Rail transit exception processing and rail transit exception processing model training method

Abstract

This invention discloses a method for handling rail transit anomalies and training a rail transit anomaly handling model. The method includes: acquiring a knowledge base of current train anomalies and train schedule adjustments; determining an anomaly text string based on the current train anomaly; performing vector transformation on each phrase in the anomaly text string to obtain at least one first phrase vector; performing vector transformation on each phrase of each entry in the train schedule adjustment knowledge base to obtain at least one second phrase vector; calculating the similarity between each first phrase vector and each second phrase vector; filtering entries in the train schedule adjustment knowledge base based on the similarity between each first phrase vector and each second phrase vector to obtain the optimal schedule adjustment entry; and processing the current train anomaly based on the optimal schedule adjustment entry. This invention can improve the efficiency of rail transit train operation schedule adjustments.

Description

Technical Field

[0001] This invention relates to the field of rail transit technology, and in particular to a method, apparatus, equipment and medium for handling rail transit anomalies and training a rail transit anomaly handling model. Background Technology

[0002] With the rapid development of rail transit, rail trains need to follow precise operating plans to ensure efficient transportation during normal operation. When rail trains encounter emergencies or malfunctions, adaptive adjustments to the operating plans are necessary.

[0003] Currently, adjustments to rail train operation plans can be made through manual experience-based adjustments or preset rule adjustments.

[0004] However, the above method has the drawback of low efficiency. Summary of the Invention

[0005] This invention provides a method, apparatus, equipment, and medium for handling rail transit anomalies and training a rail transit anomaly handling model. The embodiments of this invention can improve the efficiency of adjusting rail train operation plans.

[0006] In a first aspect, embodiments of the present invention provide a method for handling anomalies in rail transit, the method comprising:

[0007] Access the current train anomaly information and train schedule adjustment knowledge base;

[0008] Feature extraction is performed on the current abnormal train situation to obtain at least one abnormal feature vector, which includes word vectors and / or sentence vectors;

[0009] Obtain at least one knowledge feature vector for each knowledge item in the train schedule adjustment knowledge base;

[0010] The knowledge feature vector is enhanced to obtain the enhanced knowledge feature vector;

[0011] Calculate the similarity between the anomaly feature vector and each augmented knowledge feature vector;

[0012] For each piece of knowledge in the train schedule adjustment knowledge base, the knowledge is filtered based on the similarity between the abnormal feature vector and the feature vector of each enhanced knowledge to obtain the optimal schedule adjustment knowledge.

[0013] Based on the knowledge of the optimal plan, handle the current abnormal situation of the train.

[0014] Secondly, embodiments of the present invention provide a training method for a rail transit anomaly handling model, the method comprising:

[0015] The acquired basic knowledge of rail transit is processed to obtain training samples;

[0016] Based on the training samples, the first baseline model is pre-trained to obtain the second track knowledge model;

[0017] Access the train schedule adjustment knowledge base;

[0018] The knowledge base is adjusted according to the train schedule, and the second track knowledge model is fine-tuned to obtain the rail transit anomaly handling model.

[0019] Thirdly, embodiments of the present invention also provide a rail transit anomaly handling device, the device comprising:

[0020] The knowledge base acquisition module is used to acquire the current train anomaly status and train schedule adjustment knowledge base;

[0021] The abnormal feature extraction module is used to extract features from the current abnormal situation of the train and obtain at least one abnormal feature vector, which includes word vectors and / or sentence vectors;

[0022] The knowledge feature extraction module is used to obtain at least one knowledge feature vector for each piece of knowledge in the train plan adjustment knowledge base;

[0023] The knowledge feature enhancement module is used to enhance the knowledge feature vector to obtain an enhanced knowledge feature vector.

[0024] The similarity calculation module is used to calculate the similarity between the abnormal feature vector and each augmented knowledge feature vector;

[0025] The knowledge filtering module is used to filter the knowledge in the train plan adjustment knowledge base based on the similarity between each abnormal feature vector and the enhanced knowledge feature vector to obtain the optimal plan adjustment knowledge.

[0026] The abnormal situation handling module is used to handle abnormal situations of the current train based on the knowledge of the optimal plan adjustment.

[0027] Fourthly, embodiments of the present invention also provide a training device for a rail transit anomaly handling model, the device comprising:

[0028] The sample acquisition module is used to process the acquired basic knowledge of rail transit to obtain training samples;

[0029] The pre-training module is used to pre-train the first benchmark model based on the training samples to obtain the second track knowledge model.

[0030] The data acquisition module is used to acquire the train schedule adjustment knowledge base;

[0031] The model fine-tuning module is used to adjust the knowledge base according to the train schedule, fine-tuning the second track knowledge model to obtain the rail transit anomaly handling model.

[0032] Fifthly, embodiments of the present invention also provide a computer device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method.

[0033] In a sixth aspect, embodiments of the present invention also provide a computer-readable storage medium storing computer instructions that cause a processor to execute the steps in the above-described method.

[0034] The technical solution of this invention, by acquiring the current train anomaly status and train schedule adjustment knowledge base, ensures that the model can access all necessary data related to train operation in real time, providing basic data support for subsequent train schedule adjustments. By extracting features from the current train anomaly status, complex train anomaly data can be converted into a numerical form that can be calculated and compared. Obtaining anomaly feature vectors facilitates subsequent matching and comparison with information in the knowledge base, improving processing efficiency. Using word vectors and / or sentence vectors enhances the semantic understanding capability of feature extraction, enabling the processing of more complex text data. By acquiring the feature vectors of each piece of knowledge in the train schedule adjustment knowledge base, it ensures that the adjustment scheme has a clear representation in the vector space. This facilitates subsequent calculations and matching; by vectorizing each entry in the knowledge base, different adjustment schemes can be uniformly represented in the computer system, thereby improving the efficiency of retrieval and filtering; by enhancing the knowledge feature vectors, the model can adapt to more diverse train anomaly situations, thus providing more accurate adjustment schemes; by calculating the similarity between anomaly feature vectors and enhanced knowledge feature vectors, semantic matching between train anomalies and adjustment schemes can be achieved; by filtering adjustment schemes with high similarity, it can be ensured that train anomalies receive the most suitable handling scheme; by adjusting the knowledge based on the optimal plan to handle the current train anomaly, it can be ensured that the implementation of the anomaly handling scheme is most suitable for the current situation, thereby improving the efficiency of railway train operation plan adjustment.

