1+w2*s2;

[0071] Among them, R4 is the risk score of the environmental safety branch, wi is the weight of the i-th risk score, si is the i-th risk score,

[0072] According to the prediction accuracy and importance of each branch model, different weights are assigned to the prediction results of each branch model, and weighted average is performed to obtain a comprehensive risk score. The calculation formula for the comprehensive risk score is as follows:

[0073] B=(W1*R1)+(W2*R2)+(W3*R3)+(W4*R4)

[0074] Among them, B represents the comprehensive risk score, the weight of the operation safety branch is W1, and the comprehensive risk score is R1; the weight of the component safety branch is W2, and the comprehensive risk score is R2; the weight of the passenger safety branch is W3, and the comprehensive risk score is R3; the weight of the environmental safety branch is W4, and the comprehensive risk score is R4. Finally, the railway passenger station elevator safety prediction model based on the multi-branch fusion algorithm can comprehensively analyze multiple parameters and evaluation indicators, comprehensively evaluate the safety and operation risks of elevators, improve the accuracy and reliability of predictions, and ensure the safety of passengers during peak hours and emergencies. Travel.

[0075] A linear regression model is established and the regression coefficient is determined by the least squares method. The following are the detailed calculation steps and formulas:

[0076] Calculate the mean of feature variable X and target variable y:

[0077]

[0078]

[0079] Calculate the slope β1, the calculation formula is as follows:

[0080]

[0081] Calculate the intercept β0, the calculation formula is as follows:

[0082]

[0083] The regression equation calculation formula is as follows:

[0084] y = β0 + β1 * X;

[0085] Among them, y is the predicted target variable value, is the characteristic variable value, β0 and β1 are the intercept and slope respectively. The regression equation formula obtained by calculation is used to compare and analyze the degree of wear of ordinary passenger station elevators and railway passenger station elevators.

[0086] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0087] The present invention integrates multiple indicators for comprehensive analysis, which can more comprehensively and accurately reflect the operating status of the elevator and improve the accuracy and reliability of the prediction. In addition, the present invention introduces a multi-branch fusion algorithm, which can effectively capture the complex conditions and abnormal modes in the operation of the elevator through branch processing and fusion of different indicator data. Compared with the traditional single algorithm, the present invention has significant advantages in processing high-dimensional data and nonlinear relationships, and can more accurately predict elevator failures. In view of the unique usage characteristics of high passenger flow and high usage frequency of railway passenger station elevators, the present invention specially considers peak hours and sudden passenger flow factors, and provides more reliable safety predictions for elevators under high-intensity and high-frequency usage, effectively improving the safety and reliability of elevators. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0089] Figure 1 It is a framework diagram of the railway passenger station elevator safety prediction system of the present invention;

[0090] Figure 2 It is a schematic diagram of the construction of the multi-index system in the present invention;

[0091] Figure 3 It is a schematic diagram of the safety assessment sub-evaluation indicators and monitoring parameters in the present invention;

[0092] Figure 4 It is a feature engineering and extraction flow chart of the present invention;

[0093] Figure 5 It is a flow chart of the multi-branch fusion algorithm in the present invention;

[0094] Figure 6 is a flow chart of training and evaluation of the branch model in the present invention;

[0095] Figure 7 It is a flow chart of the prediction results of the multi-branch fusion algorithm in the present invention. DETAILED DESCRIPTION

[0096] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0097] See also Figure 1-Figure 7 , the present invention provides a technical solution:

[0098] A railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm comprises the following steps:

[0099] S1: Build an index system based on basic elevator information;

[0100] S2: Collect elevator operation data, component status data, passenger behavior data and environmental monitoring data;

[0101] S3: Based on the constructed indicator system and the collected elevator operation data, component status data, passenger behavior data and environmental monitoring data, the safety status of the elevator is classified and the evaluation indicators and monitoring parameters of each category are analyzed;

[0102] S4: Generate and select new features based on the detection parameters of each category;

[0103] S5: Construct operation safety branch, component safety branch, passenger safety branch and environmental safety branch models based on the generated new features;

[0104] S6: Train and evaluate the operation safety branch, component safety branch, passenger safety branch and environmental safety branch models, and combine the multi-branch fusion algorithm to predict the safety of railway passenger station elevators.

