Ferry transport capacity resource allocation method, system, device and medium based on multi-source data

Abstract

The application provides a ferry transport capacity resource allocation method, system, device and medium based on multi-source data, which comprises the following steps: acquiring ferry historical reservation data; dividing the historical reservation data into multiple basic adjustment periods, and constructing a space-time demand portrait; collecting external behavior factors; processing the external behavior factors according to the space-time demand portrait to obtain user travel sensitivity weights corresponding to the basic adjustment periods; according to the historical reservation data, obtaining the total value of the historical bookable ticket number and the historical successful booking ticket number value corresponding to each basic adjustment period, and calculating the historical booking success rate to obtain a booking success rate weight; according to the user travel sensitivity weight, the booking success rate weight, and the preset demand adjustment coefficient and the supply-demand adjustment coefficient, the first-stage transport capacity adjustment coefficient of the corresponding basic adjustment period is determined by calculation; and the guide ticket price of each basic adjustment period of the ferry is determined according to the first-stage transport capacity adjustment coefficient.

Description

Technical Field

[0001] This application relates to the field of resource allocation technology, and in particular to a method, system, equipment and medium for allocating ferry capacity resources based on multi-source data. Background Technology

[0002] With the rapid development of waterway transportation and island tourism in my country, ferries, as a core mode of transportation connecting the mainland and islands and traversing rivers and waterways, have experienced explosive growth in passenger and freight volume. Due to the obvious cyclical and seasonal characteristics of ferry operations, fixed transport capacity resources such as ships and manpower face extremely uneven demand shocks in different seasons, dates, and even different times of the day. How to utilize modern information technology to scientifically and rationally allocate and guide ferry transport capacity resources, achieving peak-hour diversion and off-peak-hour demand stimulation, thereby improving the overall resource utilization rate, has become a key issue urgently needing to be addressed in the intelligent construction of waterway transportation.

[0003] In related technologies, ferry capacity allocation and pricing strategies typically employ fixed patterns or rely solely on single, localized historical ferry booking data for fine-tuning. This approach severely limits dynamic perception capabilities, making it difficult to adapt in real-time to complex external behavioral factors such as sudden weather changes, large-scale surrounding events, and instantaneous holiday traffic. Summary of the Invention

[0004] In view of this, it is necessary to provide a method, system, equipment and medium for allocating ferry capacity resources based on multi-source data, which can at least overcome one of the above-mentioned defects.

[0005] In a first aspect, embodiments of this application provide a method for allocating ferry capacity resources based on multi-source data, the method comprising:

[0006] Obtain historical ferry reservation data;

[0007] The historical reservation data is subjected to pattern recognition to divide it into multiple basic adjustment periods, and a spatiotemporal demand profile is constructed.

[0008] Collect external behavioral factors;

[0009] The external behavioral factors are processed based on the spatiotemporal demand profile to obtain the user travel sensitivity weights corresponding to the basic adjustment period.

[0010] Based on the historical reservation data, obtain the total number of historically bookable tickets and the number of historically successfully booked tickets for each of the basic adjustment periods;

[0011] The historical booking success rate is calculated based on the total number of historically available tickets and the number of historically successfully booked tickets, so as to obtain the booking success rate weight that characterizes the historical supply and demand of the spatiotemporal demand profile.

[0012] The first-stage capacity adjustment coefficient corresponding to the basic adjustment period is determined by calculating the user travel sensitivity weight, the booking success rate weight, and the preset demand adjustment coefficient and supply-demand adjustment coefficient.

[0013] The guiding fare for each basic adjustment period of the ferry is determined based on the capacity adjustment coefficient of the first stage.

[0014] In one embodiment, processing the external behavioral factors based on the spatiotemporal demand profile to obtain the user travel sensitivity weights corresponding to the basic adjustment period includes:

[0015] Generate corresponding time period labels based on each of the basic adjustment time periods, and obtain the first sensitivity weight based on the time period label mapping;

[0016] Obtain the external behavior feature indicators corresponding to each of the basic adjustment periods, and generate corresponding multi-path dynamic environment sensitivity weights based on the external behavior feature indicators.

[0017] The first sensitivity weight and each of the multi-path dynamic environment sensitivity weights are normalized and weighted to obtain the user travel sensitivity weight corresponding to the basic adjustment period.

[0018] In one embodiment, the external behavioral characteristic indicators include weather and meteorological data corresponding to each of the basic adjustment periods, activity information data for future periods, and calendar information data;

[0019] The process of generating corresponding multi-path dynamic environment sensitivity weights based on the external behavioral feature indicators includes:

[0020] Climate labels are generated based on the weather and meteorological data, and a second sensitivity weight is obtained by mapping the climate labels.

[0021] Activity tags are generated based on the activity information data, and a third sensitivity weight is obtained by mapping the activity tags.

[0022] Holiday tags are generated based on the calendar information data, and a fourth sensitivity weight is obtained based on the mapping of the holiday tags.

[0023] In one embodiment, the climate label, the activity label, and the holiday label are respectively weighted according to the degree of travel obstruction, the scale of the activity, and the duration of the holiday;

[0024] The higher the degree of travel obstruction for a climate label, the greater the corresponding second sensitivity weight; the larger the scale of the activity and the higher the travel necessity for the activity label, the smaller the corresponding third sensitivity weight; and the longer the holiday duration and the greater the travel aggregation effect for a holiday label, the smaller the corresponding fourth sensitivity weight.

[0025] In one embodiment, the method further includes:

[0026] Set time thresholds based on the scheduled departure times of ferry flights;

[0027] When the countdown exceeds the time threshold, real-time traffic data of the area near the ferry port is acquired, and the real-time traffic data is analyzed to obtain the current traffic congestion situation.

[0028] Based on the traffic congestion situation, determine the traffic congestion situation label and the corresponding traffic congestion situation weight;

[0029] The guiding fare is adjusted a second time based on the weight of the traffic congestion situation to obtain the second-stage guiding fare.

[0030] In one embodiment, the step of acquiring real-time traffic data of the area near the ferry port and analyzing the real-time traffic data to obtain the current traffic congestion situation includes:

[0031] Real-time surveillance video data of the area near the ferry port is acquired, and the real-time surveillance video data is processed by frame segmentation to obtain real-time surveillance image data.

[0032] The real-time monitoring image data is segmented using an image segmentation algorithm to obtain multiple image blocks;

[0033] The image patches are classified according to the decision tree classification algorithm and based on the multivariate features of each image patch to determine the category of each image patch;

[0034] The region growing algorithm is used to merge image patches of the same category to obtain road regions and vehicle regions under the same monitoring image;

[0035] The traffic congestion situation is obtained by jointly analyzing the spatial and state relationships of the road area and the vehicle area.

