CN-121117518-B - Railway ticket station-level safety operation and maintenance method and platform based on RWKV model

CN121117518BCN 121117518 BCN121117518 BCN 121117518BCN-121117518-B

Abstract

The invention discloses a railway ticket station-level safety operation and maintenance method and platform based on RWKV models, which comprise the steps of collecting multi-source heterogeneous time sequence data generated in the operation of ticket terminal equipment, preprocessing the collected multi-source heterogeneous time sequence data, inputting the preprocessed multi-source heterogeneous time sequence data into a pre-trained RWKV model, utilizing a time mixing module and a channel mixing module of the RWKV model to conduct reasoning, outputting a quantization index for evaluating the health state of the equipment, wherein the quantization index comprises a fault early warning probability value for representing the health state of the equipment, generating and executing operation and maintenance instructions based on the quantization index, and the operation and maintenance instructions comprise the functions of triggering grading alarm, generating a treatment work order or starting and stopping the equipment. The railway ticket station-level safety operation and maintenance platform based on RWKV model has the advantages that the fault diagnosis efficiency and accuracy can be improved, the high availability and safety of a railway ticket vending system are guaranteed, and the operation and maintenance efficiency is remarkably improved.

Inventors

QI JIANHUAI
HU JINHUA
XU GUOQIAN
ZHENG WEIFAN
HAN DANDAN
CHENG YANG

Assignees

深圳市永达电子信息股份有限公司

Dates

Publication Date: 20260512
Application Date: 20251113

Claims (9)

1. The railway ticket station-level safety operation and maintenance method based on RWKV model is characterized by comprising the following steps: The method comprises the steps of collecting multi-source heterogeneous time sequence data generated in the operation of passenger ticket terminal equipment, wherein the passenger ticket terminal equipment comprises a self-service ticket vending machine and a ticket checking gate; The preprocessing comprises analyzing text log data, converting the original log into normalized word sequence by dynamic variable replacement and word segmentation filtering, performing normalization, time alignment, missing value interpolation and Z-score normalization on the digital sensor data to eliminate dimension influence, constructing high-order time sequence characteristics, and calculating statistical characteristics and frequency domain characteristics in a sliding window based on the normalized digital data; inputting the preprocessed multi-source heterogeneous time sequence data into a pre-trained RWKV model, and carrying out joint reasoning by utilizing a time mixing module and a channel mixing module of the RWKV model to output a quantization index for evaluating the health state of equipment, wherein the construction of the RWKV model comprises the steps of adopting a time attenuation factor to dynamically adjust the weight of historical information so as to enhance the memory of a long-term key event of the equipment; receiving and analyzing a quantization index output by the RWKV model, wherein the quantization index comprises a fault early warning probability value representing the health state of equipment; And generating and executing operation and maintenance instructions based on the quantitative indexes, wherein the operation and maintenance instructions comprise triggering grading alarm, generating a treatment work order or starting and stopping equipment function, and triggering equipment temporary degradation strategy during high risk early warning.
2. The railway ticket station-level security operation and maintenance method based on RWKV model as in claim 1, wherein the construction of RWKV model further comprises the step of training the model using historical time series data containing equipment failure labels.
3. The RWKV model-based railway ticket station-level security operation and maintenance method according to claim 2 is characterized in that the time mixing module and the channel mixing module of the RWKV model are utilized to carry out joint reasoning, specifically, outputs of a RWKV submodel, an LSTM-CNN submodel, a support vector machine SVM submodel and a random forest submodel are subjected to weighted fusion through a multi-model fusion system so as to generate the fault early warning probability value in a reasoning way.
4. The railway ticket station level security operation and maintenance method based on RWKV model as claimed in claim 1, further comprising the steps of: introducing a device fault case library and device structure knowledge, wherein the device fault case library comprises a historical fault type, corresponding fault characteristics and a fault processing scheme; Adopting a CNN-transducer model to process waveform data when faults occur, and extracting local and global features of a fault waveform graph to form fault feature vectors; And calculating the similarity between the fault characteristic vector and a case in the equipment fault case library, and carrying out reverse deduction by combining the equipment structure knowledge so as to accurately position a fault module.
5. The railway ticket station-level security operation and maintenance method based on RWKV model as set forth in claim 1, further comprising the step of establishing a station security risk assessment model by using personnel activity data, equipment state data and environmental factor data and adopting a fuzzy comprehensive evaluation method or a hierarchical analysis method.
6. The railway ticket station-level security operation and maintenance method based on RWKV model as set forth in claim 1, further comprising the steps of periodically collecting feedback data generated during execution of operation and maintenance operations, and performing incremental training on the RWKV model with the feedback data to optimize model parameters, predicting future traffic load peaks according to historical traffic data, and automatically adjusting computing resources, spare part inventory, or personnel scheduling plans according to the prediction results.
7. A railway ticket station level security operation and maintenance platform based on RWKV model, which is used for realizing the railway ticket station level security operation and maintenance method based on RWKV model according to any one of claims 1 to 6, comprising: The system comprises a station-level ticket system, a data acquisition layer, a ticket terminal device and a network communication layer, wherein the station-level ticket system is used for acquiring multi-source heterogeneous time sequence data of the station-level ticket system, the ticket terminal device comprises a self-service ticket vending machine and a ticket checking gate; The RWKV model reasoning layer is deployed on a station edge server and comprises a pre-trained RWKV model, wherein the pre-trained RWKV model is used for carrying out real-time reasoning on the preprocessed data and outputting fault probability and risk early warning, and the RWKV model is constructed by adopting a time attenuation factor to dynamically adjust the weight of historical information so as to enhance the memory of a long-term key event of equipment; the safety operation and maintenance service layer provides functions of equipment state monitoring, fault intelligent diagnosis, safety event handling and operation and maintenance resource scheduling and is used for driving a closed-loop operation and maintenance flow according to the output result of the RWKV model reasoning layer; And the man-machine interaction layer is used for displaying early warning information, operation and maintenance work orders and statistical reports and supporting voice interaction with operation and maintenance personnel.
8. The railway ticket station-level security operation and maintenance platform based on RWKV model as claimed in claim 7, wherein the data acquisition layer specifically comprises: the equipment state monitoring module is used for collecting the operation parameters of the ticket vending terminal and the ticket gate; the environment monitoring module is used for monitoring the equipment operation environment through a temperature and humidity sensor and a smoke sensor; the network probe module is used for collecting network traffic, port access records and abnormal log-in logs; And the data preprocessing module is used for cleaning, converting, integrating and reducing the acquired multi-source heterogeneous data to form a standardized input feature vector.
9. The railway ticket station-level security operation and maintenance platform based on RWKV model according to claim 8, the RWKV model reasoning layer specifically includes: the model training module is used for training the RWKV model by using historical equipment time sequence data and using Focal Loss as a Loss function; The real-time reasoning engine is used for inputting the preprocessed feature vector into the trained RWKV model, capturing a time sequence dependency relationship through the time mixing module, performing cross-feature correlation analysis through the channel mixing module, and outputting the health state score and the fault probability of the equipment; The RWKV model reasoning layer is also integrated with a multi-model fusion module, and the multi-model fusion module is used for receiving the output results of the RWKV submodel, the machine learning submodel based on SVM and random forest and the deep learning submodel based on LSTM-CNN.