[0105] Further, in step S1: based on the basic information of the elevator, including operation information, component information, fault information, event information and statistical information, three groups of elevator indicator systems are constructed, namely, an operation indicator system, a management indicator system and a safety indicator system. The operation indicator system is constructed based on the operation information and component information, and monitors and analyzes the operation status, component operation status and passenger behavior pattern of the elevator in detail, and comprehensively evaluates the safety, efficiency and reliability of the elevator. The operation information refers to the real-time status, parameters and environmental data of the elevator, specifically including the following: operation status, service mode, operation speed, number of passengers, passenger behavior, machine room temperature and shaft temperature. The component information refers to the component and device data that constitutes and controls the operation and safety system of the elevator, specifically including the following: traction machine, car, control system, motor, power system, car door and unlocking area status;

[0106] The safety index system is built based on fault information and event information, recording and analyzing the frequency of abnormal operation, component failure, safety system failure and safety events, and comprehensively identifying various abnormalities and safety hazards in elevator operation. Fault information is to describe and record the faults and abnormal conditions in elevator operation, including the following: safety circuit break, door opening and closing failure, overspeed, power failure and high temperature alarm. Elevator event information is to describe and record the events and abnormal conditions in elevator operation, ensuring the safety and smoothness of elevator operation, including the following: abnormal behavior in the car, water ingress into the machine room and shaft, and abnormal temperature events;

[0107] The management indicator system is constructed based on basic information and statistical information. It collects and counts basic properties, operating time, energy consumption, maintenance cycle and number of switches to comprehensively evaluate the operation quality and maintenance of elevators. Basic information refers to the basic data and identification information of the elevator, including the following: registration code, category, model, factory and installation date, manufacturer, installation unit, factory number, rated speed and load capacity. Statistical information refers to the summary and analysis of data collected during the operation of the elevator in order to evaluate the performance and maintenance of the elevator, including the following: cumulative operating time and number of times, cumulative number of door openings and closings, power consumption statistics and maintenance cycle.

[0108] Furthermore, in step S2: the operation data, component status data, passenger behavior data and environmental monitoring data of the elevator are obtained from the elevator control system, monitoring equipment, sensors and event records; the operation data includes: operation status data, speed data, operation mode data and time data; the component safety data includes: operation time data, fault data and operation time data; the passenger behavior data includes: number of passengers, passenger flow and event alarm data; the environmental monitoring data includes: temperature and humidity data and environmental abnormality data.

[0109] Furthermore, in step S2: the collected data is normalized to ensure the quality and consistency of the data. Z-Score Normalization is used to normalize the data to a mean of 0 and a standard deviation of 1. The formula is as follows:

[0110]

[0111] Among them, X is the elevator running speed data set, X' is the normalized elevator running speed data, μ is the mean of the elevator running speed data set, and σ is the overall standard deviation of the elevator running speed data set;

[0112] The mean calculation formula is as follows:

[0113]

[0114] Among them, μ is the mean of the elevator speed data set, N is the number of elevator speed data sets, and X i Indicates the data value of the running speed of the i-th elevator;

[0115] The formula for calculating the population standard deviation of a data set is as follows:

[0116]

[0117] Among them, σ is the overall standard deviation of the elevator running speed data set, N represents the number of elevator running speed data sets, and X i represents the i-th elevator running speed data value, μ is the mean value of the elevator running speed data set;

[0118] Use the Min-Max Scaling method to scale the running speed data to the range of 0 to 1 to achieve normalization. The formula is as follows:

[0119]

[0120] Among them, X is the elevator running speed data set, X' is the normalized elevator running speed data set, and X min is the minimum value in the elevator speed data set, X max It is the maximum value in the elevator running speed data set.

[0121] Further, in step S3: based on the constructed indicator system and the collected elevator operation data, component status data, passenger behavior data and environmental monitoring data, the safety status of the elevator is classified, and the evaluation indicators and monitoring parameters of each category are analyzed;

[0122] The operational safety assessment indicators and specific monitoring parameters are as follows:

[0123] Evaluation indicators include: actual operating state change frequency and duration, actual operating speed deviation rate, operating mode switching and operating time information collection; specific monitoring parameters include: operating speed, actual operating speed, operating mode and cumulative operating time;

[0124] The component safety assessment indicators and specific monitoring parameters are as follows:

[0125] Evaluation indicators include: component wear degree and component failure rate; specific monitoring parameters include: component operation data and component failure data;

[0126] Passenger safety assessment indicators and specific monitoring parameters are as follows:

[0127] Evaluation indicators include: the impact of the number of passengers on operating efficiency and safety; specific monitoring parameters include: the number of passengers, real-time passenger flow, expected passenger flow and passenger event alarm;

[0128] The environmental safety assessment indicators and specific monitoring parameters are as follows:

[0129] Evaluation indicators include: analysis of the relationship between internal and external environmental temperature and humidity; specific monitoring parameters include: ambient temperature, humidity and environmental event alarms.