[0036] In one embodiment, the step of jointly analyzing the spatial and state relationships between the road area and the vehicle area to obtain the traffic congestion situation includes:

[0037] The Hough transform algorithm is used in the road area to identify and obtain lane line position information and traffic sign information;

[0038] Instantaneous traffic flow is counted in the vehicle area, and the real-time speed of the vehicle is calculated based on the displacement and time difference of the vehicle between consecutive frames.

[0039] Using the K-Means algorithm, multi-dimensional feature clustering analysis is performed on the lane line position information, the traffic sign information, the traffic flow, and the vehicle speed to form multiple data clusters;

[0040] Based on the average feature value within each data cluster, each data cluster is divided and mapped to different congestion levels to obtain the traffic congestion situation.

[0041] Secondly, embodiments of this application provide a ferry capacity resource allocation system based on multi-source data, applied to implement the method described in the first aspect, the system comprising:

[0042] The information acquisition module is used to acquire historical ferry reservation data and collect external behavioral factors;

[0043] The profile generation module is used to perform pattern recognition on the historical reservation data to divide it into multiple basic adjustment periods and construct a spatiotemporal demand profile.

[0044] The resource allocation module is used to process the external behavioral factors according to the spatiotemporal demand profile to obtain the user travel sensitivity weight corresponding to the basic adjustment period; based on the historical reservation data, it obtains the total number of historically bookable tickets and the number of historically successfully booked tickets for each of the basic adjustment periods; it calculates the historical booking success rate based on the total number of historically bookable tickets and the number of historically successfully booked tickets to obtain the booking success rate weight that represents the historical supply and demand of the spatiotemporal demand profile; and it calculates and determines the first-stage capacity adjustment coefficient corresponding to the basic adjustment period based on the user travel sensitivity weight, the booking success rate weight, and preset demand adjustment coefficients and supply and demand adjustment coefficients.

[0045] The fare generation module is used to determine the guiding fare for each of the basic adjustment periods of the ferry based on the capacity adjustment coefficient of the first stage.

[0046] Thirdly, embodiments of this application provide an electronic device, including:

[0047] Processor; and

[0048] A memory having computer-readable instructions stored thereon for controlling the processor to perform the method as described in the first aspect.

[0049] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the method described in the first aspect at runtime.

[0050] This application provides a method, system, equipment, and medium for allocating ferry capacity resources based on multi-source data. By acquiring long-term historical ferry reservation data and performing pattern recognition, it scientifically divides multiple basic adjustment periods, including peak, off-peak, and low-peak periods, and constructs a precise spatiotemporal demand profile. Based on this, the system not only collaboratively processes dynamically changing external behavioral factors such as weather and events to dynamically assess the rigidity of user travel needs (sensitivity weight), but also deeply mines the booking success rate weight, which characterizes historical supply and demand, by combining historical available and successful booking numbers. By performing multi-dimensional linear calculations using the weights of multi-source internal and external data and preset coefficients, it accurately determines the first-stage capacity adjustment coefficient for the corresponding period and converts it into differentiated guiding fares. This method effectively overcomes the limitations of traditional allocation methods that rely solely on fixed fares or single local reservation data, leading to uneven distribution of capacity resources and low overall utilization. In a normal, long-term dimension, it successfully achieves spatiotemporal diversion through price leverage, effectively diverting congested demand during peak periods and fully stimulating the utilization of idle resources during off-peak periods, thus achieving optimized allocation and efficient utilization of ferry baseline capacity resources. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating a ferry capacity resource allocation method based on multi-source data provided in an embodiment of this application.

[0052] Figure 2 This is a schematic diagram of the modules of a ferry capacity resource allocation method provided in an embodiment of this application.

[0053] Figure 3 This is a schematic diagram of the modules of an electronic device provided in an embodiment of this application. Detailed Implementation

[0054] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0055] It should be noted that, in the embodiments of this application, "at least one" refers to one or more, and "more than one" refers to two or more. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the specification of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application.

[0056] It should be noted that in the embodiments of this application, the terms "first," "second," etc., are used only for descriptive purposes and should not be construed as indicating or implying relative importance, nor as indicating or implying order. Features specified as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0057] Based on the embodiments described in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0058] With the rapid development of waterway transportation and island tourism in my country, ferries, as a core mode of transportation connecting the mainland and islands and traversing rivers and waterways, have experienced explosive growth in passenger and freight volume. Due to the obvious cyclical and seasonal characteristics of ferry operations, fixed transport capacity resources such as ships and manpower face extremely uneven demand shocks in different seasons, dates, and even different times of the day. How to utilize modern information technology to scientifically and rationally allocate and guide ferry transport capacity resources, achieving peak-hour diversion and off-peak-hour demand stimulation, thereby improving the overall resource utilization rate, has become a key issue urgently needing to be addressed in the intelligent construction of waterway transportation.

[0059] In related technologies, ferry capacity allocation and pricing strategies typically employ fixed patterns or rely solely on single, localized historical ferry booking data for fine-tuning. This approach severely limits dynamic perception capabilities, making it difficult to adapt in real-time to complex external behavioral factors such as sudden weather changes, large-scale surrounding events, and instantaneous holiday traffic.

[0060] In view of this, the ferry capacity resource allocation method based on multi-source data provided in this application scientifically divides multiple basic adjustment periods, including peak, off-peak, and low-peak periods, by acquiring long-term historical ferry reservation data for pattern recognition, and constructs a precise spatiotemporal demand profile. Based on this, the system not only collaboratively processes dynamically changing external behavioral factors such as weather and events to dynamically assess the rigidity of user travel needs (sensitivity weight), but also deeply mines the reservation success rate weight, which characterizes historical supply and demand, by combining historical available and successful ticket numbers. By performing multi-dimensional linear calculations with the weights of multi-source internal and external data and preset coefficients, the first-stage capacity adjustment coefficient for the corresponding period is accurately determined and converted into differentiated guiding fares. This method effectively overcomes the limitations of traditional allocation methods that rely solely on fixed fares or single local reservation data, leading to uneven distribution of capacity resources and low overall utilization. In the long-term, normal timeframe, it successfully achieves spatiotemporal diversion through price leverage, effectively diverting congested demand during peak periods and fully stimulating the utilization of idle resources during off-peak periods, thus achieving optimized allocation and efficient utilization of ferry baseline capacity resources.

