A high-speed rail business travel service personnel scheduling method based on data analysis

Abstract

The application discloses a high-speed rail business travel service personnel scheduling method based on data analysis, and relates to the technical field of personnel scheduling management. The method comprises the following steps: data collection and integration: collecting high-speed rail station operation data, passenger information, service personnel information and environment data; performing stream data processing and space-time alignment on the collected data, and constructing a dynamic main data matrix; service demand prediction: extracting passenger multi-dimensional features from the dynamic main data matrix, and generating passenger multi-dimensional feature labels; based on the passenger multi-dimensional feature labels, combining environment data, performing bimodal demand prediction, and obtaining initial passenger service demand. Through the establishment of the bimodal service demand prediction method, the extraction of passenger multi-dimensional features, the combination of environment data and the prediction of passenger demand, the application realizes the transition from passive response to active prediction, accelerates the response speed, reduces the waiting time of passengers and improves the service quality and efficiency.

Description

Technical Field

[0001] This invention relates to the field of personnel scheduling and management technology, specifically a method for scheduling personnel for high-speed rail business travel services based on data analysis. Background Technology

[0002] With the expansion of the high-speed rail network, the proportion of business travel has increased significantly, and traditional extensive scheduling methods can no longer meet the differentiated service needs. Business travelers have higher requirements for service response speed and personalized experiences, necessitating the establishment of a refined scheduling system. Currently, there is a structural imbalance in scheduling, with no distinction in service priority between ordinary and business travelers, leading to a decline in the experience for high-value passengers. It is necessary to achieve precise service and efficient operation through scientific resource allocation.

[0003] For example, patent publication number CN103581299A discloses a service scheduling method, apparatus, and system, including: receiving a service request sent by a terminal; obtaining the terminal's geographical location based on the service request; querying a distributed server matching the terminal's geographical location based on pre-stored geographical locations of various distributed servers; and sending a service response to the terminal, the service response carrying connection information of the distributed server matching the terminal's geographical location. This invention, by scheduling the terminal to a matching distributed server based on its geographical location, solves the problem that manually preset scheduling strategies might configure one or more IP addresses to servers located far away; it improves access speed and accuracy, reduces network transmission costs, and minimizes network latency.

[0004] However, the above and similar technical methods are based on service requests for personnel scheduling, which has a poor proactive service awareness, is prone to response delays, and results in poor service performance after scheduling. Personnel scheduling relies on simple priority rules, which are difficult to adapt to dynamic needs and may lead to resource mismatch. Summary of the Invention

[0005] The purpose of this invention is to provide a method for scheduling high-speed rail business travel service personnel based on data analysis, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for scheduling high-speed rail business travel service personnel based on data analysis, comprising:

[0007] Data collection and integration: Collect high-speed railway station operation data, passenger information, service personnel information, and environmental data; perform streaming data processing and spatiotemporal alignment on the collected data, and construct a dynamic master data matrix;

[0008] Service demand forecasting: Extract multi-dimensional features of passengers from the dynamic master data matrix and generate multi-dimensional feature labels for passengers; based on the multi-dimensional feature labels of passengers and combined with environmental data, perform bimodal demand forecasting to obtain the initial passenger service demand;

[0009] Multidimensional dynamic scheduling: Based on the dynamic master data matrix, service personnel are matched in multiple dimensions according to the initial passenger service needs to generate an initial service plan;

[0010] Service execution and monitoring: Establish a real-time passenger-service mapping space, where service personnel monitor service gaps in real time while executing the initial service plan, evaluate service quality, dynamically adjust the initial service plan, and obtain service execution effect reports;

[0011] Feedback and optimization: Update and optimize the service personnel scheduling process based on the service performance report.

[0012] Furthermore, the method for constructing the dynamic master data matrix includes:

[0013] Data processing: Establish unique linear mileage coordinates and a unified time base, and map all data events to the same spatiotemporal coordinate system; define and unify primary keys for high-speed rail station operation data, passenger data, service personnel information, and environmental data, and establish a primary key mapping table; and perform streaming data processing, aligning data by train number, time window, and spatial window to generate unified aligned data.

[0014] Create an empty matrix template: Select the primary key and predefine an empty matrix based on the primary key;

[0015] Dynamic matrix filling: Map each aligned data entry to a corresponding cell. For multiple records in the same cell, mapping is performed by overwriting and aggregating the latest value to obtain a dynamic master data matrix.

[0016] Verification and Update: Real-time detection of data integrity, marking of anomalies, updating of the dynamic master data matrix, and retention of historical versions.

[0017] Furthermore, the method for obtaining the multidimensional feature labels includes:

[0018] Feature Dimension Definition: Define the core feature dimensions of passengers. The core feature dimensions are divided into static dimensions and dynamic dimensions.

[0019] Feature extraction: Based on the dynamic master data matrix, multi-dimensional features of passengers are extracted according to the core feature dimensions of passengers and a sliding time window is used to calculate the feature values ​​to obtain multi-dimensional features of passengers;

[0020] Tag generation: Based on the feature values, all tags are aggregated by passenger ID number to generate a dynamic tag table. Each row corresponds to a passenger ID number, and the column fields are the multi-dimensional feature tags of the specific passenger.

[0021] Furthermore, the bimodal prediction method includes:

[0022] Demand Category Determination: Based on passenger multidimensional feature tags and environmental data, demand categories are determined, including: regular demand and sudden demand;

[0023] Routine demand forecasting: Based on the multi-dimensional feature tags of passengers, routine demand is broken down into several sub-tasks, the probability of each sub-task is calculated, and routine service demand records are generated. The routine service demand records include: passenger ID number, sub-task list, probability, and valid time.

[0024] Sudden Demand Forecast: Determine the type of sudden event, combine passenger multi-dimensional feature tags to generate several sub-tasks, assign a severity level to each sub-task, and generate a sudden service demand record. The sudden service demand record includes: passenger ID number, sub-task list, severity level, event location, and effective time.

[0025] Service task generation: Simultaneously receive regular service demand records and emergency service demand records. If the two types of service demands overlap in the time window for the same passenger, the emergency service demand takes priority and the regular service demand is postponed. Calculate the priority of each subtask, generate the initial passenger service demand, and write it into the dynamic master data matrix, marking it as a predicted demand.