[0035] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 A flowchart of a rail transit anomaly handling method provided in an embodiment of the present invention;

[0038] Figure 2 A flowchart of a rail transit anomaly handling method provided in an embodiment of the present invention;

[0039] Figure 3 A flowchart illustrating a training method for a rail transit anomaly handling model provided in an embodiment of the present invention;

[0040] Figure 4 This is a schematic diagram of the structure of a rail transit anomaly handling device provided in an embodiment of the present invention;

[0041] Figure 5 A schematic diagram of the structure of a training device for a rail transit anomaly handling model provided in an embodiment of the present invention;

[0042] Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0043] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0044] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0045] The acquisition, storage, and application of current train anomaly information, train schedule adjustment knowledge base, and basic rail transit knowledge involved in the technical solutions of this invention comply with relevant laws and regulations and do not violate public order and good morals.

[0046] Figure 1 This is a flowchart illustrating a method for handling rail transit anomalies according to an embodiment of the present invention. This embodiment is applicable to rail transit anomaly handling and can be executed by a rail transit anomaly handling device, which can be implemented in hardware and / or software.

[0047] See Figure 1 The rail transit anomaly handling method shown includes:

[0048] S101. Obtain the current train abnormality status and train schedule adjustment knowledge base.

[0049] The current train anomaly can refer to any deviation from normal train operation during the journey. Anomalies typically include equipment malfunctions, weather conditions, traffic accidents, track problems, and / or signaling issues. Current train anomalies may prevent the train from operating as scheduled. They form the basis for triggering train schedule adjustments; only by identifying and obtaining information about the current train anomaly can a reasonable emergency response plan be developed. For example, a current train anomaly could mean the train cannot continue operating due to a power system failure; it could mean the train is encountering extreme weather and reducing its speed; or it could mean the signaling system malfunctions, causing a delay at the station.

[0050] The train schedule adjustment knowledge base refers to a dataset containing multiple train anomaly information, adjustment plans, priorities, and related information. It includes various possible train anomalies, corresponding train adjustment plans, train operation priorities, and associated equipment. Train adjustment plans can refer to how to adjust train operation schedules, timetables, and routes based on different anomalies. The knowledge base provides emergency response solutions for various train anomalies, avoiding human error and improving efficiency and effectiveness in responding to anomalies by using standardized solutions. For example, when a train stops due to equipment failure, the knowledge base records the handling steps, how to dispatch other trains, and how to provide alternative services to passengers. When a train encounters extreme weather, the knowledge base records the train's speed and prioritizes train operation, such as long-distance trains passing through extreme weather sections first, while short-distance trains detour via alternative routes.

[0051] S102. Extract features from the current abnormal train situation to obtain at least one abnormal feature vector, which includes word vectors and / or sentence vectors.

[0052] Here, the anomaly feature vector refers to a vector representation describing key information in the current train anomaly situation. The anomaly feature vector is used to transform the current train anomaly situation into a representation that can be understood by the deep learning model, and it can serve as the basis vector for matching with the train schedule adjustment knowledge base.

[0053] Here, word vectors can refer to the feature vectors converted from each word group in the current train anomaly situation. Word group vectors are used to represent the feature vectors of a single word group in the train anomaly situation.

[0054] Here, the statement vector can refer to the feature vector transformed from the entire statement in the current train anomaly situation. The statement vector is used to represent the contextual information of the entire statement in the train anomaly situation.

[0055] S103. Obtain at least one knowledge feature vector of each knowledge in the train plan adjustment knowledge base.

[0056] Here, "knowledge" can refer to a specific piece of information or record in the train schedule adjustment knowledge base. Each piece of knowledge typically corresponds to a specific train adjustment plan or processing step. Each piece of knowledge is a basic building block of the train schedule adjustment knowledge base, and each piece of knowledge describes how to adjust the train operation plan when a specific train encounters an anomaly. Adjusting the train operation plan includes the required operational steps, priorities, and relevant rules.

[0057] In this context, the knowledge feature vector refers to the vector representation of each phrase in every piece of knowledge within the train schedule adjustment knowledge base. The knowledge feature vector provides a mathematical representation for the matching process, allowing deep learning models to identify the most relevant adjustment scheme between the train schedule adjustment knowledge base and the current train anomaly by calculating the similarity between the anomaly feature vector and the knowledge feature vector.

[0058] S104. Enhance the knowledge feature vector to obtain the enhanced knowledge feature vector.

[0059] Enhanced knowledge feature vectors refer to the process of augmenting the knowledge feature vectors in the train schedule adjustment knowledge base to obtain vector representations with richer information and stronger expressive power. Through this augmentation, these knowledge feature vectors can better adapt to current train anomalies, helping the model make more accurate decisions when handling train scheduling and anomaly adjustments. Enhanced knowledge feature vectors enrich the original knowledge vectors, incorporating more contextual information and external factors. This enables the model to understand more complex train anomalies and provide more accurate adjustment solutions.

[0060] S105. Calculate the similarity between the abnormal feature vector and each enhanced knowledge feature vector.

[0061] Similarity, in this context, refers to the degree of similarity between different feature vectors in a vector space, used to determine the correlation between them. Cosine similarity can be used to calculate the similarity between anomalous feature vectors and various augmented knowledge feature vectors. Cosine similarity measures the angle between two feature vectors; the closer the cosine similarity value is to 1, the more similar the two feature vectors are, meaning the semantic content they represent is closer. Calculating similarity is used to select the most suitable solution for the current train anomaly from multiple possible train adjustment plan knowledge bases. A higher similarity value indicates a higher degree of matching between the relevant knowledge in the train plan adjustment knowledge base and the current train anomaly.

[0062] Specifically, calculating the similarity between the abnormal feature vector and each enhanced knowledge feature vector is equivalent to calculating the similarity between the abnormal feature vector and the enhanced knowledge feature vector corresponding to the phrases in each knowledge in each train plan adjustment knowledge base.