[0061] This application constructs a two-stage, closed-loop allocation system that coordinates "macro-level long-term guidance" and "micro-level near-term peak shaving." The first stage integrates historical reservation supply and demand saturation with multi-source external environmental indicators at a normalized level to generate a reasonable basic guiding fare for spatiotemporal diversion. The second stage directly addresses the pain point of traditional solutions' inability to dynamically respond to sudden changes in traffic conditions. At critical time points approaching ferry departures, it creatively introduces digital spatial reconstruction technology from port perimeter surveillance videos. Through image segmentation and decision tree classification, road and vehicle areas are accurately separated, and the K-Means algorithm is used to perform multi-dimensional feature clustering of lane lines, traffic signs, instantaneous traffic flow, and vehicle speed, thereby intelligently and automatically identifying real-time traffic congestion around the port. By injecting real-time congestion weights into the system to perform secondary dynamic fare adjustments, in the event of severe congestion, the price lever is used to decisively suppress excessive instantaneous demand, forcibly diverting nearby travel vehicles, fundamentally alleviating traffic paralysis on roads around the port, and ensuring that the ferry capacity system as a whole possesses extremely high risk resistance and intelligent collaborative allocation efficiency.

[0062] Figure 1 This is a flowchart illustrating a ferry capacity resource allocation method based on multi-source data, provided in an embodiment of this application. Figure 1The ferry capacity resource allocation method based on multi-source data, as shown, includes at least the following steps: S100: Obtain historical ferry reservation data; S200: Perform pattern recognition on the historical reservation data to divide it into multiple basic adjustment periods and construct a spatiotemporal demand profile; S300: Collect external behavioral factors; S400: Process the external behavioral factors according to the spatiotemporal demand profile to obtain the user travel sensitivity weight for the corresponding basic adjustment period; S500: Obtain the total number of historically bookable tickets and the number of historically successfully booked tickets for each basic adjustment period based on the historical reservation data; S600: Calculate the historical booking success rate based on the total number of historically bookable tickets and the number of historically successfully booked tickets to obtain the booking success rate weight, which characterizes the historical supply and demand of the spatiotemporal demand profile; S700: Calculate and determine the first-stage capacity adjustment coefficient for the corresponding basic adjustment period based on the user travel sensitivity weight, the booking success rate weight, and preset demand adjustment coefficients and supply and demand adjustment coefficients; S800: Determine the guiding ticket price for each basic adjustment period of the ferry based on the first-stage capacity adjustment coefficient.

[0063] S100: Obtain historical ferry reservation data.

[0064] In this embodiment of the application, the ferry capacity resource allocation method based on multi-source data includes obtaining historical ferry reservation data in step S100.

[0065] Specifically, the historical reservation data includes, but is not limited to, historical travel dates, flight departure times, user order times, number of tickets booked, location of the ticketing account, and final order payment status. To ensure the accuracy of subsequent pattern recognition, abnormal order data generated by system failures or malicious ticket scalping can be removed. The extracted reservation data can be denoised and standardized for alignment to form a continuous and structured historical reservation time series.

[0066] Understandably, acquiring long-term historical ferry reservation data is crucial for uncovering patterns hidden behind massive amounts of travel behavior at a macro level. Traditional allocation methods, lacking in-depth analysis of long-term historical reservation patterns, cannot accurately grasp the cyclical evolution of passenger flow. The data accumulation achieved through this step provides solid data support for subsequently accurately characterizing user travel characteristics at different temporal and spatial points.

[0067] S200: Perform pattern recognition on historical reservation data to divide it into multiple basic adjustment periods and construct a spatiotemporal demand profile.

[0068] In this embodiment of the application, the ferry capacity resource allocation method based on multi-source data includes, in step S200, performing pattern recognition on historical reservation data to divide it into multiple basic adjustment periods and constructing a spatiotemporal demand profile.

[0069] Specifically, the pattern recognition process can automatically divide a 24-hour day or an operating cycle into multiple basic adjustment periods, including peak, off-peak, and low-peak periods, based on the peak and trough distribution characteristics of historical reservation volumes. For example, the period from 10:00 AM to 4:00 PM is identified as a peak period, the period from late night to early morning with extremely low reservation volumes is identified as a low-peak period, and the rest is considered an off-peak period. At the same time, spatial flow characteristics are further superimposed by date attributes (such as weekdays and weekends) to ultimately construct a multi-dimensional spatiotemporal demand profile.

[0070] It is understandable that the spatiotemporal demand profile is not a fixed, rigid division, but a dynamic classification of macro-market demand through pattern recognition technology. Dividing the market into multiple "basic adjustment periods" can break the limitations of traditional fixed ticket prices and one-size-fits-all capacity scheduling, allowing subsequent resource allocation and price levers to be precisely focused on different supply and demand scenarios, providing a benchmark spatiotemporal reference for differentiated allocation.

[0071] S300: Collect external behavioral factors.

[0072] In this embodiment of the application, the ferry capacity resource allocation method based on multi-source data includes collecting external behavioral factors in step S300.

[0073] Specifically, the external behavioral factors include, but are not limited to, weather and meteorological data of the area where the ferry port is located (such as temperature, wind speed, precipitation, visibility, etc. in the future), future activity information data of the port's radiation economic circle or the destination on the opposite bank (such as the holding time and expected scale of concerts, sports events, large-scale exhibitions), and calendar information data corresponding to the current date (such as whether it is a statutory holiday, winter or summer vacation, or a long weekend).

[0074] Understandably, ferries, as a special mode of cross-water transportation, have extremely high and sensitive external dependence on capacity demand. Traditional dispatching methods only focus on historical sea-crossing data, lacking the ability to detect sudden or periodic external disturbances. This step, by introducing external behavioral factors such as weather, events, and holidays from multiple sources, can effectively capture temporary and sudden demand fluctuations caused by changes in the external environment, avoiding lag in the dispatching system.

[0075] S400: Process external behavioral factors based on spatiotemporal demand profiles to obtain the user travel sensitivity weights for the corresponding basic adjustment period.

[0076] In this embodiment of the application, the ferry capacity resource allocation method based on multi-source data includes, in step S400, processing external behavioral factors according to the spatiotemporal demand profile to obtain the user travel sensitivity weight for the corresponding basic adjustment period.

[0077] Specifically, the system first generates a time period label based on the current adjustment period and matches it with a first sensitivity weight; then, it converts weather data into climate labels (such as typhoon, rainstorm, sunny, etc.) using a corresponding recognition model and matches them with a second sensitivity weight; it converts activity data into activity labels (such as large-scale events, small exhibitions) and matches them with a third sensitivity weight; it converts calendar data into holiday labels and matches them with a fourth sensitivity weight; finally, it normalizes the sensitivity weights corresponding to the above multiple labels and performs a weighted summation to calculate the final user travel sensitivity weight.