Description

Railway ticket station-level safety operation and maintenance method and platform based on RWKV model Technical Field The invention relates to the technical field of railway informatization, in particular to a railway ticket station-level safety operation and maintenance method based on RWKV models. In addition, the invention also relates to a railway ticket station-level safety operation and maintenance platform based on RWKV model. Background The railway ticket system is a core component of key traffic infrastructure, station-level nodes bear key functions such as ticket selling terminal management, local transaction processing, data synchronous verification and the like, and the stable operation of terminal equipment such as self-service ticket vending machines, ticket gate checking machines and the like deployed in the nodes directly relates to transportation order and passenger travel experience. At present, the safety operation and maintenance of railway ticket station level mainly depends on a monitoring system based on fixed rules and manual experience, and the prior art scheme is generally composed of the following parts of firstly collecting state data, network flow and operation logs of equipment through deployed sensors and a log system, secondly carrying out abnormality judgment by utilizing a preset static threshold value, and finally generating an alarm and carrying out fault diagnosis and treatment depending on operation and maintenance personnel intervention after the threshold value is triggered. However, the prior art scheme has inherent structural defects, the prior art scheme depends on a static threshold value, long-sequence multi-source heterogeneous time sequence data generated during the running of equipment cannot be effectively processed, meanwhile, the fault diagnosis efficiency is low, the accuracy is poor, passenger detention is easy to occur in a passenger transport peak period, namely, the prior art has the technical problems of early warning lag, low diagnosis efficiency and insufficient intellectualization. Therefore, how to provide a railway ticket station-level safety operation and maintenance method based on RWKV model, which can improve the fault diagnosis efficiency and accuracy, and improve the operation and maintenance efficiency through an automatic process, and ensure the high availability and safety of a railway ticket vending system, is a technical problem to be solved by those skilled in the art. Disclosure of Invention In order to solve the technical problems, the invention provides a railway ticket station-level safety operation and maintenance method based on RWKV models, which can improve fault diagnosis efficiency and accuracy, improve operation and maintenance efficiency through an automatic process, ensure high availability and safety of a railway ticket vending system, and also provides a railway ticket station-level safety operation and maintenance platform based on RWKV models, and the method has the same beneficial effects. The technical scheme provided by the invention is as follows: The invention provides a railway ticket station-level safety operation and maintenance method based on RWKV model, which comprises the steps of collecting multi-source heterogeneous time sequence data generated in the running of ticket terminal equipment; Preprocessing the acquired multi-source heterogeneous time sequence data; Inputting the preprocessed multi-source heterogeneous time sequence data into a pre-trained RWKV model, and utilizing a time mixing module and a channel mixing module of the RWKV model to perform joint reasoning and output a quantization index for evaluating the health state of equipment; receiving and analyzing a quantization index output by the RWKV model, wherein the quantization index comprises a fault early warning probability value representing the health state of equipment; and generating and executing operation and maintenance instructions based on the quantization indexes, wherein the operation and maintenance instructions comprise the steps of triggering a grading alarm, generating a treatment work order or starting and stopping equipment functions. Further, in a preferred mode of the invention, preprocessing the acquired multi-source heterogeneous time sequence data comprises analyzing text log data, converting an original log into a normalized word sequence through dynamic variable replacement and word segmentation filtration, normalizing digital sensor data, performing time alignment, missing value interpolation and Z-score normalization processing to eliminate dimension influence, constructing high-order time sequence characteristics, and calculating statistical characteristics and frequency domain characteristics of the digital sensor data in a sliding window based on the normalized digital data. Further, in a preferred mode of the invention, the construction of the RWKV model comprises the steps of adopting a time attenuation facto