[0026] Furthermore, the multidimensional dynamic matching method includes:

[0027] Dataset extraction: Extract real-time and predicted demand from the dynamic master data matrix to establish a set of passenger demand to be responded to; extract the set of currently schedulable service personnel from the dynamic master data matrix;

[0028] Define matching dimensions and quantification rules: Define the matching dimensions between service personnel and passenger service needs, as well as the quantification rules for these matching dimensions; the matching dimensions include: spatial and temporal accessibility, skill matching degree, service value ratio, and staff fatigue level;

[0029] Candidate matching: Based on hard filtering rules, the initial screening calculation yields a candidate matching set of demand and personnel;

[0030] Multidimensional dynamic matching: For each demand-person candidate matching set, the matching dimension index is calculated and normalized to obtain a comprehensive score and an initial service plan;

[0031] Service plan generation and output: Based on the initial service plan, a structured scheduling plan table is generated, with each row including: service requirements, service personnel, estimated arrival time, skill matching degree, service value score, and fatigue risk level.

[0032] Furthermore, the service execution and monitoring steps include:

[0033] Establish a real-time passenger-service mapping space: Link passenger information and service personnel information to the dispatch plan table, and dynamically display passenger location, service personnel location and service status through a digital dashboard; generate a real-time service tag for each passenger and mark the service personnel status;

[0034] Service execution: The scheduling plan table is pushed to the service personnel. After receiving it, the service personnel update the service status to "in progress" and start the timer.

[0035] Real-time monitoring of service gaps: Real-time collection of high-speed rail station operation data, passenger information, service personnel information, and environmental data to calculate time gaps, demand gaps, and personnel gaps;

[0036] Dynamic evaluation and adjustment of service quality: Real-time evaluation of service timeliness, completeness and satisfaction, and root cause analysis to obtain analysis results; Based on the analysis results, the initial service plan is dynamically adjusted and synchronously returned to the mapping space;

[0037] Service execution effectiveness report generation: After each service request is completed, the passenger status and service personnel performance are automatically updated to generate a service execution effectiveness report.

[0038] Furthermore, the method for constructing the real-time passenger-service mapping space includes:

[0039] Data preparation: Import the dispatch plan table, passenger information and service personnel information into the passenger-service real-time mapping database, and unify the passenger ID number and service personnel number as the primary key;

[0040] Passenger layer construction: Each passenger record generates a dynamic object with fields including: ID number, train number, schedule node, schedule time, real-time location, and service tag;

[0041] Service personnel layer construction: Each service personnel record generates a dynamic object with fields including: employee number, skills, real-time location, task list, and service status;

[0042] Node layer construction: Convert all service nodes of the high-speed rail station into spatial points, with capacity thresholds, current number of people and queuing time;

[0043] Coordinate access: Real-time access to the real-time location of passengers and service personnel.

[0044] Furthermore, the feedback optimization step includes:

[0045] Collection and Cleaning: Service performance reports are collected into the feedback pool, and unstructured feedback is converted into a unified score to generate a passenger-event-score-evaluation record;

[0046] Record Analysis: Performs attribution analysis on passenger-event-score-evaluation records, automatically labels root causes, and generates root cause reports;

[0047] Optimization and Update: Based on the root cause report, update and optimize the service personnel scheduling process.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] A data-driven method for scheduling service personnel in high-speed rail business travel is proposed. By establishing a service execution and monitoring method, service gaps can be detected and service quality evaluated in real time, and service plans can be dynamically adjusted to improve the passenger service experience.

[0050] Meanwhile, by establishing a bimodal service demand forecasting method, multidimensional features of passengers are extracted and combined with environmental data to predict passenger demand, realizing the shift from passive response to proactive forecasting, accelerating response speed, reducing passenger waiting time, and improving service quality and efficiency. By establishing a multidimensional dynamic scheduling method, unavailable personnel are automatically avoided, candidate personnel are initially identified, and candidate personnel are dynamically matched in multiple dimensions to adapt to dynamic demand while reducing resource mismatch. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the high-speed rail business travel service personnel dispatching method of the present invention;

[0052] Figure 2 This is a schematic diagram of the method for constructing the dynamic master data matrix according to the present invention;

[0053] Figure 3 This is a schematic diagram of the dual-modal service demand prediction method of the present invention;

[0054] Figure 4 This is a schematic diagram of the multi-dimensional dynamic scheduling method of the present invention;

[0055] Figure 5 This is a schematic diagram of the service execution and monitoring method of the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.

[0057] like Figure 1 As shown, the present invention provides a technical solution: a method for scheduling high-speed rail business travel service personnel based on data analysis, comprising:

[0058] Step 1: Data Collection and Integration: Collect high-speed rail station operation data, passenger information, service personnel information, and environmental data; perform streaming data processing and spatiotemporal alignment on the collected data, and construct a dynamic master data matrix.

[0059] like Figure 2 As shown, the present invention provides a method for constructing a dynamic master data matrix;

[0060] Specifically:

[0061] Data processing: Establish unique linear mileage coordinates and a unified time base, and map all data events to the same spatiotemporal coordinate system; define and unify primary keys for high-speed rail station operation data, passenger data, service personnel information, and environmental data, and establish a primary key mapping table; and perform streaming data processing, aligning data by train number, time window, and spatial window to generate unified aligned data.

[0062] Create an empty matrix template: Select the primary key and predefine an empty matrix based on the primary key;

[0063] Dynamic matrix filling: Map each aligned data entry to a corresponding cell. For multiple records in the same cell, mapping is performed by overwriting and aggregating the latest value to obtain a dynamic master data matrix.

[0064] Verification and Update: Real-time detection of data integrity, marking of anomalies, updating of the dynamic master data matrix, and retention of historical versions.

[0065] It is important to note the following regarding data collection: The collected high-speed rail station operation data includes: train timetables, stops, carriage layout, real-time location, delay status, security check logs, etc.; this data is obtained through railway dispatching system APIs (such as CTC / TDMS), onboard sensors (IoT devices), and the 12306 railway platform. Passenger information collected includes: passenger ID, ticket purchase records, historical behavior and preferences, job level, real-time needs, real-time location, service records, and feedback; this information is obtained through enterprise authentication systems, ticketing system databases (SQL / Oracle), mobile app logs, and onboard Wi-Fi / Bluetooth probe data. Service personnel information collected includes: employee number, position, skills, shift schedule, real-time location, service status, and service records; this information is obtained through the high-speed rail station personnel system database (SQL / Oracle), indoor positioning systems (UWB base stations), and smart badges (IoT sensors). The collected environmental data includes: temperature and humidity in the carriages, air quality, weather, and holiday event information; environmental data is obtained through onboard environmental sensors, meteorological bureau data, and social media sentiment analysis (Python web crawlers). Lightweight streaming engines (such as Flink Edge) are deployed at train gateways or station edge nodes to filter invalid data, perform preliminary aggregation, and compress data volume to reduce cloud load. All data is ultimately fed into a central streaming platform (such as Flink) for further processing.