[0078] Understandably, the user travel sensitivity weight reflects users' acceptance and tolerance of fare adjustments under specific time periods and external environments. When faced with scenarios with strong travel demand, such as large-scale events or long holidays, users are less sensitive to price changes, resulting in a lower sensitivity weight. In this case, the system has more room to raise prices to implement traffic diversion. Conversely, during off-peak periods or recovery periods from severe weather, users have low travel intentions and high price sensitivity, resulting in a higher sensitivity weight. The system then needs to stimulate domestic demand by lowering the barriers to entry, thereby using price levers to achieve flexible allocation.

[0079] S500: Based on historical reservation data, obtain the total number of historically available tickets and the number of historically successfully booked tickets for each basic adjustment period.

[0080] In this embodiment of the application, the ferry capacity resource allocation method based on multi-source data includes, in step S500, obtaining the total number of historically bookable tickets and the number of historically successfully booked tickets corresponding to each basic adjustment period based on historical reservation data.

[0081] Specifically, for each specific basic adjustment period (such as the peak period of a weekend in a specific season) divided in step S200, the system backtracks and retrieves data of all flights that have been opened in the same historical period. It then adds up the total number of passenger / vehicle seats that were allowed to be booked by the outside world when the system was first opened, to obtain the total number of historically bookable tickets. At the same time, it adds up the actual number of tickets that were eventually successfully booked and paid for by users and completed the trip, to obtain the number of historically successfully booked tickets.

[0082] Understandably, the total number of historically available tickets obtained in this step represents the maximum capacity that the ferry system can provide during that period, while the number of historically successful ticket bookings represents the actual resource demand being met in the market. Extracting these two absolute values ​​is a crucial data bridge for quantitatively assessing whether this baseline adjustment period is frequently characterized by resource idleness or extreme resource strain under historical normal conditions.

[0083] S600: Calculate the historical booking success rate based on the total number of historically available tickets and the number of historically successfully booked tickets to obtain the booking success rate weight that represents the historical supply and demand of the spatiotemporal demand profile.

[0084] In this embodiment of the application, the ferry capacity resource allocation method based on multi-source data includes, in step S600, calculating the historical booking success rate based on the total number of historically bookable tickets and the number of historically successfully booked tickets, so as to obtain the booking success rate weight that characterizes the historical supply and demand of the spatiotemporal demand profile.

[0085] Specifically, the system uses the historical number of successfully booked tickets within a specific basic adjustment period as the numerator and the total number of historically available tickets as the denominator for division, and then performs a division operation to obtain the overall historical booking success rate for that period. If this success rate is close to 100%, it indicates that the period has historically been in a state of supply shortage and extremely tight capacity; conversely, if the success rate is low, it indicates that capacity resources are idle. Subsequently, the system uses a mapping function to convert this booking success rate into corresponding booking success rate weights based on its magnitude.

[0086] Understandably, the booking success rate weight is a quantitative indicator that depicts the long-term supply and demand imbalance within the spatiotemporal demand profile. Introducing this weight allows the system to fully absorb and inherit the experience of supply and demand saturation under similar past scenarios when adjusting current ticket prices, making the baseline allocation of resources more in line with the real response of the long-term market and avoiding the phenomenon of simply relying on external forecasts and deviating from the port's own historical carrying capacity.

[0087] S700: The first-stage capacity adjustment coefficient for the corresponding basic adjustment period is determined by calculating the user's travel sensitivity weight, booking success rate weight, and preset demand adjustment coefficient and supply-demand adjustment coefficient.

[0088] In this embodiment of the application, the ferry capacity resource allocation method based on multi-source data includes, in step S700, calculating and determining the first-stage capacity adjustment coefficient for the corresponding basic adjustment period based on the user travel sensitivity weight, the booking success rate weight, and the preset demand adjustment coefficient and supply-demand adjustment coefficient.

[0089] Specifically, the system uses the user travel sensitivity weight obtained in step S400 and the booking success rate weight obtained in step S600 as independent variables. It also introduces demand adjustment coefficients and supply-demand adjustment coefficients pre-set by the system backend based on long-term operational goals. These weights and coefficients are then substituted into preset linear or nonlinear combination formulas for matrix multiplication and weighted superposition calculations, thereby customizing and outputting a first-stage capacity adjustment coefficient for each basic adjustment period.

[0090] Understandably, the first-stage capacity adjustment coefficient is the culmination of this scheme's core technology, achieved through the integration of multi-source data at a macro level. It not only comprehensively considers internal supply and demand factors such as historical ticket purchase success rates, but also takes into account the dynamic impact of external factors like temporary weather, large-scale events, and holidays on ticket purchase sensitivity. It distills the complex internal and external game relationships into a unified standard allocation coefficient, providing direct numerical support for subsequent refined pricing.

[0091] S800: Determine the guiding ticket price for each basic adjustment period of the ferry based on the capacity adjustment coefficient of the first stage.

[0092] In this embodiment of the application, the ferry capacity resource allocation method based on multi-source data includes, in step S800, determining the guiding ticket price for each basic adjustment period of the ferry based on the first-stage capacity adjustment coefficient.

[0093] Specifically, the system retrieves the preset initial ticket price from the backend, multiplies this initial ticket price by the first-stage capacity adjustment coefficient determined in step S700, or performs a preset function transformation to calculate and output the first ferry guidance ticket price effective for each basic adjustment period. The system updates the ticketing frontend with the first ferry guidance ticket price as a supply and demand lever. During peak hours, the system raises the entry threshold through the adjustment coefficient to guide passenger flow to off-peak or low-peak periods on both sides. During low-peak periods, the system lowers the entry threshold through the adjustment coefficient to attract potential travel traffic.

[0094] It is understandable that determining the guiding fare is the final technical loop for this application to achieve the spatiotemporal optimization and allocation of ferry benchmark capacity resources on a macro-long-term perspective. This application transforms the complex technical problem of "capacity resource allocation" into a quantitative solution of "multi-source data adjustment coefficients and corresponding guiding fares," using dynamically adjustable fares as a spatiotemporal diversion adjustment valve. Without blindly increasing or decreasing the input of ships and human resources, it achieves flexible peak-shifting control of cross-sea passenger and vehicle flows, thereby technically solving the industry pain points of capacity overload and congestion during peak periods and idle and wasted capacity resources during off-peak periods.

[0095] In this embodiment, external behavioral factors are processed based on spatiotemporal demand profiles to obtain user travel sensitivity weights for corresponding basic adjustment periods. This includes: generating corresponding time period labels for each basic adjustment period and mapping a first sensitivity weight based on the time period labels; obtaining external behavioral characteristic indicators for each basic adjustment period and mapping corresponding multi-path dynamic environment sensitivity weights based on these indicators; and performing a normalized weighted sum of the first sensitivity weight and each multi-path dynamic environment sensitivity weight to obtain the user travel sensitivity weight for the corresponding basic adjustment period.