[0130] Further, in step S4: new features are generated and selected based on the detection parameters of each category:

[0131] The feature generation includes: operating state change rate, speed deviation rate, mode switching rate and accumulated operating information; component operating information and failure frequency; number of passengers and passenger flow; temperature and humidity change rate, temperature and humidity abnormalities and environmental abnormalities;

[0132] The feature selection includes: using the RFE recursive feature elimination method to select the operating state change rate, fault frequency, passenger density and temperature and humidity change rate from the generated multiple features.

[0133] Furthermore, in step S5: based on the generated new features, operation safety branch, component safety branch, passenger safety branch and environmental safety branch models are constructed, wherein the operation safety branch model includes processing and analyzing the operation status data of the elevator: the frequency and duration of operation status changes, the operation speed deviation rate and the operation mode switching status; the component safety branch includes processing and analyzing the status data of the elevator components: component operation time and fault data; the passenger safety branch includes processing and analyzing the passenger data: the number of passengers, passenger flow, safety incidents and abnormal behavior; the environmental safety branch includes processing and analyzing the data of the elevator operation environment: temperature and humidity and environmental anomaly monitoring.

[0134] Furthermore, in step S6: during the training of the multi-branch model, a method combining supervised learning and unsupervised learning is adopted, and historical data and real-time data are comprehensively used for training. The specific steps are as follows:

[0135] Divide the data set into training set, validation set and test set to ensure the generalization ability and prediction accuracy of the model;

[0136] According to different security assessment categories, LSTM, SVM, K-Means and ARIMA algorithms are used for model training:

[0137] The accuracy, stability, and interpretability of the model are evaluated using precision, recall, and F1-score indicators. When evaluating the elevator operation safety model, precision and recall are calculated to ensure that the model can accurately predict changes in the elevator's operating status.

[0138] The formula for calculating accuracy is as follows:

[0139]

[0140] Among them, P represents precision, TP represents the number of elevators predicted as "in operation", and FP represents the number of elevators incorrectly predicted as "in operation".

[0141] The formula for calculating recall is as follows:

[0142]

[0143] Where R represents the recall rate, TP represents the number of elevators predicted to be in the "running" state, and FN represents the number of elevators incorrectly predicted to be in the "stopped" state.

[0144] F1-score comprehensively measures the precision and recall of the model. The formula is as follows:

[0145]

[0146] Among them, F1-score represents the harmonic mean of precision and recall, P represents precision, and R represents recall.

[0147] Furthermore, in step S6: each branch model is fused by a weighted average method in the prediction stage, and the prediction results of each branch are combined to obtain the final elevator safety assessment result, wherein the prediction results of each branch are as follows:

[0148] The operation safety branch prediction results include: operation status prediction risk score, speed deviation prediction risk score, mode switching prediction risk score and maintenance prediction risk score;

[0149] The component safety branch prediction results include: component status prediction risk score and failure probability prediction risk score;

[0150] The passenger safety branch prediction results include: peak-hour operation efficiency prediction risk score, sudden passenger flow operation efficiency prediction risk score, and passenger behavior pattern prediction risk score;

[0151] The prediction results of the environmental safety branch include: environmental temperature and humidity prediction risk score and environmental anomaly prediction risk score;

[0152] The risk score of the running security branch is calculated by weighted average, and the formula is as follows:

[0153] R1=w1*s1+w2*s2+w3*s3+w4*s4;

[0154] Where R1 is the risk score of the running security branch, wi is the weight of the ith risk score, si is the ith risk score,

[0155] The risk scores of component safety branches are calculated by weighted average, and the formula is as follows:

[0156] R2=w1*s1+w2*s2;

[0157] Where R2 is the risk score of the component safety branch, wi is the weight of the i-th risk score, si is the i-th risk score,

[0158] The risk score of the passenger safety branch is calculated by taking a weighted average, as follows:

[0159] R3=w1*s1+w2*s2+w3*s3;