[0066] Data Processing: Spatiotemporal Alignment: The high-speed rail line is decomposed into a linear reference system of [line ID, mileage marker]. Geographic coordinate transformation is achieved using GIS tools (such as ArcGIS), and the position of moving trains is processed using interpolation algorithms. UTC timestamps are used, and the time of each subsystem is synchronized through an NTP server to solve the problem of cross-time zones. Time windows are divided according to business needs, and data within the same time window is considered as the same batch. Spatial windows are divided according to carriages or mileage segments. A piece of data must belong to a combination of [train number, time window, spatial window]. Primary Key Mapping: The primary key structure is [train number, time window, spatial window]. A high-performance key-value database (such as Redis) is used to store the relationship between the primary key and the original data. Flink window functions are used to aggregate data by primary key.

[0067] Create an empty matrix data template: First, design the matrix dimensions, using a unified primary key as the row index and data attributes as the column index. Use sparse matrices to pre-allocate space to avoid memory waste in all-zero matrices.

[0068] Dynamic matrix filling: Data mapping rules: Latest value overwrites, suitable for real-time status data, such as temperature and location; aggregation calculation, suitable for statistical data, such as cumulative passenger counts and average temperature and humidity. Multiple data records under the same primary key are processed according to priority, such as sensor data taking precedence over manually reported data.

[0069] Verification and Updates: Integrity checks employ a dynamic matrix windowing integrity verification protocol to check the integrity of the dynamic master data matrix, ensuring that each window contains data; if no data is detected, it is marked as an anomaly. Consistency checks use Drools to check for logical contradictions; data under the same primary key is not allowed to have logical conflicts, such as the number of passengers not being both 10 and 20. Delta Lake saves the matrix state every 10 minutes, supporting time travel queries; Kafka retains data update events for backtracking.

[0070] Step 2: Service demand prediction. Extract multi-dimensional features of passengers from the dynamic master data matrix and generate multi-dimensional feature labels for passengers. Based on the multi-dimensional feature labels of passengers and combined with environmental data, perform bimodal demand prediction to obtain the initial passenger service demand.

[0071] like Figure 3 As shown, the present invention provides a bimodal service demand forecasting method;

[0072] Specifically:

[0073] Feature Dimension Definition: Define the core feature dimensions of passengers. The core feature dimensions are divided into static dimensions and dynamic dimensions.

[0074] Feature extraction: Based on the dynamic master data matrix, multi-dimensional features of passengers are extracted according to the core feature dimensions of passengers and a sliding time window is used to calculate the feature values ​​to obtain multi-dimensional features of passengers;

[0075] Tag generation: Based on the feature values, aggregate all tags by passenger ID to generate a dynamic tag table. Each row corresponds to a passenger ID, and the column fields are the multi-dimensional feature tags of the specific passenger.

[0076] Demand Category Determination: Based on passenger multidimensional feature tags and environmental data, demand categories are determined, including: regular demand and sudden demand;

[0077] Routine demand forecasting: Based on the multi-dimensional feature tags of passengers, routine demand is broken down into several sub-tasks, the probability of each sub-task is calculated, and a routine service demand record is generated. The routine service demand record includes: passenger ID, sub-task list, probability, and effective time.

[0078] Sudden Demand Forecast: Determine the type of sudden event, combine passenger multi-dimensional feature tags to generate several sub-tasks, assign a severity level to each sub-task, and generate a sudden service demand record. The sudden service demand record includes: passenger ID, sub-task list, severity level, event location, and effective time.

[0079] Service task generation: Simultaneously receive regular service demand records and emergency service demand records. If the two types of service demands overlap in the time window for the same passenger, the emergency service demand takes priority and the regular service demand is postponed. Calculate the priority of each subtask, generate the initial passenger service demand, and write it into the dynamic master data matrix, marking it as a predicted demand.

[0080] It is important to note that the acquisition of multi-dimensional passenger feature tags involves two dimensions: Static dimensions, consisting of low-frequency updated data, serve as the basis for service resource pre-allocation and include: basic identity information such as ID number, job level, and industry; fixed service needs such as VIP membership status, VIP membership level, and historical service preferences; and physical characteristics such as disability certification and frequently carried baggage specifications. Dynamic dimensions, consisting of real-time / near-real-time updated data, serve as the basis for scheduling decisions and include: travel behavior such as ticket purchase frequency in the past week, average advance booking time, and whether a transfer is required; consumption characteristics such as the percentage of business class seats in the past month and additional service spending; spatiotemporal characteristics such as high-frequency travel times and frequently used stations; and service feedback such as the types of the last three service complaints and positive review tags.

[0081] Feature extraction involves removing invalid trips, such as refunded tickets, from the dynamic master data matrix. Sliding window configurations include: short-cycle windows (real-time scheduling), which count service request counts in 15-minute windows; and long-cycle windows (predictive scheduling), which calculate business class travel frequency in 30-day windows. Feature values ​​are calculated, including: service dependency = historical service request counts / historical travel counts; price sensitivity = discounted ticket purchase counts / total ticket purchase counts; and environmental adaptability = temperature and humidity adjustment request counts / travel hours. All feature values ​​are then min-max normalized, with the normalized value calculated as (original value - minimum value) / (maximum value - minimum value).

[0082] Tag generation includes the following types: identity characteristic tags, such as cross-border business travelers; behavioral preference tags, such as cold food preference; service sensitivity tags, such as time-sensitive and quality-priority; equipment dependence tags, such as requiring power bank delivery; and health tags, such as accessibility service needs. Features are aggregated according to traveler ID, and pre-defined tag generation rules are used. Each rule clearly defines the mapping relationship between specific characteristic conditions and corresponding tags. For example, if a traveler's cold food service requests exceed 60% in the past 30 days, they are tagged with "cold food preference"; if their three most recent service reviews all give negative feedback on service timeliness, they are tagged with "time-sensitive." If the same traveler's characteristics simultaneously meet multiple conflicting tag conditions, the highest priority tag is retained according to pre-defined priorities, or a composite tag is generated. All multi-dimensional characteristic tags of all travelers are stored in a dynamic tag table, with each row corresponding to one traveler, using the traveler ID as the primary key; each column represents a tag type, with field values ​​being Boolean (yes / no); and includes the last update timestamp to ensure data timeliness and traceability. Dynamic feature changes trigger timely tag updates. Service personnel can manually add special tags, but the operator and reason must be noted.