[0096] Specifically, the system first assigns a corresponding time period label to the current basic adjustment period based on the peak, off-peak, or low-peak period attributes defined in step S200, and then matches the corresponding first sensitivity weight using a preset discrete mapping table. Based on this, the system extracts multi-dimensional feature data from the port's external environment mapped to this adjustment period as external behavioral feature indicators through multiple parallel data acquisition channels. Each indicator is processed by its respective feature mapping model, decoupled, and independently outputs its corresponding multi-path dynamic environment sensitivity weights. Finally, the system uses a normalization operator to perform dimensionless processing on the first sensitivity weight and the multi-path dynamic environment sensitivity weights, and then retrieves a preset global weight matrix to perform a weighted summation, outputting a user travel sensitivity weight that comprehensively represents the intertwined influence of the current time period and the environment.

[0097] Understandably, travel sensitivity weighting is a digital measure of a user's essential need for travel. This step, through a combination of "time-period label basic mapping" and "multi-environment dynamic correction," can finely depict the psychological tolerance boundaries of users for adjustment coefficients and guiding fares at different times and in different environments. Decomposing a single external factor into multiple environmental weights and then weighting and summing them can prevent the adjustment coefficients from being distorted due to drastic fluctuations in a single environmental indicator, greatly improving the accuracy of the sensitivity weight output.

[0098] In this embodiment, the external behavioral characteristic indicators include weather and meteorological data corresponding to each basic adjustment period, activity information data for future periods, and calendar information data. Based on these external behavioral characteristic indicators, corresponding multi-path dynamic environmental sensitivity weights are generated, including: generating climate labels based on weather and meteorological data, and obtaining a second sensitivity weight based on the climate labels; generating activity labels based on activity information data, and obtaining a third sensitivity weight based on the activity labels; and generating holiday labels based on calendar information data, and obtaining a fourth sensitivity weight based on the holiday labels.

[0099] Specifically, the system monitors and captures multi-source external data in real time through cross-platform data interfaces: For weather and meteorological data, it uses a meteorological recognition model to extract textual and numerical features, outputting climate labels such as typhoon, rainstorm, fog, and sunny, which are then mapped to obtain the second sensitivity weight; For event information data for future periods, it extracts key dimensions such as the event time, location, and expected number of participants, inputs them into the event recognition model to generate corresponding event labels (such as concerts, international competitions, and small exhibitions), which are then mapped to obtain the third sensitivity weight; For calendar information data, it generates corresponding holiday labels (such as Spring Festival, National Day, and non-holidays) by matching the standard legal calendar, which are then mapped to obtain the fourth sensitivity weight.

[0100] Understandably, the multi-dimensional refinement and labeling of external behavioral characteristic indicators are core steps in transforming unstructured external environmental factors into computer-recognizable technical parameters. Weather, events, and calendars intertwine from different dimensions to determine the external triggers for ferry travel. By establishing climate labels, event labels, and holiday labels, and independently mapping them to second, third, and fourth sensitivity weights, the system can establish a clear technical logic chain, accurately identifying and quantifying the local impact of different external behaviors on the ferry capacity system.

[0101] In this embodiment, climate tags, activity tags, and holiday tags are weighted according to their respective degrees of travel disruption, activity scale, and holiday duration. Specifically, climate tags with higher degrees of travel disruption have a greater second sensitivity weight; activity tags with larger activity scales and higher travel inelasticity have a smaller third sensitivity weight; and holiday tags with longer holiday durations and travel aggregation effects have a smaller fourth sensitivity weight.

[0102] Specifically, in the design of the weight mapping rules, the parameter guidance rules are as follows: For climate labels, labels with high travel obstruction (such as typhoons, rainstorms, etc.) are assigned larger values ​​for their corresponding second sensitivity weights, while labels with low travel obstruction (such as sunny days) are assigned smaller values ​​for their corresponding weights; For event labels, labels with large event scale and high travel demand (such as concerts, etc.) are assigned smaller values ​​for their corresponding third sensitivity weights; For holiday labels, labels with long holiday duration and travel aggregation effects (such as National Day, Spring Festival, etc.) are assigned smaller values ​​for their corresponding fourth sensitivity weights.

[0103] Understandably, the aforementioned weighting mapping rules follow a rigorous reverse smoothing technical logic: the magnitude of the sensitivity weight is positively correlated with users' sensitivity to ticket prices, and negatively correlated with the subsequent adjustment range of the system's entry threshold. When facing large-scale, long-holiday, and highly essential travel scenarios (such as long holidays or major events), users have a very high willingness to travel and a high acceptance of price changes. In this case, the smaller the mapped sensitivity weight, the more conducive it is to move the capacity adjustment coefficient to a higher level in the first stage, thereby technically creating a higher entry threshold to implement peak-shifting and diversion of large traffic volumes. Conversely, when the travel barrier caused by weather labels is extremely high, it indicates that travel is not driven by subjective essential needs, and users are extremely sensitive to prices. The larger the mapped sensitivity weight, the more it can limit the upward movement of the adjustment coefficient, thereby quickly dispersing the stranded traffic with a low threshold after the severe weather ends, achieving refined allocation of ferry capacity resources.

[0104] In this embodiment, the method further includes: setting a time threshold based on the scheduled departure time of the ferry flight. When the countdown exceeds the time threshold, real-time traffic data of the area near the ferry port is acquired, and the real-time traffic data is analyzed to obtain the current traffic congestion situation. A traffic congestion label and corresponding traffic congestion weight are determined based on the traffic congestion situation. The guiding fare is then adjusted a second time based on the traffic congestion weight to obtain the second-stage guiding fare.

[0105] Specifically, the system presets a time threshold (e.g., 5 hours before departure) for each ferry trip based on the ferry schedule. Once the countdown to departure exceeds this time threshold, the system switches from normal long-cycle adjustment to a special dynamic adjustment period closer to departure. It then activates data interfaces to access real-time traffic data from sensors, cameras, or road network monitoring in the port's periphery for online digital processing. The analyzed congestion status is assigned corresponding traffic congestion labels (e.g., high congestion, medium congestion, low congestion), and a built-in weight matrix is ​​used to assign corresponding traffic congestion weights based on these labels. Finally, the system uses these traffic congestion weights as correction parameters, adding them to the previously determined normalized guiding fare. Through mathematical calculations, it performs a second-stage fare correction, outputting a second-stage guiding fare for vehicles with upcoming travel needs.