[0160] Where R3 is the risk score of the passenger safety branch, wi is the weight of the ith risk score, si is the ith risk score,

[0161] The risk scores of the environmental security branches are calculated by weighted average, and the formula is as follows:

[0162] R4=w1*s1+w2*s2;

[0163] Among them, R4 is the risk score of the environmental safety branch, wi is the weight of the i-th risk score, si is the i-th risk score,

[0164] According to the prediction accuracy and importance of each branch model, different weights are assigned to the prediction results of each branch model, and weighted average is performed to obtain a comprehensive risk score. The calculation formula for the comprehensive risk score is as follows:

[0165] B=(W1*R1)+(W2*R2)+(W3*R3)+(W4*R4)

[0166] Among them, B represents the comprehensive risk score, the weight of the operation safety branch is W1, and the comprehensive risk score is R1; the weight of the component safety branch is W2, and the comprehensive risk score is R2; the weight of the passenger safety branch is W3, and the comprehensive risk score is R3; the weight of the environmental safety branch is W4, and the comprehensive risk score is R4. Finally, the railway passenger station elevator safety prediction model based on the multi-branch fusion algorithm can comprehensively analyze multiple parameters and evaluation indicators, comprehensively evaluate the safety and operation risks of elevators, improve the accuracy and reliability of predictions, and ensure the safety of passengers during peak hours and emergencies. Travel.

[0167] A linear regression model is established and the regression coefficient is determined by the least squares method. The following are the detailed calculation steps and formulas:

[0168] Calculate the mean of feature variable X and target variable y:

[0169]

[0170] Calculate the slope β1, the calculation formula is as follows:

[0171]

[0172] Calculate the intercept β0, the calculation formula is as follows:

[0173]

[0174] The regression equation calculation formula is as follows:

[0175] y = β0 + β1 * X;

[0176] Among them, y is the predicted target variable value, is the characteristic variable value, β0 and β1 are the intercept and slope respectively. The regression equation formula obtained by calculation is used to compare and analyze the degree of wear of ordinary passenger station elevators and railway passenger station elevators.

[0177] Example 1: Collected operation safety data, railway station elevator data:

[0178] time Running status Uplink speed Downlink speed Operation Mode Maintenance requirements 2024-07-01 08：00 Running 1.2 1.1 Normal mode none 2024-07-01 08：05 stop 0.0 0.0 Normal mode none 2024-07-01 08：10 Running 1.3 1.2 Peak Mode none 2024-07-01 08：15 Running 1.2 1.0 Peak Mode have 2024-07-01 08：20 stop 0.0 0.0 Normal mode none

[0179] Component safety data, railway station elevator data:

[0180] part Run time Fault Fault description Motor 500 no none pulley 450 yes Pulley wear Wire Rope 600 no none Control System 400 yes Control system failure Door systems 550 no none

[0181] Ordinary passenger station elevator data

[0182] part Run time Fault Fault description Motor 300 no none pulley 280 no none Wire Rope 350 no none Control System 320 no none Door systems 300 no none

[0183] Passenger safety data, railway station elevator data:

[0184] time Number of passengers Customer flow Event Alarm 2024-07-01 08：00 50 1000 none 2024-07-01 08：05 55 1100 none 2024-07-01 08：10 60 1200 Overload alarm 2024-07-01 08：15 65 1300 none 2024-07-01 08：20 45 900 none

[0185] Environmental safety data, railway station elevator data:

[0186] time temperature humidity Abnormal environment 2024-07-01 08：00 30 55 none 2024-07-01 08：05 31 56 none 2024-07-01 08：10 32 57 high temperature 2024-07-01 08：15 31 55 none 2024-07-01 08：20 30 54 none

[0187] Normalize the collected data and use Z-Score Normalization to normalize the data to a mean of 0 and a standard deviation of 1 to achieve normalization. Calculate the mean and standard deviation:

[0188] Get a set of uplink speed data: X = {1.2, 1.3, 1.1, 1.4, 1.0},

[0189] Calculate the sum of the upstream speed data:

[0190] Calculate the number of uplink speed data: N = 5,

[0191] Calculate the mean uplink speed μ:

[0192] Overall standard deviation of uplink speed σ:

[0193] The mean μ of the downlink speed is 1.1, and the overall standard deviation σ of the downlink speed is 0.141. The mean and overall standard deviation of the temperature are calculated: the mean of the temperature is 30.8, and the overall standard deviation is 0.748;

[0194] Normalized calculation:

[0195] Calculate the normalized data X' for the uplink speed of 1.2:

[0196] The normalized data of the upstream speed of 1.3 is 0.71, the normalized data of the upstream speed of 1.1 is -0.71, the normalized data of the upstream speed of 1.4 is 1.41, and the normalized data of the upstream speed of 1.0 is -1.41.