[0083] Demand Category Determination: The system reads passenger multi-dimensional feature tags and environmental monitoring data streams, and determines the demand category based on established rules. These rules include: Regular Demand Determination Conditions (must be met simultaneously): The passenger's multi-dimensional feature tag indicates "Business Traveler" or "Frequent Traveler"; the current time is during weekday morning / evening rush hours (7:00-9:00 AM and 5:00-7:00 PM); there are no unforeseen events, such as train delays. Unexpected Demand Determination Conditions (any one must be met): High-speed rail station operation data detects a train delay of ≥15 minutes; the passenger's multi-dimensional feature tag displays "Abnormal Heart Rate"; the security check system reports a lost baggage alarm. A demand category label is generated for each business traveler, categorized into three types: Regular Demand, Unexpected Demand, and No Service Required.

[0084] Standard Demand Forecasting: Subtasks include VIP lounge guidance, exclusive meal delivery, and business equipment rental. When a member's level is ≥ Gold Card and their historical usage rate is >30%, the VIP lounge guidance task is triggered, with a probability of 0.7 + 0.1 × member level coefficient, where the member level coefficient is 0 for regular cards, 0.3 for silver cards, 0.6 for gold cards, and 1.0 for platinum cards. When a traveler's multidimensional feature tag includes "dietary preference" and the current time is mealtime, the exclusive meal delivery task is triggered, using a logistic regression model to output the probability. When electronic device usage ranks in the top 20%, the business equipment rental task is triggered, using a Bayesian network to infer the probability.

[0085] The specific prediction steps include: matching the applicable sub-task type based on the passenger's multi-dimensional feature labels; calling the pre-set probability model to calculate the probability of each sub-task occurring; filtering low-confidence predictions with a probability value < 0.6; and generating regular service demand records, each of which includes: passenger ID, a list of sub-tasks (including type and probability value), and an effective time window (calculated based on the train timetable).

[0086] Sudden Demand Forecasting: Identifying sudden event types, including: medical events identified through multi-dimensional passenger feature tags; itinerary changes identified through delay information pushed by the train dispatch system and passengers' subsequent transfer tickets; and lost items identified through the duration of RFID baggage tag loss. Then, the severity level of the event is quantified. For example, the severity level of a medical emergency is calculated as: base level + n × 0.1, where the base level is 2, and n is the number of minutes of treatment delay; the severity level range for medical emergencies is 2-5. The severity level of a train delay is calculated as: delay minutes / 10; the severity level range for train delays is 1-3. The severity level of lost baggage is fixed at 3. The event location is automatically associated, such as gate 13A. The effective time is calculated as: event detection time + estimated processing time; the estimated processing time for medical events is 30 minutes by default.

[0087] Service task generation includes: a demand merging strategy: time conflict determination—if the overlap rate of the effective time windows of two types of service demands is greater than 50%, a time conflict occurs; priority rules—emergency demands always take precedence over regular demands, and tasks of the same type are executed in descending order of priority score. The priority of regular demands = probability × passenger value coefficient, where passenger value coefficient = (0.5 × membership level coefficient + 0.5 × normalized value of historical consumption); the priority of emergency demands = severity level × location urgency coefficient, where location urgency coefficient = 1 / (distance from service point in meters + 1). The final demand record is written to the "Predicted Demand" partition of the dynamic master data matrix, with fields including: demand ID, task list, and effective time. The demand ID includes a timestamp and passenger ID hash; the task list includes type, priority, and location; and the effective time includes start and end times. The passenger ID hash refers to a fixed-length numeric value generated by applying a hash function to passenger information, i.e., a hash value.

[0088] Step 3: Multi-dimensional dynamic scheduling: Based on the dynamic master data matrix, service personnel are matched in multiple dimensions according to the initial passenger service needs to generate an initial service plan.

[0089] like Figure 4 As shown, this invention provides a multi-dimensional dynamic scheduling method;

[0090] Specifically:

[0091] Dataset extraction: Extract real-time and predicted demand from the dynamic master data matrix to establish a set of passenger demand to be responded to; extract the set of currently schedulable service personnel from the dynamic master data matrix;

[0092] Define matching dimensions and quantification rules: Define the matching dimensions between service personnel and passenger service needs, as well as the quantification rules for these matching dimensions; the matching dimensions include: spatial and temporal accessibility, skill matching degree, service value ratio, and staff fatigue level;

[0093] Candidate matching: Based on hard filtering rules, the initial screening calculation yields a candidate matching set of demand and personnel;

[0094] Multidimensional dynamic matching: For each demand-person candidate matching set, the matching dimension index is calculated and normalized to obtain a comprehensive score and an initial service plan;

[0095] Service plan generation and output: Based on the initial service plan, a structured scheduling plan table is generated, with each row including: service requirements, service personnel, estimated arrival time, skill matching degree, service value score, and fatigue risk level.

[0096] It is important to note the following data extraction process: From the real-time demands marked "Pending Response" in the dynamic master data matrix, extract unexpired service demands (planned service times later than the current time), excluding cancelled or completed historical service demands; merge these with predicted demands to form a set of pending passenger demands, sorted in descending order of urgency to ensure high-priority demands are processed first. Then, extract dispatchable service personnel from the dynamic master data matrix, selecting currently on-duty personnel, excluding those on leave or time off, and ensuring their current task load does not exceed their maximum capacity (e.g., handling a maximum of 4 tasks simultaneously), and that there have been no overload alarm records in the past hour; load personnel skill information and historical service ratings. Convert passenger demand locations and service personnel locations into a unified station grid code, such as grid area H-2 for the 3rd floor business waiting area. When a passenger location is missing, infer the default location based on ticket information; when service personnel skill information expires, temporarily mark it as "requires manual review"; thus establishing a set of dispatchable service personnel. Add a timestamp version to the extracted data.

[0097] Output a set of passenger requests awaiting response and a set of dispatchable service personnel. Each record in the passenger request set includes: a unique request ID, passenger location and service request, urgency level and deadline, and a list of required skills. Each record in the dispatchable service personnel set includes: a personnel ID and current location, current workload, a list of skills and qualifications, and real-time fatigue status. Scan for data changes every 30 seconds, processing only newly added or modified records. When multiple requests from the same passenger conflict, retain the request with the highest urgency. Ensure that the timestamps of the request and service personnel data are consistent to avoid matching errors.