[0106] Understandably, setting time thresholds and dividing the allocation process into two phases is intended to supplement long-term static forecasting with a short-term, instantaneous micro-adjustment mechanism. The first phase of fare guidance primarily addresses the issue of supply and demand misalignment based on historical reservations and the macro environment, while the second phase focuses on resolving potential localized vehicle congestion and delays around the port as departure approaches. By introducing real-time traffic conditions as a secondary adjustment variable for pricing as departure approaches, the system can decisively raise the entry threshold when the surrounding road network is paralyzed, forcibly suppressing and delaying the sudden increase in non-pre-purchased vehicles, thus establishing a closed-loop mechanism for instantaneous peak shaving to cope with sudden traffic pressure from a technical perspective.

[0107] In this embodiment, real-time traffic data of the area near the ferry port is acquired and analyzed to obtain the current traffic congestion situation. This includes: acquiring real-time surveillance video data of the area near the ferry port and performing frame-by-frame processing on the real-time surveillance video data to obtain real-time surveillance image data; using an image segmentation algorithm to segment the real-time surveillance image data to obtain multiple image blocks; classifying the image blocks according to a decision tree classification algorithm and based on the multivariate features of each image block to determine the category of each image block; merging image blocks of the same category using a region growing algorithm to obtain road areas and vehicle areas under the same surveillance image; and performing joint analysis based on the spatial and state relationships of the road areas and vehicle areas to obtain the traffic congestion situation.

[0108] Specifically, the system acquires high-definition video stream data from surveillance cameras at key intersections around the port and uses a video decoding operator to break it down into static surveillance image data arranged in a time series according to a preset time step. Then, a deep image segmentation network is used to segment the entire image into multiple non-overlapping image blocks composed of local pixels. The system extracts multi-dimensional feature vectors for each image block in parallel, including texture complexity, RGB color mean, perimeter-to-area ratio, and average grayscale value, and inputs them into a pre-trained decision tree classifier to identify and classify each image block as belonging to the physical category of "road" or "vehicle." Next, using image blocks of the same category as seed nodes, a region growing algorithm is used to expand and merge spatially connected regions, accurately delineating large continuous road areas and independent vehicle target areas in the surveillance image. Finally, the current traffic congestion status is extracted based on the overlap and dynamic changes of roads and vehicles.

[0109] Understandably, this step introduces AI-based image processing technology to transform unstructured surveillance video into a high-precision road network spatial topology. Through a rigorous computer vision chain of "frame segmentation-segmentation-decision tree classification-region growth and merging," the system can eliminate interference from background noise, shadows, and weather reflections on image recognition. This significantly lowers the technical barrier to digital reconstruction of port road conditions without relying on high-cost roadside sensors, and lays a reliable physical representation foundation for the subsequent fully automated extraction of multi-dimensional traffic flow core features and congestion level determination.

[0110] In this embodiment, traffic congestion is determined through joint analysis of the spatial and state relationships between road and vehicle areas. This includes: identifying and acquiring lane position information and traffic sign information in the road area using the Hough transform algorithm; calculating instantaneous traffic flow in the vehicle area and determining real-time vehicle speeds based on displacement and time differences between consecutive frames; and performing multi-dimensional feature clustering analysis on lane position information, traffic sign information, traffic flow, and vehicle speed using the K-Means algorithm to form multiple data clusters. Based on the average feature values ​​within each data cluster, each data cluster is mapped to different congestion levels to obtain the traffic congestion situation.

[0111] Specifically, the system performs parallel mining of spatial and dynamic features: within the identified road area, by setting distance and angle parameters in the Hough space, the Hough transform algorithm is applied to filter short interference lines to accurately extract the position of long lane lines, and color and shape thresholds are used to extract traffic sign information such as speed limits; within the identified vehicle area, a sliding window counting mechanism is used to count the instantaneous traffic flow within the monitoring area, and target tracking technology is used to lock the pixel displacement of the same vehicle in consecutive image frames, and the actual average speed of the vehicle is calculated by combining the frame time difference. Subsequently, the four-dimensional indicators of lane line position, traffic signs, traffic flow, and vehicle speed are concatenated and used as a unified feature vector input into the K-Means clustering model for multiple iterations of optimization, resulting in multiple data clusters representing different traffic congestion states after convergence. Finally, by calculating the average value of the feature parameters within each data cluster, each data cluster is strictly mapped to discrete congestion levels such as high congestion, medium congestion, and low congestion.

[0112] Understandably, this step integrates static geometric features (lane lines, signs) with dynamic physical flow features (traffic volume, vehicle speed) to construct a multi-dimensional comprehensive clustering space. Single features (such as only considering traffic volume or only considering vehicle speed) are prone to failure in complex port queuing scenarios. This step utilizes the K-Means algorithm for multi-dimensional unsupervised clustering, which can automatically and accurately capture severe congestion patterns such as "restricted lane lines + extremely high traffic volume + extremely low vehicle speed." By calculating the cluster mean and stably mapping it to the corresponding congestion level, the final output traffic congestion label ensures extremely high decision-making reference value, making the subsequent secondary capacity allocation strategy more scientific and accurate.

[0113] The complete workflow of the ferry capacity resource allocation method based on multi-source data provided in this application is described below with an exemplary embodiment.

[0114] This example uses the operational scheduling of a ferry port across a strait on a complex day in the summer of 2026 (involving a large concert and a thunderstorm recovery period) to illustrate in detail the complete workflow of the method provided in this application:

[0115] Phase 1: Profiling and macro-level guidance of big data demand in a normal, long-term cycle (steps S100 to S800)

[0116] First, the system executes step S100, automatically retrieving all historical ferry reservation orders from the port ticketing management platform backend for the past 30 days. Next, step S200 is executed, where the system inputs this denoised historical reservation time-series data into a pre-trained neural network model for temporal feature analysis and pattern recognition. The model identifies the daily peak and trough patterns of passenger flow, automatically dividing the daily operating cycle into multiple basic adjustment periods: 10:00 AM to 4:00 PM is designated as the peak period, 4:00 PM to 10:00 PM as the off-peak period, and the remaining time periods as the low-peak period, thus constructing a normalized spatiotemporal demand profile for this operating cycle.

[0117] Subsequently, the system executes steps S300 and S400, synchronously collecting external behavioral factors through a cross-platform data interface, and performing labeling and weighting processing based on the pre-defined baseline adjustment periods. In this example, the system captures the following multi-source external feature indicators and completes multi-path mapping:

[0118] Time period characteristics: For the peak period from 10:00 AM to 4:00 PM currently being processed, the system generates peak period labels and matches and maps them to the first sensitivity weight;

[0119] Weather and meteorology: Meteorological data detects that there will be severe convective thunderstorms in the future. Based on this, the meteorological identification model outputs a rainstorm climate label and maps it to obtain the corresponding second sensitivity weight.