[0197] The normalized data of the downstream speed of 1.1 is 0.0, the normalized data of the downstream speed of 1.2 is 0.71, the normalized data of the downstream speed of 1.0 is -0.71, the normalized data of the downstream speed of 1.3 is 1.41, and the normalized data of the downstream speed of 0.9 is -1.41.

[0198] The normalized data for the calculated temperature of 30 is -1.07, the normalized data for the calculated temperature of 31 is 0.27, the normalized data for the calculated temperature of 30=2 is 1.6, the normalized data for the calculated temperature of 31 is 0.27, and the normalized data for the calculated temperature of 30 is -1.07.

[0199] Then use the Min-Max Scaling method to normalize the Z-Score result to the range of 0-1:

[0200] Determine the minimum and maximum values ​​of the Z-Score data:

[0201] Uplink speed Z-Score data: 0.0, 0.71, -0.71, 1.41, -1.41, minimum: -1.41, maximum: 1.41,

[0202] Downlink speed Z-Score data: 0.0, 0.71, -0.71, 1.41, -1.41, minimum: -1.41, maximum: 1.41,

[0203] use Normalize the Z-Score data of the upstream speed, the downstream speed, and the temperature:

[0204] Normalize the uplink speed and get the normalized uplink speed data: 0.5, 0.75, 0.25, 1.0, 0.0,

[0205] Normalize the downlink speed to get the normalized downlink speed data: 0.5, 0.75, 0.25, 1.0, 0.0,

[0206] Normalize the temperature data and get the normalized temperature data: 0.12, 0.59, 1.0, 0.59, 0.21,

[0207] The railway passenger station elevator data in the component safety data is used to calculate the β1=-0.006,β0=3.4,so the regression equation of railway passenger station elevator is: y=3.4-0.006*X;

[0208] The common passenger station elevator data in the component safety data is used to calculate the common passenger station elevator β1=0, β0=0, so the regression equation of the ordinary passenger station elevator is: y=0; from the regression equation, it can be seen that the slope of the railway passenger station elevator is negative, which means that as the running time increases, the probability of failure decreases slightly; while the slope of the ordinary passenger station elevator is 0, indicating that its running time has no relationship with the probability of failure, that is, no matter how long it runs, the probability of failure is always 0.

[0209] Calculate the wear index at different operating times of 100 hours, 300 hours, 500 hours and 700 hours, and compare the wear index of the two at the same operating time:

[0210] The regression equation for the railway station elevator is: y = 3.4-0.006*X,

[0211] Running time X = 100 hours, y = 3.4-0.006*100 = 2.8, running time X = 300 hours, y = 3.4-0.006*300 = 1.6,

[0212] Running time X = 500 hours, y = 3.4-0.006*500 = 0.4, running time X = 700 hours, y = 3.4-0.006*700 = -0.8,

[0213] Since the wear index cannot be negative, the minimum value of the wear index is 0;

[0214] The regression equation of the ordinary passenger station elevator is: y = 0,

[0215] Running time X = 100 hours, y = 0, running time X = 300 hours, y = 0, running time X = 500 hours, y = 0, running time X = 700 hours, y = 0,

[0216] By comparing the wear index under the same running time, the following conclusions can be drawn:

[0217] Railway station elevators have more significant wear: Under the same operating time, the wear index of railway station elevators is significantly higher than that of ordinary passenger station elevators, indicating that the parts of railway station elevators have more significant wear.

[0218] Ordinary passenger station elevators have no obvious wear: The wear index of ordinary passenger station elevators is always 0, indicating that the probability of failure under different operating times is not affected by the operating time.

[0219] Data division: Divide the data set into training set, validation set and test set to ensure the generalization ability and prediction accuracy of the model. Divide the elevator operation safety data into training set, validation set and test set in a ratio of 7:2:1, of which 70% is used for training the model, 20% is used for validating the model, and 10% is used for testing the model to ensure that the model performs consistently on different data sets.