[0098] Define matching dimensions and quantification rules: Spatiotemporal accessibility: Using high-speed rail station GIS maps and personnel location data, calculate the shortest path and estimated arrival time from the service personnel's current location to the passenger's desired point; dynamically adjust the time window based on real-time train arrival information to ensure arrival within the passenger's effective service time. Skill matching degree: Vectorize passenger service needs, such as foreign language services and medical emergencies, with service personnel's skills, and calculate matching scores using a cosine similarity algorithm. For complex needs, such as English plus wheelchair assistance, a weighted fusion calculation is performed.

[0099] Service value ratio is assessed based on historical data. Value ratio = (Average historical service rating of personnel × Current urgency coefficient) / Personnel unit time labor cost; where, the current urgency coefficient = base weight × time decay factor; the base weight is a preset value, such as 1.5 for emergency medical care and 1.0 for general needs; the time decay factor = ... The response time threshold is 15 minutes; the personnel unit time labor cost = (monthly salary / monthly working hours) × job coefficient, where the job coefficient includes: 1.0 for general service personnel, 1.5 for foreign language specialists, and 2.0 for medical personnel;

[0100] Staff fatigue assessment integrates real-time physiological monitoring data, such as heart rate monitoring on smart work badges, with scheduling system data, setting three fatigue risk levels (red / yellow / green) to reduce workload scheduling. Fatigue level = current continuous working hours × task intensity coefficient + daily cumulative service time penalty. The task intensity coefficient is calculated as follows: guidance services have a base intensity of 0.8, increasing by 0.1 for every 100 meters of walking distance; medical treatment tasks have a base intensity of 1.5, increasing by 0.3 for every 10 minutes of operation time; multilingual translation services have a base intensity of 1.2, increasing by 0.2 for every 15 minutes of continuous dialogue. The daily cumulative service time penalty is calculated as follows: daily cumulative working hours ≤ 4 hours, no penalty; 4 hours ≤ daily cumulative working hours ≤ 6 hours, for the portion exceeding 4 hours, an additional penalty of 0.5 is applied; daily cumulative working hours > 6 hours, for the portion exceeding 6 hours, an additional penalty of 0.8 is applied.

[0101] Candidate Matching: Based on strict filtering rules, the current set of passenger demands awaiting response and the set of available personnel are initially screened. These rules include: personnel arrival time ≤ maximum effective time; personnel skill tags covering core skills required, such as mandatory first aid certification; and personnel fatigue levels not under a red alert. This initial screening yields a set of demand-personnel candidate matches.

[0102] Multidimensional dynamic matching: For each demand-person candidate pairing in the demand-person candidate pairing set, a four-dimensional index score is calculated: Spatiotemporal accessibility score, 1 - (estimated arrival time / maximum allowable time); Skill matching score, cosine similarity between the demand skill vector and the service personnel skill vector; Service value ratio score, using the TVER model to obtain the service value ratio score; Personnel fatigue score, fatigue decay score = The above four-dimensional scores are normalized. The thresholds for spatiotemporal accessibility and fatigue decay scores are 0-1, so no normalization is needed. The theoretical threshold for skill matching score is [-1, 1], but in practice, similarity is usually ∈ [0, 1]. If a negative value appears, normalization is required: (skill matching score + 1) / 2. Service value ratio score is normalized as (service value ratio score - minimum within the group) / (maximum within the group - minimum within the group), where the minimum / maximum within the group refers to the lower / upper limit of the historical service value ratio scores of all service personnel in the same shift. The scores are divided into configuration dimension weight vectors, such as spatiotemporal accessibility being 0.4, skill matching score being 0.3, service value ratio being 0.2, and personnel fatigue being 0.1. The comprehensive score is calculated as: Comprehensive score = w1 × spatiotemporal accessibility score + w2 × skill matching score + w3 × service value ratio score + w4 × fatigue decay score. Then, the Hungarian algorithm is used to solve the comprehensive score matrix to obtain the optimal initial service plan: establish a demand-personnel bipartite graph with the comprehensive score as the edge weight; and maximize the overall service benefits through maximum weight matching.

[0103] Service plan generation and output: Generate a structured dispatch plan table, with each row including service requirements, service personnel, estimated arrival time, skill matching degree, service value score, and fatigue risk level. Real-time push of the dispatch plan table to mobile terminals: Service personnel receive navigation routes and service requirements via their apps; the dispatch center's large screen monitors the trajectories and task status of all service personnel across the station.

[0104] Step 4: Service Execution and Monitoring: Establish a real-time passenger-service mapping space. While executing the initial service plan, service personnel monitor service gaps in real time, evaluate service quality, dynamically adjust the initial service plan, and obtain a service execution effect report.

[0105] like Figure 5 As shown, the present invention provides a service execution and monitoring method;

[0106] Specifically:

[0107] Establish a real-time passenger-service mapping space: Link passenger information and service personnel information to the dispatch plan table, and dynamically display passenger location, service personnel location and service status through a digital dashboard; generate a real-time service tag for each passenger and mark the service personnel status;

[0108] Service execution: The scheduling plan table is pushed to the service personnel. After receiving it, the service personnel update the service status to "in progress" and start the timer.

[0109] Real-time monitoring of service gaps: Real-time collection of high-speed rail station operation data, passenger information, service personnel information, and environmental data to calculate time gaps, demand gaps, and personnel gaps;

[0110] Dynamic evaluation and adjustment of service quality: Real-time evaluation of service timeliness, completeness and satisfaction, and root cause analysis to obtain analysis results; Based on the analysis results, the initial service plan is dynamically adjusted and synchronously returned to the mapping space;

[0111] Service execution effectiveness report generation: After each service request is completed, the passenger status and service personnel performance are automatically updated to generate a service execution effectiveness report.

[0112] It's important to note that the first step, establishing the real-time passenger-service mapping space, involves data preparation. This involves batch importing the scheduling plan table, passenger information, and service personnel information into the passenger-service real-time mapping database, such as a Redis+Neo4j graph database. The primary keys are set as follows: the passenger-side primary key is [ID number, train number], and the service-side primary key is [employee number, shift]. Then, the data is cleaned, multiple service requests from the same passenger are merged, and the service personnel's skill tags are verified to match the scheduling plan.