[0120] Event Information: The system detects that a large-scale concert with tens of thousands of participants will be held at the destination on the other side of the strait that evening. Based on this, the event recognition model outputs a large-scale performing arts event label, which is mapped to a smaller third sensitivity weight (representing that users have a high need for travel and low price sensitivity).

[0121] Calendar information: The day is a regular weekend. The calendar model generates weekend and holiday labels and maps them to obtain the fourth sensitivity weight.

[0122] The system retrieves the built-in weight matrix, normalizes and weights the first to fourth sensitivity weights, and finally solves for the user travel sensitivity weight corresponding to the current peak period.

[0123] Next, the system executes steps S500 and S600, backtracks to the historical reservation database, aggregates and statistically analyzes the supply and demand status of all flights during the same historical period (i.e., past weekend peak periods), and accumulates the total number of historically available tickets and the number of historically successfully booked tickets. The historical comprehensive booking success rate for that period is calculated by dividing the two, and then further mapped and converted into a booking success rate weight that represents the historical supply and demand saturation.

[0124] Finally, the system executes steps S700 and S800, substituting the calculated user travel sensitivity weight and booking success rate weight, along with the pre-set demand adjustment coefficient and supply-demand adjustment coefficient, into a linear combination matrix equation to accurately determine the first-stage capacity adjustment coefficient for the current peak period. The system multiplies the initial base fare by this first-stage capacity adjustment coefficient, calculates and outputs the first ferry guidance fare for the peak period, and publishes it on the ticketing front end. By using a price entry threshold, it begins to guide non-essential vehicles to stagger their travel times at a macro level.

[0125] Phase Two: Micro-level Instantaneous Peak Shaving and Real-time Traffic Closed-Loop Control Near Departure

[0126] As the flight time approaches, the system automatically triggers a micro-level dynamic adjustment closed loop. The system pre-sets a time threshold of 5 hours before departure based on the scheduled flight departure time.

[0127] When the countdown to departure exceeds the 5-hour time threshold, the system immediately activates the road condition coordination and dispatch module. The system retrieves high-definition video streams from surveillance cameras on key access roads surrounding the port in real time, performing frame-by-frame processing to obtain a static image sequence. Subsequently, an image segmentation algorithm is used to divide the surveillance images into independent image blocks, extracting multi-dimensional features such as texture complexity, RGB mean, and perimeter-to-area ratio for each image block. A decision tree classification algorithm is then used to accurately identify whether each image block belongs to the "road" or "vehicle" category. Based on this, a region growing algorithm is used to merge connected components, reconstructing a high-definition, continuous area of ​​roads and vehicle targets around the port.

[0128] In the reconstructed space, the system uses the Hough transform algorithm to detect lane positions online and identify speed limit traffic signs. Simultaneously, in the vehicle area, a target tracking algorithm is used to statistically analyze the current instantaneous traffic flow, and the real-time vehicle speed is calculated based on the vehicle displacement and time difference between consecutive frames. Subsequently, the system inputs the four-dimensional feature vectors of lane lines, traffic signs, traffic flow, and vehicle speed into a K-Means clustering model for iterative optimization, forming multiple data clusters after convergence. By calculating the average feature value within each cluster, the system identifies a pattern of "extremely high traffic flow and average vehicle speed below 20 km / h," automatically mapping and classifying the data cluster as "high congestion" and outputting a high congestion label.

[0129] Because a high congestion label was triggered, the system assigned a significant weight to the traffic congestion level based on mapping rules. Ultimately, the allocation system, based on this traffic congestion weight, performed a secondary adjustment calculation on the initial guidance fare determined in the first phase, calculating the second-phase guidance fare for the approaching departure phase and updating it in real-time to the on-site ticketing terminals. By appropriately raising the instantaneous entry threshold during periods of severe congestion, the system successfully suppressed the indiscriminate influx of vehicles without pre-booked tickets, forcing them to reschedule or arrive at the port at off-peak times, effectively mitigating the instantaneous surge in demand. This completed a closed-loop optimization and allocation of ferry capacity resources, from "macro-level big data prediction" to "micro-level instantaneous traffic condition feedback."

[0130] Figure 2 This is a schematic diagram of the modules of a ferry capacity resource allocation system provided in an embodiment of this application. For example... Figure 2 The ferry capacity resource allocation system 10 shown includes at least the following components: information acquisition module 11, profile generation module 12, resource allocation module 13, and fare generation module.

[0131] In this embodiment, the information acquisition module 11 is used to acquire historical ferry reservation data and collect external behavioral factors. Please refer to [link / reference needed] for details. Figure 1 The details and their corresponding descriptions are not repeated here.

[0132] In this embodiment, the profile generation module 12 is used to perform pattern recognition on the historical reservation data to divide it into multiple basic adjustment periods and construct a spatiotemporal demand profile. For details, please refer to [link / reference needed]. Figure 1 The details and their corresponding descriptions are not repeated here.

[0133] In this embodiment, the resource allocation module 13 is used to process the external behavioral factors according to the spatiotemporal demand profile to obtain the user travel sensitivity weight corresponding to the basic adjustment period; according to the historical reservation data, it obtains the total number of historically bookable tickets and the number of historically successfully booked tickets corresponding to each of the basic adjustment periods; it calculates the historical booking success rate according to the total number of historically bookable tickets and the number of historically successfully booked tickets to obtain the booking success rate weight characterizing the historical supply and demand of the spatiotemporal demand profile; and it calculates the first-stage capacity adjustment coefficient corresponding to the basic adjustment period based on the user travel sensitivity weight, the booking success rate weight, and preset demand adjustment coefficients and supply and demand adjustment coefficients. For details, please refer to [link to relevant documentation]. Figure 1 The details and their corresponding descriptions are not repeated here.

[0134] In this embodiment, the fare generation module 14 is used to determine the guiding fare for each of the basic adjustment periods of the ferry based on the first-stage capacity adjustment coefficient. For details, please refer to [link / reference needed]. Figure 1 The details and their corresponding descriptions are not repeated here.

[0135] Figure 3 This is an electronic device 20 provided in one embodiment of this application. For example... Figure 3 As shown, the electronic device 20 includes at least the following components: a processor 21 and a memory 22.

[0136] In this embodiment, the memory 22 is used to store executable instructions of the processor 21, which, when configured to execute instructions, implement... Figure 1 The method shown in the figure.