[0220] Model selection: According to different security assessment categories, use LSTM, SVM, K-Means, and ARIMA algorithms for model training:

[0221] Use the LSTM model to train elevator operation status data and predict future operation status changes;

[0222] time Running status Future state of motion 2024-07-01 08：00 Running In motion 2024-07-01 08：05 stop stop 2024-07-01 08：10 Running In motion 2024-07-01 08：15 Running In motion 2024-07-01 08：20 stop stop

[0223] Use the SVM model to train elevator component failure data and predict future failures;

[0224] part Fault Future failures Motor no no pulley yes yes Wire Rope no no Control System yes yes Door systems no no

[0225] Use K-Means model to train passenger flow data for passenger flow cluster analysis;

[0226] Normalized passenger flow Actual situation Clustering results 0.0 Peak hours Peak hours 0.5 peak hours peak hours 0.25 Peak hours Peak hours 0.375 Peak hours Peak hours 0.625 peak hours peak hours

[0227] Use the ARIMA model to train environmental temperature and humidity data to predict future environmental change trends;

[0228] time Predicted temperature 2024-07-01 08：25 0.6 2024-07-01 08：30 0.8 2024-07-01 08：35 0.7

[0229] The accuracy of the model is evaluated using precision, recall and F1-score indicators. Each branch model is fused through the weighted average method in the prediction stage, and the prediction results of each branch are combined to obtain the final elevator safety evaluation result. The prediction results of each branch are as follows:

[0230] The prediction results of the operation safety branch include: the operation state prediction risk score is 0.2, the speed deviation prediction risk score is 0.5, the mode switching prediction risk score is 0.2 and the maintenance prediction risk score is 0.6;

[0231] The component safety branch prediction results include: component status prediction risk score is 0.3 and failure probability prediction risk score is 0.8;

[0232] The passenger safety branch prediction results include: peak hour operation efficiency prediction risk score is 0.5, sudden passenger flow operation efficiency prediction risk score is 0.7 and passenger behavior pattern prediction risk score is 0.2;

[0233] The prediction results of the environmental safety branch include: the environmental temperature and humidity prediction risk score is 0.2 and the environmental anomaly prediction risk score is 0.4;

[0234] According to the operation status prediction risk score of 0.2, the speed deviation prediction risk score of 0.5, the mode switching prediction risk score of 0.2 and the maintenance prediction risk score of 0.6, substitute into the formula R1 = w1*s1+w2*s2+w3*s3+w4*s4 = 0.4,

[0235] According to the component status prediction risk score of 0.3 and the failure probability prediction risk score of 0.8, substitute into the formula R2 = w1*s1+w2*s2 = 0.55,

[0236] According to the peak period operation efficiency prediction risk score of 0.5, the sudden passenger flow operation efficiency prediction risk score of 0.7 and the passenger behavior pattern prediction risk score of 0.2, substitute into the formula R3 = w1*s1+w2*s2+w3*s3 = 0.467,

[0237] According to the environmental temperature and humidity prediction risk score of 0.2 and the environmental anomaly prediction risk score of 0.4, substitute it into the formula R4=w1*s1+w2*s2=0.3,

[0238] According to the importance of each branch, the following weights are assigned:

[0239] The weight of the operational safety branch is W1: 0.4, the weight of the component safety branch is W2: 0.3, the weight of the passenger safety branch is W3: 0.2, and the weight of the environmental safety branch is W4: 0.1.

[0240] According to the prediction accuracy and importance of each branch model, different weights are assigned to the prediction results of each branch model, and weighted average is performed to obtain a comprehensive risk score. The calculation formula for the comprehensive risk score is as follows:

[0241] B=(W1*R1)+(W2*R2)+(W3*R3)+(W4*R4)

[0242] Among them, B represents the comprehensive risk score. The weight of the operational safety branch is W1, and the comprehensive risk score is R1; the weight of the component safety branch is W2, and the comprehensive risk score is R2; the weight of the passenger safety branch is W3, and the comprehensive risk score is R3; the weight of the environmental safety branch is W4, and the comprehensive risk score is R4.

[0243] After substituting the specific values, the comprehensive risk score is 0.4484. Based on this score, the following conclusions can be drawn:

[0244] Low risk (0-0.3): The elevator operates safely and does not require special maintenance.

[0245] Medium risk (0.3-0.6): The elevator has a certain operating risk and preventive maintenance is recommended.