[0113] The passenger layer is constructed by creating dynamic data objects for each passenger, including: basic attributes such as ID number and train number; planning information such as the expected service node (e.g., ticket gate, VIP lounge) and time window; real-time status such as current location coordinates obtained through positioning devices; and service tags such as service demand labels generated based on predictions (e.g., "requires VIP guidance"). The passenger layer dynamically updates the planned node / time every 5 minutes, synchronizing with the train number; and updates the real-time location every second via Bluetooth beacon / GPS positioning.

[0114] The service personnel layer is structured to create dynamic data objects for each service personnel, including: qualification information (employee number, skill certificates held); task information (currently assigned task list and service status); real-time location (current location coordinates obtained through UWB ultra-wideband technology in the smart ID card); and load status (current number of service personnel and maximum capacity). Service status includes idle, running, and overloaded. Service status determination rules: if the task list is empty, it is considered idle, and new tasks are automatically received; if the number of tasks is greater than zero and less than the maximum load, it is considered running, and the current task continues to be executed; if the number of tasks is greater than or equal to the maximum load, it is considered overloaded, and the assignment of new tasks is suspended.

[0115] The node layer is constructed by modeling all service stations within the high-speed rail station as spatial nodes. Each node includes: fixed attributes such as geographical location and maximum capacity; dynamic indicators such as current number of people and average queuing time; and a two-level early warning mechanism is set up, with a yellow warning when the current number of people reaches 80% of the capacity and a red warning when the current number of people reaches 95% of the capacity.

[0116] Real-time coordinate access is achieved by combining mobile phone GPS and Bluetooth beacon data to obtain real-time indoor and outdoor passenger positioning, while ultra-wideband technology is used to obtain real-time positioning of service personnel. WGS84 latitude and longitude coordinates are then converted to the station's local coordinate system, enabling data fusion. In the event of positioning loss, the current position can be estimated based on the last location and movement speed. For positioning drift data, Kalman filtering can be used to smooth the trajectory. The conversion of WGS84 latitude and longitude coordinates to the station's local coordinate system refers to the process of mapping globally unified geocentric coordinate (WGS84) data to an independent planar coordinate system specific to the engineering scenario.

[0117] Construct a dynamic relationship graph, defining node relationships: Passenger-Service Personnel, currently serving / awaiting service; Passenger-Node, current location / planned destination; Service Personnel-Node, stationed / moving; use Neo4j to build a real-time relationship network. Digital dashboards (such as GIS maps or 3D terminal models) display the congestion level of each node in heatmap format; dynamically display personnel movement trajectories and service progress; provide standard API interfaces to support querying the real-time location and service status of passengers / service personnel.

[0118] The second step, service execution, involves loading the scheduling plan into the service terminal device, prioritizing tasks, and generating a personalized task list for each service personnel. Resources are reserved for tasks requiring special equipment (such as wheelchairs), and the capacity quotas of relevant service nodes are locked in the real-time mapping space. Service personnel are notified via vibration and visual cues on their smart badge terminals. After receiving the task, the service personnel click to confirm, automatically updating the task status to "in progress" and automatically triggering UWB positioning tracking. Service progress evidence is collected every 30 seconds, such as taking photos and uploading material handover status.

[0119] The third step is to monitor service gaps in real time: Calculate time gaps using the dispatch plan table and service personnel location, Σ(current time - latest task start time) > 0; calculate demand gaps using the passenger demand list and service personnel status, calculated as the number of unresponsive demands / the number of currently available personnel; calculate personnel gaps using the set of dispatchable service personnel and the set of pending service demands, i.e., the number of tasks requiring skill A - the number of available personnel with skill A. A three-tiered early warning mechanism is established: a yellow warning is issued when the accumulated time gap exceeds 5 minutes, prompting service personnel to accelerate; an orange warning is issued when the demand gap exceeds 2:1 and persists for 10 minutes, initiating support from nearby areas; a red warning is issued when a shortage of critically skilled personnel leads to medical delays, triggering cross-shift emergency dispatch.

[0120] Step 4: Dynamic Evaluation and Adjustment of Service Quality: Timeliness Evaluation. Core indicators include: calculating the deviation rate between actual and estimated service time. For example, if a task planned to take 15 minutes actually takes 18 minutes, the deviation rate = [(18-15) / 15] × 100% = 20%. A deviation rate ≤ 10% indicates excellent timeliness, recorded in the high-efficiency case library; 10% < deviation rate < 30% indicates acceptable timeliness, with optimization suggestions pushed; a deviation rate ≥ 30% indicates unacceptable timeliness, triggering root cause analysis. Delays caused by external factors (such as train delays) will automatically adjust the evaluation baseline time.

[0121] Integrity assessment includes: key node verification, which combines IoT devices with human confirmation to verify critical actions throughout the service process. For example, in baggage assistance, mandatory verification nodes include pickup, transportation, and delivery signature. Verification methods can include scanning a code or taking a photo. It also includes integrity coefficient calculation: integrity coefficient = number of completed nodes / number of marked nodes. A integrity coefficient < 0.8 indicates a service deficiency. For missing non-critical nodes, such as greetings, the service is recorded but not interrupted; for missing critical nodes, such as medical confirmation, the service is immediately terminated and remedial measures are triggered.

[0122] Satisfaction assessment involves multimodal data collection: explicit feedback, including passenger-initiated ratings (1-5 stars) and rating tag selections (e.g., "professional," "poor attitude"); implicit analysis, including voiceprint emotion recognition and behavioral trajectory analysis. A satisfaction index is calculated based on the collected multimodal data: Satisfaction Index = 0.6 × Star Rating + 0.2 × Emotion Score + 0.1 × (1 - Repeat Request Rate) + 0.1 × Positive Keyword Ratio. The satisfaction index ranges from 0 to 1, with a score of 0.8 or higher indicating excellent service. When the satisfaction index is detected to be <0.6, the employee is automatically upgraded to a senior service staff member and receives compensation benefits.

[0123] Deviation rate, completeness coefficient, and satisfaction index are uniformly converted into scores ranging from 0 to 1 to ensure fairness in weighting. Weights are dynamically adjusted to calculate the overall service quality score: For routine services, emphasis is placed on service standards to ensure standardized service processes and avoid sacrificing quality for speed. The overall service quality score = 0.3 × Timeliness score + 0.4 × Completeness score + 0.3 × Satisfaction score. For emergencies, response speed is prioritized. In scenarios such as medical emergency, rapid arrival is more important than complete etiquette. The overall service quality score = 0.5 × Timeliness score + 0.3 × Completeness score + 0.2 × Satisfaction score. For example, in handling lost luggage (an emergency), with individual scores of: Timeliness 0.8, Completeness 0.6, and Satisfaction 0.7, the overall service quality score = 0.5 × 0.8 + 0.3 × 0.6 + 0.2 × 0.7 = 0.73.