[0063] In a specific example, suppose there are three pieces of knowledge in the train schedule adjustment knowledge base: knowledge A, knowledge B, and knowledge C. Knowledge A includes phrases a, b, and c; knowledge B includes phrases d, e, and f; and knowledge C includes phrases g, h, and i. The enhanced knowledge feature vectors corresponding to knowledge A are aa, bb, and cc, respectively; the enhanced knowledge feature vectors corresponding to knowledge B are dd, ee, and ff, respectively; and the enhanced knowledge feature vectors corresponding to knowledge C are gg, hh, and ii, respectively. The abnormal feature vectors corresponding to each phrase determined based on the current train anomaly are X, Y, and Z, respectively. The similarity between each abnormal feature vector and each enhanced knowledge feature vector is calculated by calculating the similarity between abnormal feature vector X and the enhanced knowledge feature vectors corresponding to knowledge A, the similarity between abnormal feature vector Y and the enhanced knowledge feature vectors corresponding to knowledge B, and the similarity between abnormal feature vector Z and the enhanced knowledge feature vectors corresponding to knowledge C.

[0064] S106. For each piece of knowledge in the train schedule adjustment knowledge base, the knowledge in the train schedule adjustment knowledge base is filtered according to the similarity between the abnormal feature vector and the feature vector of each enhanced knowledge to obtain the optimal schedule adjustment knowledge.

[0065] Filtering refers to the process of selecting train schedule adjustment knowledge that is most relevant to the current train anomaly based on calculated similarity. Filtering ensures that only the train schedule adjustment knowledge that best fits the current anomaly is retained, avoiding irrelevant or inappropriate adjustment schemes from interfering with the decision-making process.

[0066] In this context, optimal plan adjustment knowledge refers to the train schedule adjustment scheme that is most suitable for the current train anomaly situation, obtained through screening. Optimal plan adjustment knowledge represents the best response to the current train anomaly.

[0067] S107. Adjust knowledge according to the optimal plan and handle the current abnormal situation of the train.

[0068] In this context, "processing" refers to taking appropriate operational steps based on the optimal plan and adjusted knowledge to resolve the current abnormal situation of the train. The processing includes executing specific operations in the adjustment plan, such as train detours, turnarounds, temporary suspensions, and timetable adjustments.

[0069] As can be seen, in this embodiment, by acquiring the current train anomaly status and train schedule adjustment knowledge base, the model can ensure real-time access to all necessary data related to train operation, providing basic data support for subsequent train schedule adjustments. By extracting features from the current train anomaly status, complex train anomaly data can be converted into numerical forms that can be calculated and compared. Obtaining anomaly feature vectors facilitates subsequent matching and comparison with information in the knowledge base, improving processing efficiency. Using word vectors and / or sentence vectors enhances the semantic understanding of feature extraction, enabling the processing of more complex text data. Acquiring the feature vectors of each piece of knowledge in the train schedule adjustment knowledge base ensures that the adjustment scheme has a clear representation in the vector space. This facilitates subsequent calculations and matching; by vectorizing each entry in the knowledge base, different adjustment schemes can be uniformly represented in the computer system, thereby improving the efficiency of retrieval and filtering; by enhancing the knowledge feature vectors, the model can adapt to more diverse train anomaly situations, thus providing more accurate adjustment schemes; by calculating the similarity between anomaly feature vectors and enhanced knowledge feature vectors, semantic matching between train anomalies and adjustment schemes can be achieved; by filtering adjustment schemes with high similarity, it can be ensured that train anomalies receive the most suitable handling scheme; by adjusting the knowledge based on the optimal plan to handle the current train anomaly, it can be ensured that the implementation of the anomaly handling scheme is most suitable for the current situation, thereby improving the efficiency of railway train operation plan adjustment.

[0070] In an optional embodiment, the process of "selecting the knowledge in the train schedule adjustment knowledge base based on the similarity between the abnormal feature vector and each enhanced knowledge feature vector to obtain the optimal schedule adjustment knowledge" is refined to "extracting enhanced knowledge feature vectors whose similarity to each enhanced knowledge feature vector is greater than the similarity threshold based on the similarity between the abnormal feature vector and each enhanced knowledge feature vector to obtain at least one related word vector; obtaining the time information of each related word vector; calculating the semantic matching degree between each related word vector and each abnormal feature vector; determining the optimal word vector among the related word vectors based on the time information of each related word vector and the semantic matching degree between each related word vector and the abnormal feature vector; and determining the optimal schedule adjustment knowledge based on the optimal word vector," thereby improving the operation of rail transit anomaly handling.

[0071] It should be noted that for parts not described in detail in the embodiments of the present invention, please refer to the descriptions in other embodiments. Figure 2 This is a flowchart of a rail transit anomaly handling method provided in an embodiment of the present invention.

[0072] See Figure 2The rail transit anomaly handling method shown includes:

[0073] S201. Obtain the current train abnormality status and train schedule adjustment knowledge base.

[0074] S202. Extract features from the current abnormal train situation to obtain at least one abnormal feature vector, which includes word vectors and / or sentence vectors.

[0075] S203. Obtain at least one knowledge feature vector of each knowledge in the train plan adjustment knowledge base.

[0076] S204. Enhance the knowledge feature vector to obtain the enhanced knowledge feature vector.

[0077] S205. Calculate the similarity between the abnormal feature vector and each enhanced knowledge feature vector.

[0078] S206. Based on the similarity and similarity threshold between the abnormal feature vector and each enhanced knowledge feature vector, extract each enhanced knowledge feature vector whose similarity to the abnormal feature vector and each enhanced knowledge feature vector is greater than the similarity threshold to obtain at least one related word group vector.

[0079] The similarity threshold refers to a threshold used to measure whether the similarity between anomaly feature vectors and various augmented knowledge feature vectors reaches a certain level of similarity. By using the similarity threshold, knowledge from the train schedule adjustment knowledge base that is irrelevant to the current train anomaly or has too low a similarity can be filtered out. For example, when using cosine similarity to calculate similarity, a similarity threshold of 0.8 can be set. Only when the similarity between feature vectors is greater than this threshold will the augmented knowledge feature vector be used for further processing.

[0080] Among them, relevant phrase vectors can refer to enhanced knowledge feature vectors whose similarity to the abnormal feature vector exceeds a similarity threshold. For example, enhanced knowledge feature vectors with a similarity threshold can be used as relevant phrase vectors. During the screening phase, feature vectors corresponding to the knowledge in the train schedule adjustment knowledge base most relevant to the current abnormal situation are extracted based on the similarity threshold. These relevant phrase vectors can then be used to further evaluate their matching degree with the current train abnormal situation.