[0246] High risk (0.6-1.0): The elevator operation risk is high and requires immediate inspection and maintenance.

[0247] In this example, the overall risk score is 0.4484, which is in the medium risk range, and preventive maintenance is recommended.

[0248] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0249] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm, characterized in that: The following steps are involved: S1: Build an index system based on basic elevator information; S2: Collect elevator operation data, component status data, passenger behavior data and environmental monitoring data; S3: Based on the constructed indicator system and the collected elevator operation data, component status data, passenger behavior data and environmental monitoring data, the safety status of the elevator is classified and the evaluation indicators and monitoring parameters of each category are analyzed; S4: Generate and select new features based on the detection parameters of each category; S5: Construct operation safety branch, component safety branch, passenger safety branch and environmental safety branch models based on the generated new features; S6: Train and evaluate the operation safety branch, component safety branch, passenger safety branch and environmental safety branch models, and combine the multi-branch fusion algorithm to predict the safety of railway passenger station elevators.

2. The railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm according to claim 1 is characterized in that: In step S1: based on the basic information of the elevator, including operation information, component information, fault information, event information and statistical information, three sets of elevator indicator systems are constructed, namely, an operation indicator system, a management indicator system and a safety indicator system.

3. The railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm according to claim 1 is characterized in that: In step S2: obtaining elevator operation data, component status data, passenger behavior data and environmental monitoring data from the elevator control system, monitoring equipment, sensors and event records; The operation data includes: operation status data, speed data, operation mode data and time data; the component safety data includes: operation time data, fault data and operation time data; the passenger behavior data includes: number of passengers, passenger flow and event alarm data; the environmental monitoring data includes: temperature and humidity data and environmental abnormality data.

4. The method for predicting the safety of railway passenger station elevators based on a multi-index multi-branch fusion algorithm according to claim 1 is characterized in that: In step S2: the collected data is normalized using Z-ScoreNormalization to normalize the data to a mean of 0 and a standard deviation of 1. The formula is as follows: Among them, X is the elevator running speed data set, X' is the normalized elevator running speed data, μ is the mean of the elevator running speed data set, and σ is the overall standard deviation of the elevator running speed data set; The mean calculation formula is as follows: Among them, μ is the mean of the elevator speed data set, N is the number of elevator speed data sets, and X i Indicates the data value of the running speed of the i-th elevator; The formula for calculating the population standard deviation of a data set is as follows: Among them, σ is the overall standard deviation of the elevator running speed data set, N represents the number of elevator running speed data sets, and X i represents the i-th elevator running speed data value, μ is the mean value of the elevator running speed data set; Use the Min-Max Sca1ing method to scale the running speed data to the range of 0 to 1 to achieve normalization. The formula is as follows: Among them, X is the elevator running speed data set, X' is the normalized elevator running speed data set, and X min is the minimum value in the elevator speed data set, X max It is the maximum value in the elevator running speed data set.

5. The railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm according to claim 1 is characterized in that: In step S3: based on the constructed indicator system and the collected elevator operation data, component status data, passenger behavior data and environmental monitoring data, the safety status of the elevator is classified, and the evaluation indicators and monitoring parameters of each category are analyzed; The operational safety assessment indicators and specific monitoring parameters are as follows: Evaluation indicators include: actual operating state change frequency and duration, actual operating speed deviation rate, operating mode switching and operating time information collection; specific monitoring parameters include: operating speed, actual operating speed, operating mode and cumulative operating time; The component safety assessment indicators and specific monitoring parameters are as follows: Evaluation indicators include: component wear degree and component failure rate; specific monitoring parameters include: component operation data and component failure data; Passenger safety assessment indicators and specific monitoring parameters are as follows: Evaluation indicators include: the impact of the number of passengers on operating efficiency and safety; specific monitoring parameters include: the number of passengers, real-time passenger flow, expected passenger flow and passenger event alarm; The environmental safety assessment indicators and specific monitoring parameters are as follows: Evaluation indicators include: analysis of the relationship between internal and external environmental temperature and humidity; specific monitoring parameters include: ambient temperature, humidity and environmental event alarms.