[0124] Root cause analysis, problem identification and classification, includes: anomaly detection, automatically detecting three key service quality assessment indicators: timeliness (e.g., baggage delivery time exceeding the plan by 30%); completeness (e.g., missing key steps in the service process); and satisfaction (e.g., a drop of more than one star in passenger ratings within a day). When an indicator continuously exceeds a threshold, the analysis process is triggered. Problem classification includes: isolated, occasional problems (e.g., individual passenger complaints); and systemic anomalies (e.g., inefficiency of the entire team).

[0125] Data tracing involves reviewing the entire data chain from 15 minutes before the anomaly occurred to 5 minutes after it occurred. The data to be reviewed includes: the real-time location of service personnel, task execution records, skill qualifications, surveillance camera footage of service nodes, passenger flow within the station, train delay information, temperature and humidity, complaint text content, service evaluation tags, etc.

[0126] Multi-dimensional correlation analysis includes: personnel factor investigation, checking whether service personnel possess the necessary skills for the task, analyzing the rationality of movement paths, and assessing whether the current workload exceeds capacity. System factor investigation, verifying whether positioning devices are functioning properly, checking for delays in task dispatch, and testing whether data synchronization is real-time. Environmental factor investigation, identifying whether abnormal events are concentrated in specific areas or time periods, and detecting whether the physical environment affects service.

[0127] Root cause analysis involves tracing the problem layer by layer from the surface until the root cause is found. For example, regarding the issue of excessively long passenger wait times: Inquiry 1: Service personnel did not arrive on time; Inquiry 2: Navigation guidance was incorrect; Inquiry 3: The map was not updated for the renovation area; Inquiry 4: Construction information was not synchronized with the dispatch system; Thus, the root cause was found to be a missing cross-departmental data interface. The discovered root causes are then categorized, including: human error, system deficiencies, process loopholes, and external force majeure, leading to the analysis results.

[0128] Based on the analysis results, a solution is generated, and the initial service plan is dynamically adjusted. If the path is incorrect, the corrected path is immediately pushed and the map data is updated. The adjusted initial service plan is then synchronized back to the mapping space.

[0129] Step 5: Service Execution Performance Report Generation: After each service request is completed, service personnel mark the task as completed using a smart badge. Actual service data is automatically recorded, including: actual start and end times, completion status of key task milestones, and records of any abnormal events. If any key task milestones are not completed, the reason for the interruption is recorded, such as a passenger canceling their request, service personnel terminating service due to unforeseen circumstances, or equipment malfunction preventing service continuation. Service details are added to the passenger's individual service record, including: the type and specific content of the service received, the difference between the actual and planned service duration, the employee number responsible for the service, and a service completion quality assessment. The employee's work file accumulates the service performance data, including: service punctuality, service completeness, and service efficiency. The system automatically calculates and updates the service personnel's monthly on-time service rate, recent average service quality score, and service skill proficiency trend.

[0130] Integrate various data from the service process to generate a service execution effectiveness report, which includes: basic information, service time, location, type, and participating personnel; quality assessment, the difference between actual and planned completion time, whether all service steps were completed, and passenger evaluation results; root cause analysis, explaining the root causes of substandard services; and optimization suggestions, providing specific optimization solutions for the problems.

[0131] Step 5: Feedback and Optimization: Based on the service performance report, update and optimize the service personnel scheduling process.

[0132] Collection and Cleaning: Service performance reports are collected into the feedback pool, and unstructured feedback is converted into a unified score to generate a passenger-event-score-evaluation record;

[0133] Record Analysis: Performs attribution analysis on passenger-event-score-evaluation records, automatically labels root causes, and generates root cause reports;

[0134] Optimization and Update: Based on the root cause report, update and optimize the service personnel scheduling process.

[0135] It's important to note that NLP models are used to convert textual passenger reviews into standardized scores of 1-5. For example, "good service attitude but slow speed" would be broken down into +2 points for attitude and -1 point for timeliness. Invalid data, such as duplicate reviews or purely symbolic content, is removed. A unified record is generated, with each record including passenger ID, event type, standardized score, and original review text, such as "Passenger G123 Xiaoming, wheelchair delivery service, score 3 points, guide professional but delivery slow." Problems are categorized into three types: response timeliness (e.g., staff not arriving on time); skill matching (e.g., assigned staff not speaking a foreign language); and load balancing (e.g., one staff member handling 5 tasks). Based on the passenger-event-score-review record, the decision-making logic for staff scheduling at the time of the problem is examined, including the list of available staff and their location / skills / load; and task priority settings.

[0136] Root cause analysis reveals that, for example, if service personnel are late, it may be because the navigation route did not take elevator maintenance into account; if skills are mismatched, it may be because skill tags were not updated in a timely manner; and if tasks are piling up, it may be because there are errors in the multi-dimensional dynamic matching of service personnel.

[0137] Generate root cause reports, such as "12 response delays occurred in Area A on August 12th. The root cause is: the navigation system did not synchronize the temporary closure of the passage (80%), and the personnel load warning threshold was set too high (20%)." Optimize the service personnel dispatch process based on the root cause reports, such as adding rules to automatically check for alternatives when the real-time location of a service personnel is more than 200 meters away from the task point; and requiring foreign language service tasks to match the "recently passed language test" skill tag.

[0138] The optimization effect is verified through A / B testing, with comparison indicators including on-time response rate, passenger satisfaction, and staff load balance. If the optimized scheduling process causes a certain indicator to decrease by 10%, the system will automatically roll back to the previous version.

[0139] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended embodiments and their equivalents.