[0081] S207. Obtain the time information of each relevant word group vector.

[0082] The time information refers to the generation time or timestamp information of the knowledge in the train schedule adjustment knowledge base corresponding to the relevant term vector. Time information is used to determine whether the current knowledge belongs to the latest relevant solution, ensuring that the newest and most effective solution is selected. Time information allows for the priority selection of timely updated train adjustment solutions, identifying outdated solutions and avoiding the use of obsolete solutions. For example, assuming that an adjustment solution in the train schedule adjustment knowledge base corresponding to the relevant term vector was generated in XXXX year, and this entry has not been updated or adjusted, the time information can be used to determine whether the solution is outdated, especially after the introduction of new technologies or processes, when the solution may no longer be applicable to solving current train anomalies.

[0083] S208. Calculate the semantic matching degree between each relevant word group vector and the abnormal feature vector.

[0084] Semantic matching degree refers to an indicator that measures the semantic consistency between the abnormal feature vector corresponding to the current abnormal situation and the related word phrase vector. Semantic matching degree reflects the similarity between two feature vectors in a practical sense, rather than just the literal similarity of the text. It is used to determine which adjustment schemes corresponding to the related word phrase vectors best match the processing requirements of the current train abnormal situation corresponding to the abnormal vector, thus selecting the most suitable adjustment scheme.

[0085] S209. Based on the time information of each relevant word group vector and the semantic matching degree between each relevant word group vector and the abnormal feature vector, determine the optimal word group vector among each relevant word group vector.

[0086] The optimal phrase vector refers to the phrase vector most suitable for the current train anomaly situation, selected from relevant phrase vectors based on time information and semantic matching degree. By calculating the semantic matching degree and time information of each relevant phrase vector, the optimal phrase vector can be selected from multiple relevant phrase vectors, and the optimal phrase vector can be used as the basis for the final train schedule adjustment scheme.

[0087] S210. Based on the optimal word vector, determine the optimal plan adjustment knowledge.

[0088] Specifically, based on the optimal phrase vector, the optimal plan adjustment knowledge is determined in the vector space of the train plan adjustment knowledge base.

[0089] S211. Adjust knowledge according to the optimal plan to handle the current abnormal situation of the train.

[0090] As can be seen, in this embodiment, by setting a similarity threshold and selecting only train schedule adjustment knowledge highly related to the anomaly, knowledge irrelevant to the current train anomaly or with low matching degree can be filtered out, improving the accuracy of matching; extracting enhanced knowledge feature vectors with high similarity ensures that subsequent processing focuses primarily on items closely related to the current train anomaly, thereby improving decision-making efficiency; obtaining the time information of relevant phrase vectors ensures the timeliness of the selected adjustment scheme; calculating the semantic matching degree measures the degree of matching between relevant phrases and the current anomaly, ensuring the selection of the most suitable train schedule adjustment knowledge; accurately calculating semantic matching avoids selecting semantically irrelevant or partially matching solutions, improving the effectiveness and accuracy of anomaly handling; combining time information and semantic matching degree to determine the optimal phrase vector ensures that the final selected adjustment scheme is not only semantically matching but also time-sensitive; determining the optimal schedule adjustment knowledge through the optimal phrase vector directly outputs the solution that best suits the current train anomaly.

[0091] In some embodiments, the degree of semantic matching includes the degree of semantic similarity and the degree of contextual alignment;

[0092] Calculate the semantic matching degree between each relevant word phrase vector and the abnormal feature vector, including:

[0093] Based on the similarity between each relevant word group vector and the abnormal feature vector, the semantic similarity between each relevant word group vector and the abnormal feature vector is determined.

[0094] Based on the sentence vector of the abnormal feature vector, calculate the contextual alignment degree between each related word group vector and the abnormal feature vector;

[0095] Based on semantic similarity and contextual alignment, the semantic matching degree between each relevant word group vector and the abnormal feature vector is determined.

[0096] Semantic similarity refers to the degree of semantic similarity between the phrases corresponding to two phrase vectors. Semantic similarity is typically used to indicate the semantic consistency between two texts or phrases. For example, suppose the phrase corresponding to the relevant phrase vector is "overhead contact network failure," and the phrase corresponding to the abnormal feature vector is "power equipment damage." If the semantic similarity between the two phrase vectors is 0.85, it means that the phrases corresponding to the two phrase vectors are highly matched semantically, and their semantic meanings are quite similar.

[0097] In a specific example, the semantic matching score between the relevant word vector and the abnormal feature vector can be calculated in the following ways: by calculating the dot product of the two word vectors; by calculating the Euclidean distance between the two word vectors; or by calculating the Manhattan distance between the two word vectors.

[0098] Contextual alignment refers to the degree of alignment between the corresponding word groups of two word vectors at the contextual level. That is, whether the two word groups are semantically complementary or completely identical, especially in the context of specific problem-solving, whether they are solving the same problem. Semantic alignment emphasizes whether the roles and goals of the word groups in the solution are consistent.

[0099] As can be seen, in this embodiment, by calculating semantic similarity, a quantitative assessment of the semantic consistency of word groups can be achieved. Semantic similarity provides a standardized numerical value that can quantify the semantic similarity between different word groups. Semantic similarity can specifically quantify the degree of association between the word group corresponding to the abnormal feature vector and the word group corresponding to the related word group vector. By calculating the degree of contextual alignment, it is helpful to consider not only semantic similarity but also the coordination of semantics in the actual execution context when selecting train plan adjustment schemes. By combining semantic similarity and semantic alignment, a more comprehensive semantic matching assessment can be achieved.

[0100] In some embodiments, the current abnormal train situation is handled based on the knowledge of the optimal plan adjustment, including:

[0101] Obtain the adjustment plan, priority, and affected equipment from the optimal plan adjustment knowledge;

[0102] Obtain the priority of each affected device;

[0103] The priorities in the optimal plan adjustment knowledge are compared with the priorities of each affected device to obtain the priority comparison results;

[0104] Based on the priority comparison results and the adjustment plan in the optimal plan adjustment knowledge, the current abnormal train situation is handled.

[0105] The adjustment plan can refer to a specific scheme for adjusting train operations based on abnormal train conditions. Priority refers to the degree to which certain adjustment schemes are more important or take precedence over others among multiple train schedule adjustment schemes; priority is usually determined based on factors such as the importance of the train, service demand, passenger numbers, and / or train type. Affected equipment can refer to auxiliary data related to abnormal train conditions, such as data relating a specific piece of knowledge in the train schedule adjustment knowledge base to other knowledge.