6. The railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm according to claim 1 is characterized in that: In step S4: Generate and select new features based on the detection parameters of each category: The feature generation includes: operating state change rate, speed deviation rate, mode switching rate and accumulated operating information; component operating information and failure frequency; number of passengers and passenger flow; temperature and humidity change rate, temperature and humidity abnormalities and environmental abnormalities; The feature selection includes: using the RFE recursive feature elimination method to select the operating state change rate, fault frequency, passenger density and temperature and humidity change rate from the generated multiple features.

7. The railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm according to claim 1 is characterized in that: In step S5: constructing operation safety branch, component safety branch, passenger safety branch and environment safety branch models based on the generated new features; the operation safety branch model includes processing and analyzing the operation status data of the elevator: the frequency and duration of operation status changes, the operation speed deviation rate and the operation mode switching situation; The component safety branch includes processing and analyzing status data of elevator components: component operating time and fault data; The passenger safety branch includes processing and analyzing passenger data: number of passengers, passenger flow, safety incidents and abnormal behavior; The environmental safety branch includes processing and analyzing data on the elevator operating environment: temperature, humidity and environmental anomaly monitoring.

8. The railway passenger station elevator safety prediction method based on a multi-index multi-branch fusion algorithm according to claim 1 is characterized in that: In step S6: during the training of the multi-branch model, a method combining supervised learning and unsupervised learning is adopted, and historical data and real-time data are comprehensively used for training. The specific steps are as follows: Divide the elevator safety data into training set, validation set and test set; Use LSTM, SVM, K-Means and ARIMA algorithms to train the operation safety branch, component safety branch, passenger safety branch and environmental safety branch models; The accuracy of the model is evaluated using precision, recall, and F1-score indicators. The formula for calculating precision is as follows: Among them, P represents precision, TP represents the number of elevators predicted as "in operation", and FP represents the number of elevators incorrectly predicted as "in operation"; The formula for calculating recall is as follows: Where R represents the recall rate, TP represents the number of elevators predicted to be in the "running" state, and FN represents the number of elevators incorrectly predicted to be in the "stopped" state; F1-score comprehensively measures the precision and recall of the model. The formula is as follows: Among them, F1-score represents the harmonic mean of precision and recall, P represents precision, and R represents recall.

9. The method for predicting the safety of railway passenger station elevators based on a multi-index multi-branch fusion algorithm according to claim 1 is characterized in that: In step S6: the prediction results of each branch model are integrated by weighted average method, wherein the prediction results of each branch are integrated as follows: The operation safety branch prediction results include: operation status prediction risk score, speed deviation prediction risk score, mode switching prediction risk score and maintenance prediction risk score; The component safety branch prediction results include: component status prediction risk score and failure probability prediction risk score; The passenger safety branch prediction results include: peak-hour operation efficiency prediction risk score, sudden passenger flow operation efficiency prediction risk score, and passenger behavior pattern prediction risk score; The prediction results of the environmental safety branch include: environmental temperature and humidity prediction risk score and environmental anomaly prediction risk score; The risk score of the running security branch is calculated by weighted average, and the formula is as follows: R1=w1*s1+w2*s2+w3*s3+w4*s4; Where R1 is the risk score of the running security branch, wi is the weight of the i-th risk score, and si is the i-th risk score; The risk scores of component safety branches are calculated by weighted average, and the formula is as follows: R2=w1*s1+w2*s2; Where R2 is the risk score of the component safety branch, wi is the weight of the i-th risk score, and si is the i-th risk score; The risk score of the passenger safety branch is calculated by taking a weighted average, as follows: R3=w1*s1+w2*s2+w3*s3; Where R3 is the risk score of the passenger safety branch, wi is the weight of the i-th risk score, and si is the i-th risk score; The risk scores of the environmental security branches are calculated by weighted average, and the formula is as follows: R4=w1*s1+w2*s2; Among them, R4 is the risk score of the environmental safety branch, wi is the weight of the i-th risk score, and si is the i-th risk score; According to the prediction accuracy and importance of each branch model, different weights are assigned to the prediction results of each branch model, and weighted average is performed to obtain a comprehensive risk score. The calculation formula for the comprehensive risk score is as follows: B=(W1*R1)+(W2*R2)+(W3*R3)+(W4*R4) Among them, B represents the comprehensive risk score. The weight of the operation safety branch is W1, and the comprehensive risk score is R1; the weight of the component safety branch is W2, and the comprehensive risk score is R2; the weight of the passenger safety branch is W3, and the comprehensive risk score is R3; the weight of the environmental safety branch is W4, and the comprehensive risk score is R4.

Images