Claims

1. A method for scheduling high-speed rail business travel service personnel based on data analysis, characterized in that, include: Data collection and integration: Collect high-speed railway station operation data, passenger information, service personnel information, and environmental data; The collected data is subjected to streaming data processing and spatiotemporal alignment, and a dynamic master data matrix is ​​constructed. Service demand forecasting: Extract multi-dimensional features of passengers from the dynamic master data matrix and generate multi-dimensional feature labels for passengers; Based on multidimensional passenger feature labels and combined with environmental data, bimodal demand prediction is performed to obtain the initial passenger service demand. Multidimensional dynamic scheduling: Based on the dynamic master data matrix, service personnel are matched in multiple dimensions according to the initial passenger service needs to generate an initial service plan; Service execution and monitoring: Establish a real-time passenger-service mapping space, where service personnel monitor service gaps in real time while executing the initial service plan, evaluate service quality, dynamically adjust the initial service plan, and obtain service execution effect reports; Feedback and optimization: Update and optimize the service personnel scheduling process based on the service performance report; The dual-modal prediction method includes: N1: Demand Category Determination: Based on passenger multidimensional feature tags and environmental data, the demand category is determined. The demand categories include: regular demand and sudden demand. N2: Regular Demand Forecasting: Based on the multi-dimensional feature tags of passengers, regular demand is broken down into several sub-tasks, the probability of each sub-task is calculated, and a regular service demand record is generated. The regular service demand record includes: passenger ID number, sub-task list, probability, and valid time. N3: Sudden Demand Prediction: Determine the type of sudden event, combine passenger multi-dimensional feature tags, generate several sub-tasks, assign a severity level to each sub-task, and generate a sudden service demand record. The sudden service demand record includes: passenger ID number, sub-task list, severity level, event location, and effective time. N4: Service Task Generation: Simultaneously receives regular service demand records and emergency service demand records. If the two types of service demands overlap in the time window for the same passenger, the emergency service demand takes priority, and the regular service demand is postponed. Calculate the priority of each subtask, generate the initial passenger service demand, and write it into the dynamic master data matrix, marking it as a predicted demand. The service execution and monitoring steps include: Q1: Establishment of real-time passenger-service mapping space: Link passenger information and service personnel information to the dispatch plan table, and dynamically display passenger location, service personnel location and service status through digital dashboards; generate real-time service tags for each passenger and mark the service personnel status; Q2: Service Execution: Push the scheduling plan table to the service personnel. After receiving it, the service personnel update the service status to "in progress" and start the timer. Q3: Real-time monitoring of service gaps: Real-time collection of high-speed rail station operation data, passenger information, service personnel information, and environmental data to calculate time gaps, demand gaps, and personnel gaps; Q4: Dynamic evaluation and adjustment of service quality: Real-time evaluation of service timeliness, completeness and satisfaction, and root cause analysis to obtain analysis results; Based on the analysis results, dynamically adjust the initial service plan and synchronize it back to the mapping space; Q5: Service Execution Effectiveness Report Generation: After each service request is completed, the passenger status and service personnel performance are automatically updated to generate a service execution effectiveness report.

2. The method for scheduling high-speed rail business travel service personnel based on data analysis according to claim 1, characterized in that: The method for constructing the dynamic master data matrix includes: S1: Data Processing: Establish unique linear mileage coordinates and a unified time base, and map all data events to the same spatiotemporal coordinate system; define and unify primary keys for high-speed rail station operation data, passenger data, service personnel information, and environmental data, and establish a primary key mapping table; and perform streaming data processing on the data, align the data by train number-time window-spatial window, and generate unified aligned data. S2: Create an empty matrix template: Select the primary key and predefine an empty matrix based on the primary key; S3: Dynamic matrix filling: Maps each aligned data to the corresponding cell. For multiple records in the same cell, the mapping is performed by overwriting and aggregating the latest value to obtain a dynamic master data matrix. S4: Verification and Update: Real-time detection of data integrity, marking of anomalies, updating of the dynamic master data matrix, and retention of historical versions.

3. The method for scheduling high-speed rail business travel service personnel based on data analysis according to claim 1, characterized in that: The method for obtaining the multidimensional feature labels includes: M1: Feature Dimension Definition: Defines the core feature dimensions of passengers, which are divided into static and dynamic dimensions; M2: Feature Extraction: Based on the dynamic master data matrix, multi-dimensional features of passengers are extracted according to the core feature dimensions of passengers and a sliding time window is used to calculate the feature values ​​and obtain multi-dimensional features of passengers. M3: Tag Generation: Based on the feature values, aggregate all tags by passenger ID number to generate a dynamic tag table. Each row corresponds to a passenger ID number, and the column fields are the multi-dimensional feature tags of the specific passenger.

4. The method for scheduling high-speed rail business travel service personnel based on data analysis according to claim 1, characterized in that: The multidimensional dynamic matching method includes: P1: Dataset Extraction: Extract real-time and predicted demand from the dynamic master data matrix to establish a set of passenger demand to be responded to; extract the set of currently schedulable service personnel from the dynamic master data matrix; P2: Define matching dimensions and quantification rules: Define the matching dimensions between service personnel and passenger service needs, as well as the quantification rules for these matching dimensions; the matching dimensions include: spatial and temporal accessibility, skill matching degree, service value ratio, and personnel fatigue level; P3: Candidate Matching: Based on hard filtering rules, the initial screening calculation yields a candidate matching set of demand and personnel; P4: Multidimensional dynamic matching: For each demand-person candidate matching set, calculate the matching dimension index, normalize it, and obtain the comprehensive score and initial service plan; P5: Service Plan Generation and Output: Based on the initial service plan, generate a structured scheduling plan table. Each row includes: service requirements, service personnel, estimated arrival time, skill matching degree, service value score, and fatigue risk level.

5. The method for scheduling high-speed rail business travel service personnel based on data analysis according to claim 1, characterized in that: The method for constructing the real-time passenger-service mapping space includes: R1: Data preparation: Import the scheduling plan table, passenger information and service personnel information into the passenger-service real-time mapping database, and unify the passenger ID number and service personnel number as the primary key; R2: Passenger Layer Construction: Each passenger record generates a dynamic object with fields including: ID number, train number, schedule node, schedule time, real-time location, and service tag; R3: Service Personnel Layer Construction: Each service personnel record generates a dynamic object with fields including: employee ID, skills, real-time location, task list, and service status; R4: Node layer construction: Convert all service nodes of the high-speed rail station into spatial points, with capacity thresholds, current number of people and queuing time; R5: Coordinate Access: Real-time access to the real-time location of passengers and service personnel.

6. The method for scheduling high-speed rail business travel service personnel based on data analysis according to claim 1, characterized in that: The feedback optimization steps include: E1: Collection and Cleaning: Service performance reports are collected into the feedback pool, and unstructured feedback is converted into a unified score to generate a passenger-event-score-evaluation record; E2: Record Analysis: Performs attribution analysis on passenger-event-score-evaluation records, automatically labels root causes, and generates root cause reports; E3: Optimization and Update: Based on the root cause report, update and optimize the service personnel scheduling process.

Images