[0106] In a specific example, the train plan knowledge base can be stored in a semi-structured format. For instance, the semi-structured format of a certain piece of knowledge in the train plan knowledge base can be adjusted as follows:

[0107] "Knowledge Item 1":{

[0108] "Train Abnormality Information": "A signal system malfunction has caused some signal lights to go out in certain sections."

[0109] "Adjusted Plan": "Immediately switch trains on the affected section to manual driving mode, operate at limited speeds, and simultaneously arrange maintenance personnel to urgently repair the signaling system. Trains following behind should appropriately increase their headway to avoid rear-end collisions."

[0110] Priority: "High",

[0111] "Affected Equipment 1": ["Signal System"],

[0112] "Image Equipment 2": "Trains within and adjacent sections of the faulty section"

[0113] The priority comparison result refers to an assessment of how to prioritize different tasks or resources by comparing the priorities in the optimal plan adjustment knowledge with the priorities of the affected equipment. Priority comparisons are typically conducted based on factors such as business rules, scheduling requirements, and / or emergencies. The priority comparison result is used to determine which tasks or resources should be prioritized, ensuring that train anomalies can be resolved in the shortest possible time.

[0114] As can be seen, in this embodiment, by acquiring the adjustment plan, priority, and affected equipment from the optimal plan adjustment knowledge, a comprehensive understanding of the handling scheme required for the current train anomaly can be obtained; by acquiring relevant information on the adjustment plan, priority, and affected equipment, valuable reference can be provided for the subsequent decision-making process; by acquiring the priority of each affected equipment, it can be identified which equipment is the key equipment that needs to be given priority in fault handling; by processing according to the priority comparison results and the adjustment plan, a rapid response to the train anomaly can be achieved, and train scheduling and resource allocation can be optimized.

[0115] In some embodiments, the current train anomaly is handled based on the priority comparison results and the adjustment plan in the optimal plan adjustment knowledge, including:

[0116] Based on the priority comparison results, determine the adjustment plan and the execution order of each affected device in the optimal plan adjustment knowledge;

[0117] According to the execution sequence, handle the current abnormal situation of the train.

[0118] The execution order can refer to the sequential execution order between the adjustment plan and the influencing devices in the most planned adjustment knowledge.

[0119] As can be seen, in this embodiment, by determining the adjustment plan and the equipment execution sequence, train anomalies can be handled efficiently according to priority order, reducing delays and operational disruptions; by using priority-driven execution order, resources can be reasonably allocated among multiple tasks, minimizing the impact of train anomalies on operations; by handling train anomalies in sequence, overall scheduling capabilities can be improved, reducing passenger congestion and operational losses.

[0120] In some embodiments, the knowledge feature vector is enhanced to obtain an enhanced knowledge feature vector, including:

[0121] Based on the knowledge feature vector, determine the similarity feature vector and context feature vector corresponding to the knowledge feature vector;

[0122] Based on similarity feature vectors and context feature vectors, the enhanced knowledge feature vectors are determined.

[0123] In this context, similarity feature vectors refer to feature vectors that are semantically similar to the word groups corresponding to the knowledge feature vectors. These similarity feature vectors are usually derived from synonyms or near-synonyms of the word group, and new vectors are generated by expanding these similar words, thereby enhancing the expressive power of the original knowledge feature vectors. By expanding similar words, more possible expressions of a word can be obtained, avoiding reliance on knowledge features with only a single expression, thus improving the model's ability to handle complex problems. For example, suppose there is an adjustment scheme in the train scheduling adjustment knowledge base called "alternate train scheduling". By generating similarity feature vectors, near-synonyms related to "alternate train", such as "alternate train" and "emergency train", can be introduced to generate feature vectors for these words. These enhanced vectors can help the model recognize that "alternate train scheduling" is also a suitable adjustment scheme.

[0124] In this context, the feature vector can refer to a feature vector generated based on the contextual information of the sentence in which the phrase is located, as found in the knowledge feature vector. This feature vector considers not only the semantics of the words, but also their position in the sentence, grammatical relationships, and other contextual information.

[0125] As can be seen, in this embodiment, by determining similarity feature vectors and context feature vectors based on knowledge feature vectors, the information in the train plan adjustment knowledge base can be expanded, making it more comprehensive and accurate. By determining enhanced knowledge feature vectors, more refined and multi-dimensional knowledge representations can be obtained, making train anomaly handling more intelligent. By combining similarity feature vectors and context feature vectors, the enhanced feature vectors can provide effective solutions for more diverse train anomaly scenarios.

[0126] Figure 3 This is a flowchart illustrating a training method for a rail transit anomaly handling model provided in an embodiment of the present invention. This embodiment is applicable to the training of rail transit anomaly handling models. The method can be executed by a training device for the rail transit anomaly handling model, which can be implemented in hardware and / or software.

[0127] See Figure 3 The rail transit anomaly handling method shown includes:

[0128] S301. Process the acquired basic knowledge of rail transit to obtain training samples.

[0129] Basic knowledge of rail transit refers to fundamental information and principles related to rail transit systems. This includes knowledge of train operation principles, dispatching and management, equipment failures, traffic regulations, and passenger flow patterns. This basic knowledge provides the foundational knowledge base for training large-scale models, helping them understand the fundamental operations and requirements of rail transit systems.

[0130] Optionally, the basic knowledge of rail transit can be in a semi-structured format. In a specific example, the semi-structured format for basic knowledge of rail transit is as follows:

[0131] {"Knowledge Category":"Orbital System",}

[0132] Subject: "Orbital Composition"

[0133] Key Point: "Rail sleepers are an important component of railway tracks. Their main function is to support the rails and maintain their position and gauge. Common sleeper materials include wood and concrete. Wooden sleepers are highly elastic but have a shorter lifespan; concrete sleepers have high strength and a long lifespan, but relatively poor elasticity."

[0134] Related knowledge: ["rail", "ballast bed", "track elasticity"]

[0135] Importance level: "Medium"

[0136] S302. Based on the training samples, pre-train the obtained first benchmark model to obtain the second track knowledge model.

[0137] Here, the first baseline model can refer to an initial large model that has already been trained and possesses certain functionalities. The first baseline model is typically a relatively general large model, serving as the starting version for handling rail transit anomaly processing tasks. For example, large models such as GPT (Generative Pretrained Transformer) or BERT can be selected as the first baseline model.

[0138] Pre-training refers to the process of initially training a model based on fundamental knowledge of rail transit. This allows the model to learn basic patterns and knowledge of rail transit, which is then fine-tuned for specific tasks. Through pre-training, the initial baseline model learns some general features of rail transit fundamentals on large-scale data, making subsequent training on specific tasks more efficient.

[0139] The second track knowledge model can refer to the model obtained after pre-training the first baseline model. Based on this, the second track knowledge model can already grasp the basic knowledge of rail transit.

[0140] S303, Obtain the train schedule adjustment knowledge base.

[0141] S304. Adjust the knowledge base according to the train schedule, fine-tune the second track knowledge model, and obtain the rail transit anomaly handling model.

[0142] Fine-tuning refers to retraining a pre-trained model using task-specific data to better adapt the second track knowledge model to the needs of that specific task. Fine-tuning further optimizes the model parameters based on the pre-trained model, enabling it to better solve specific problems. Fine-tuning can involve making minor adjustments based on existing basic knowledge of rail transit to adapt to specific rail transit anomaly handling tasks.

[0143] Among them, the rail transit anomaly handling model refers to a model that can provide the optimal plan adjustment items based on the current train anomaly situation. By analyzing and modeling train anomaly data in the rail transit system, the rail transit anomaly handling model can automatically select appropriate handling solutions in the face of emergencies, thereby improving the efficiency and accuracy of train scheduling.

[0144] As can be seen, in this embodiment, by processing basic knowledge of rail transit, the original knowledge can be transformed into a data format suitable for model training, ensuring the standardization and structuring of training samples, thereby improving the model's learning effect; by using training samples to pre-train the first benchmark model, the capabilities of the general model can be extended to the field of rail transit, enabling it to master basic knowledge related to rail transit and enhancing the model's professionalism; by acquiring the train schedule adjustment knowledge base, specific problems and solutions within the domain can be provided for subsequent fine-tuning and enhancement, further enriching the model's knowledge reserves; by enhancing train anomaly information, the expression of anomaly handling solutions can be expanded and refined, improving the relevance and completeness of the train schedule adjustment knowledge base, providing high-quality input for subsequent model fine-tuning; by using the train schedule adjustment knowledge base and enhanced information to fine-tune the model, it can further adapt to the specific needs of rail transit anomaly handling, making the model more efficient and accurate in handling actual rail train schedule adjustment problems.

[0145] In some embodiments, the second track knowledge model includes a bottom network layer and a top network layer; the output of the bottom network layer is the input of the top network layer.

[0146] Based on the train schedule, the knowledge base is adjusted, and the second track knowledge model is fine-tuned to obtain the rail transit anomaly handling model, including:

[0147] Based on the train schedule, the knowledge base and enhanced information are adjusted, and the training parameters of the shallow network layer in the second track knowledge model are adjusted to obtain the rail transit anomaly handling model.

[0148] In this context, the bottom layer refers to the network layer at the bottom of a deep learning model. The bottom layer is typically located near the input layer and has a large number of parameters. It is usually responsible for learning low-level features, and its parameters remain unchanged during fine-tuning.

[0149] In this context, the top-level network layer refers to the top network layer in a deep learning model. The top-level network layer is typically located near the output layer and has fewer parameters than the bottom-level network layers. It is responsible for learning more complex high-level features. During fine-tuning, the parameters of the top-level network layer are updated to adapt the second-track knowledge model to the task of handling rail transit anomalies.

[0150] As can be seen, in this embodiment, by including a bottom network layer and a top network layer, the second track knowledge model can achieve hierarchical feature learning when handling rail transit anomalies; by adjusting only the parameters of the top network layer, the model can adapt to new task requirements, while avoiding excessive changes to the learning process of low-level features. The top network layer has fewer parameters and mainly learns high-level features. Adjusting these parameters during fine-tuning helps to better adapt to the task of handling rail transit anomalies.

[0151] Figure 4 This invention provides a schematic diagram of a rail transit anomaly handling device. This invention is applicable to rail transit anomaly handling situations, and the device can execute rail transit anomaly handling methods. The device can be implemented in hardware and / or software.

[0152] See Figure 4 The rail transit anomaly handling device shown includes: a knowledge base acquisition module 401, an anomaly feature extraction module 402, a knowledge feature extraction module 40, a knowledge feature enhancement module 404, a similarity calculation module 405, a knowledge filtering module 406, and an anomaly handling module 407, wherein...

[0153] The knowledge base acquisition module 401 is used to acquire the current train abnormality status and train plan adjustment knowledge base;

[0154] The abnormal feature extraction module 402 is used to extract features from the current abnormal situation of the train and obtain at least one abnormal feature vector, which includes word vectors and / or sentence vectors.

[0155] The knowledge feature extraction module 403 is used to obtain at least one knowledge feature vector of each knowledge in the train plan adjustment knowledge base;

[0156] The knowledge feature enhancement module 404 is used to enhance the knowledge feature vector to obtain an enhanced knowledge feature vector;

[0157] The similarity calculation module 405 is used to calculate the similarity between the abnormal feature vector and each enhanced knowledge feature vector;

[0158] The knowledge filtering module 406 is used to filter the knowledge in the train plan adjustment knowledge base based on the similarity between each abnormal feature vector and the enhanced knowledge feature vector to obtain the optimal plan adjustment knowledge.

[0159] The abnormal situation handling module 407 is used to handle the current abnormal situation of the train based on the knowledge of the optimal plan adjustment.

[0160] The technical solution of this invention, by acquiring the current train anomaly status and train schedule adjustment knowledge base, ensures that the model can access all necessary data related to train operation in real time, providing basic data support for subsequent train schedule adjustments. By extracting features from the current train anomaly status, complex train anomaly data can be converted into a numerical form that can be calculated and compared. Obtaining anomaly feature vectors facilitates subsequent matching and comparison with information in the knowledge base, improving processing efficiency. Using word vectors and / or sentence vectors enhances the semantic understanding capability of feature extraction, enabling the processing of more complex text data. By acquiring the feature vectors of each piece of knowledge in the train schedule adjustment knowledge base, it ensures that the adjustment scheme has a clear representation in the vector space. This facilitates subsequent calculations and matching; by vectorizing each entry in the knowledge base, different adjustment schemes can be uniformly represented in the computer system, thereby improving the efficiency of retrieval and filtering; by enhancing the knowledge feature vectors, the model can adapt to more diverse train anomaly situations, thus providing more accurate adjustment schemes; by calculating the similarity between anomaly feature vectors and enhanced knowledge feature vectors, semantic matching between train anomalies and adjustment schemes can be achieved; by filtering adjustment schemes with high similarity, it can be ensured that train anomalies receive the most suitable handling scheme; by adjusting the knowledge based on the optimal plan to handle the current train anomaly, it can be ensured that the implementation of the anomaly handling scheme is most suitable for the current situation, thereby improving the efficiency of railway train operation plan adjustment.

[0161] In some embodiments, regarding the selection of optimal plan adjustment knowledge from the knowledge base for train plan adjustment based on the similarity between the abnormal feature vector and the feature vector of each enhanced knowledge, the knowledge selection module 406 is specifically used for:

[0162] Based on the similarity and similarity threshold between the abnormal feature vector and each augmented knowledge feature vector, each augmented knowledge feature vector whose similarity to the abnormal feature vector and each augmented knowledge feature vector is greater than the similarity threshold is extracted to obtain at least one related word phrase vector;

[0163] Obtain the time information of each relevant word phrase vector;

[0164] Calculate the semantic matching degree between each relevant word group vector and the abnormal feature vector;

[0165] Based on the time information of each relevant word group vector and the semantic matching degree between each relevant word group vector and the abnormal feature vector, the optimal word group vector is determined among each relevant word group vector;

[0166] Based on the optimal word vector, determine the optimal plan to adjust knowledge.

[0167] In some embodiments, the semantic matching degree includes semantic similarity and contextual alignment; in calculating the semantic matching degree between each relevant word group vector and the abnormal feature vector, the knowledge filtering module 406 is specifically used for:

[0168] Based on the similarity between each relevant word group vector and the abnormal feature vector, the semantic similarity between each relevant word group vector and the abnormal feature vector is determined.

[0169] Based on the sentence vector of the abnormal feature vector, calculate the contextual alignment degree between each related word group vector and the abnormal feature vector;

[0170] Based on semantic similarity and contextual alignment, the semantic matching degree between each relevant word group vector and the abnormal feature vector is determined.

[0171] In some embodiments, in handling current train anomalies based on knowledge of the optimal plan, the anomaly handling module 407 is specifically used for:

[0172] Obtain the adjustment plan, priority, and affected equipment from the optimal plan adjustment knowledge;

[0173] Obtain the priority of each affected device;

[0174] The priorities in the optimal plan adjustment knowledge are compared with the priorities of each affected device to obtain the priority comparison results;

[0175] Based on the priority comparison results and the adjustment plan in the optimal plan adjustment knowledge, the current abnormal train situation is handled.

[0176] In some embodiments, in handling current train anomalies based on priority comparison results and the adjustment plan in the optimal plan adjustment knowledge, the anomaly handling module 407 is specifically used for:

[0177] Based on the priority comparison results, determine the adjustment plan and the execution order of each affected device in the optimal plan adjustment knowledge;

[0178] According to the execution sequence, handle the current abnormal situation of the train.

[0179] In some embodiments, in enhancing the knowledge feature vector to obtain an enhanced knowledge feature vector, the knowledge feature enhancement module 404 is specifically used for:

[0180] Based on the knowledge feature vector, determine the similarity feature vector and context feature vector corresponding to the knowledge feature vector;

[0181] Based on similarity feature vectors and context feature vectors, the enhanced knowledge feature vectors are determined.

[0182] The rail transit anomaly handling device provided in the embodiments of the present invention can execute the rail transit anomaly handling method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the rail transit anomaly handling method.

[0183] Figure 5 This invention provides a schematic diagram of a training device for a rail transit anomaly handling model. This invention is applicable to the training of rail transit anomaly handling models. The device can execute training methods for rail transit anomaly handling models and can be implemented in hardware and / or software.

[0184] See Figure 5 The training device for the rail transit anomaly handling model shown includes: a sample acquisition module 501, a pre-training module 502, a data acquisition module 503, an information enhancement module 504, and a model fine-tuning module 505, wherein...

[0185] The sample acquisition module 501 is used to process the acquired basic knowledge of rail transit to obtain training samples;

[0186] The pre-training module 502 is used to pre-train the acquired first benchmark model based on training samples to obtain the second track knowledge model;

[0187] Data acquisition module 503 is used to acquire the train schedule adjustment knowledge base;

[0188] The model fine-tuning module 504 is used to adjust the knowledge base according to the train plan, fine-tune the second track knowledge model, and obtain the rail transit anomaly handling model.

[0189] The technical solution of this invention, by processing basic knowledge of rail transit, can transform the original knowledge into a data format suitable for model training, ensuring the standardization and structuring of training samples, thereby improving the model's learning effect; by using training samples to pre-train the first benchmark model, the capabilities of the general model can be extended to the field of rail transit, enabling it to master basic knowledge related to rail transit and enhancing the model's professionalism; by acquiring a train schedule adjustment knowledge base, specific problems and solutions within the domain can be provided for subsequent fine-tuning and enhancement, further enriching the model's knowledge reserves; by enhancing train anomaly information, the expression of anomaly handling solutions can be expanded and refined, improving the relevance and completeness of the train schedule adjustment knowledge base, providing high-quality input for subsequent model fine-tuning; by using the train schedule adjustment knowledge base and enhanced information to fine-tune the model, it can further adapt to the specific needs of rail transit anomaly handling, making the model more efficient and accurate in handling actual rail train schedule adjustment problems.

[0190] In some embodiments, the second track knowledge model includes a bottom network layer and a top network layer; the output of the bottom network layer is the input of the top network layer.

[0191] In adjusting the knowledge base according to the train schedule and fine-tuning the second track knowledge model to obtain the rail transit anomaly handling model, the model fine-tuning module 504 is specifically used for:

[0192] Based on the train schedule, the knowledge base and enhanced information are adjusted, and the training parameters of the shallow network layer in the second track knowledge model are adjusted to obtain the rail transit anomaly handling model.

[0193] The training device for the rail transit anomaly handling model provided in this embodiment of the invention can execute the training method for the rail transit anomaly handling model provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing the training method for the rail transit anomaly handling model.

[0194] Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention.

[0195] like Figure 6 As shown, computer device 600 includes at least one processor 601 and memory, such as read-only memory (ROM) 602 and random access memory (RAM) 603, communicatively connected to at least one processor 601. The memory stores computer programs executable by at least one processor. Processor 601 can perform various appropriate actions and processes based on the computer program stored in ROM 602 or loaded into RAM 603 from storage unit 608. RAM 603 may also store various programs and data required for the operation of computer device 600. Processor 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0196] Multiple components in computer device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of monitors, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows computer device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0197] Processor 601 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 601 performs the various methods and processes described above, such as methods for handling rail traffic anomalies and training rail traffic anomaly handling models.

[0198] In some embodiments, the methods for handling rail transit anomalies and training rail transit anomaly models can be implemented as computer programs tangibly contained in a computer-readable storage medium, such as storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed on computer device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by processor 601, one or more steps of the methods for handling rail transit anomalies and training rail transit anomaly models described above can be performed. Alternatively, in other embodiments, processor 601 can be configured to execute the methods for handling rail transit anomalies and training rail transit anomaly models by any other suitable means (e.g., by means of firmware).

[0199] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0200] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0201] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0202] To provide interaction with the user, the systems and techniques described herein can be implemented on an operational detection device, which includes: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the XXXX device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0203] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0204] A computing system can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system. It addresses the shortcomings of traditional physical hosts and VPS (Virtual Private Server) services, such as high management difficulty and weak business scalability.

[0205] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0206] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for handling anomalies in rail transit, characterized in that, The method includes: Access the current train anomaly information and train schedule adjustment knowledge base; Feature extraction is performed on the current train anomaly to obtain at least one anomaly feature vector, the anomaly feature vector including word vectors and / or sentence vectors; Obtain at least one knowledge feature vector for each piece of knowledge in the train plan adjustment knowledge base; The knowledge feature vector is enhanced to obtain an enhanced knowledge feature vector; Calculate the similarity between the abnormal feature vector and each of the enhanced knowledge feature vectors; For each piece of knowledge in the train schedule adjustment knowledge base, the knowledge in the train schedule adjustment knowledge base is filtered according to the similarity between the abnormal feature vector and each of the enhanced knowledge feature vectors to obtain the optimal schedule adjustment knowledge; Based on the knowledge of the optimal plan adjustment, the current abnormal train situation is handled.

2. The method according to claim 1, characterized in that, The step involves filtering the knowledge in the train schedule adjustment knowledge base based on the similarity between the abnormal feature vector and each of the enhanced knowledge feature vectors to obtain the optimal schedule adjustment knowledge, including: Based on the similarity and similarity threshold between the abnormal feature vector and each of the enhanced knowledge feature vectors, each of the enhanced knowledge feature vectors whose similarity to the abnormal feature vector and each of the enhanced knowledge feature vectors is greater than the similarity threshold is extracted to obtain at least one related word group vector; Obtain the time information of each of the aforementioned related word group vectors; Calculate the semantic matching degree between each of the related word group vectors and the abnormal feature vectors; Based on the time information of each related word vector and the semantic matching degree between each related word vector and the abnormal feature vector, the optimal word vector is determined among each related word vector; Based on the optimal word vector, determine the optimal plan adjustment knowledge.

3. The method according to claim 2, characterized in that, The degree of semantic matching includes the degree of semantic similarity and the degree of contextual alignment; The calculation of the semantic matching degree between each of the relevant word group vectors and the abnormal feature vectors includes: Based on the similarity between each of the related word vectors and the abnormal feature vectors, the semantic similarity between each of the related word vectors and the abnormal feature vectors is determined; Based on the sentence vector of the abnormal feature vector, calculate the context alignment degree between each of the related word group vectors and the abnormal feature vector; Based on the semantic similarity and contextual alignment, the semantic matching degree between each related word phrase vector and the abnormal feature vector is determined.

4. The method according to claim 1, characterized in that, The step of adjusting the knowledge based on the optimal plan to handle the current train anomaly includes: Obtain the adjustment plan, priority, and affected equipment from the optimal plan adjustment knowledge; Obtain the priority of each of the affected devices; The priorities in the optimal plan adjustment knowledge are compared with the priorities of each of the affected devices to obtain the priority comparison results; Based on the priority comparison results and the adjustment plan in the optimal plan adjustment knowledge, the current train abnormality is handled.

5. The method according to claim 4, characterized in that, The step of handling the current train anomaly based on the priority comparison result and the adjustment plan in the optimal plan adjustment knowledge includes: Based on the priority comparison results, the adjustment plan in the optimal plan adjustment knowledge and the execution order of each affected device are determined; The current train abnormality is handled according to the execution order.

6. The method according to claim 1, characterized in that, The enhancement of the knowledge feature vector to obtain the enhanced knowledge feature vector includes: Based on the knowledge feature vector, determine the similarity feature vector and the context feature vector corresponding to the knowledge feature vector; The enhanced knowledge feature vector is determined based on the similarity feature vector and the context feature vector.

7. A training method for a rail transit anomaly handling model, characterized in that, include: The acquired basic knowledge of rail transit is processed to obtain training samples; Based on the training samples, the first benchmark model is pre-trained to obtain the second track knowledge model; Access the train schedule adjustment knowledge base; Based on the train plan adjustment knowledge base, the second track knowledge model is fine-tuned to obtain the rail transit anomaly handling model.

8. The method according to claim 7, characterized in that, The second orbital knowledge model includes a bottom network layer and a top network layer; the output of the bottom network layer is the input of the top network layer. The step of adjusting the knowledge base according to the train plan and fine-tuning the second track knowledge model to obtain the rail transit anomaly handling model includes: Based on the train plan adjustment knowledge base and the enhanced information, the training parameters of the shallow network layer in the second track knowledge model are adjusted to obtain the rail traffic anomaly handling model.

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the rail transit anomaly handling method according to any one of claims 1 to 6, or the steps of the rail transit anomaly handling model training method according to any one of claims 7 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the steps of the rail transit anomaly handling method according to any one of claims 1 to 6, or the steps of the rail transit anomaly handling model training method according to any one of claims 7 to 8.

Images