[0137] In one embodiment of this application, the program operating in the electronic device 20 may be a program that controls a central processing unit (CPU) or similar device to achieve the functions of the above-described embodiments of the present invention (a program that enables the computer to function). The information processed by these devices is then temporarily stored in random access memory (RAM) during processing, and subsequently stored in various ROMs such as read-only memory (Flash ROM) and hard disk drives (HDDs), and read, corrected, and written by the CPU as needed.

[0138] It should be noted that a portion of the electronic device 20 described above can also be implemented using a computer. In this case, the program for implementing the control function can be recorded on a computer-readable recording medium, and the program recorded on the recording medium can be read into the computer and executed.

[0139] It should be noted that the term "computer" as used here refers to a computer built into electronic device 20, employing hardware including an operating system and peripheral devices. Furthermore, "computer-readable recording media" refers to removable media such as floppy disks, magneto-optical disks, ROMs, and CD-ROMs, as well as storage devices such as hard drives built into the computer.

[0140] Furthermore, a "computer-readable recording medium" can include: a medium that dynamically stores a program for a short period of time, such as a communication line used when transmitting a program via a network such as the Internet or a communication line such as a telephone line; or a medium that stores a program for a fixed period of time, such as volatile memory inside a computer that serves as a server or client in this case. In addition, the aforementioned program can be a program used to implement the above-mentioned functions, or it can be a program that can implement the above-mentioned functions by combining with programs already recorded in the computer.

[0141] It is understood that the ferry capacity resource allocation method based on multi-source data provided in this application's embodiments scientifically divides multiple basic adjustment periods, including peak, off-peak, and low-peak periods, by acquiring long-term historical ferry reservation data for pattern recognition, and constructs an accurate spatiotemporal demand profile. Based on this, the system not only collaboratively processes dynamically changing external behavioral factors such as weather and events to dynamically assess the rigidity of user travel needs (sensitivity weight), but also deeply mines the reservation success rate weight, which characterizes historical supply and demand, by combining historical bookable and successfully booked tickets. By performing multi-dimensional linear calculations with the weights of multi-source internal and external data and preset coefficients, the first-stage capacity adjustment coefficient for the corresponding period is accurately determined and converted into differentiated guiding fares. This method effectively overcomes the limitations of traditional allocation methods that rely solely on fixed fares or single local reservation data, leading to uneven distribution of capacity resources and low overall utilization. In the normal long-term dimension, it successfully achieves spatiotemporal diversion through price leverage, effectively diverting congested demand during peak periods and fully stimulating the utilization of idle resources during off-peak periods, thus achieving optimized allocation and efficient utilization of ferry baseline capacity resources.

[0142] Those skilled in the art should recognize that the above embodiments are only used to illustrate this application and are not intended to limit this application. Any appropriate changes and variations made to the above embodiments within the essential spirit and scope of this application fall within the scope of protection claimed in this application.

Claims

1. A method for allocating ferry capacity resources based on multi-source data, characterized in that, The method includes: Obtain historical ferry reservation data; The historical reservation data is subjected to pattern recognition to divide it into multiple basic adjustment periods, and a spatiotemporal demand profile is constructed. Collect external behavioral factors; The external behavioral factors are processed based on the spatiotemporal demand profile to obtain the user travel sensitivity weights corresponding to the basic adjustment period. Based on the historical reservation data, obtain the total number of historically bookable tickets and the number of historically successfully booked tickets for each of the basic adjustment periods; The historical booking success rate is calculated based on the total number of historically available tickets and the number of historically successfully booked tickets, so as to obtain the booking success rate weight that characterizes the historical supply and demand of the spatiotemporal demand profile. The first-stage capacity adjustment coefficient corresponding to the basic adjustment period is determined by calculating the user travel sensitivity weight, the booking success rate weight, and the preset demand adjustment coefficient and supply-demand adjustment coefficient. The guiding fare for each basic adjustment period of the ferry is determined based on the capacity adjustment coefficient of the first stage.

2. The ferry capacity resource allocation method based on multi-source data according to claim 1, characterized in that, The process of processing the external behavioral factors based on the spatiotemporal demand profile to obtain the user travel sensitivity weights corresponding to the basic adjustment period includes: Generate corresponding time period labels based on each of the basic adjustment time periods, and obtain the first sensitivity weight based on the time period label mapping; Obtain the external behavior feature indicators corresponding to each of the basic adjustment periods, and generate corresponding multi-path dynamic environment sensitivity weights based on the external behavior feature indicators. The first sensitivity weight and each of the multi-path dynamic environment sensitivity weights are normalized and weighted to obtain the user travel sensitivity weight corresponding to the basic adjustment period.

3. The ferry capacity resource allocation method based on multi-source data according to claim 2, characterized in that, The external behavioral characteristic indicators include weather and meteorological data corresponding to each of the basic adjustment periods, activity information data for future periods, and calendar information data; The process of generating corresponding multi-path dynamic environment sensitivity weights based on the external behavioral feature indicators includes: Climate labels are generated based on the weather and meteorological data, and a second sensitivity weight is obtained by mapping the climate labels. Activity tags are generated based on the activity information data, and a third sensitivity weight is obtained by mapping the activity tags. Holiday tags are generated based on the calendar information data, and a fourth sensitivity weight is obtained based on the mapping of the holiday tags.

4. The ferry capacity resource allocation method based on multi-source data according to claim 3, characterized in that, The climate label, the activity label, and the holiday label are respectively weighted and mapped according to the degree of travel obstruction, the scale of the activity, and the length of the holiday; The higher the degree of travel obstruction for a climate label, the greater the corresponding second sensitivity weight; the larger the scale of the activity and the higher the travel necessity for the activity label, the smaller the corresponding third sensitivity weight; and the longer the holiday duration and the greater the travel aggregation effect for a holiday label, the smaller the corresponding fourth sensitivity weight.

5. The ferry capacity resource allocation method based on multi-source data according to claim 1, characterized in that, The method further includes: Set time thresholds based on the scheduled departure times of ferry flights; When the countdown exceeds the time threshold, real-time traffic data of the area near the ferry port is acquired, and the real-time traffic data is analyzed to obtain the current traffic congestion situation. Based on the traffic congestion situation, determine the traffic congestion situation label and the corresponding traffic congestion situation weight; The guiding fare is adjusted a second time based on the weight of the traffic congestion situation to obtain the second-stage guiding fare.

6. The ferry capacity resource allocation method based on multi-source data according to claim 5, characterized in that, The process of acquiring real-time traffic data in the area near the ferry port and analyzing the real-time traffic data to obtain the current traffic congestion situation includes: Real-time surveillance video data of the area near the ferry port is acquired, and the real-time surveillance video data is processed by frame segmentation to obtain real-time surveillance image data. The real-time monitoring image data is segmented using an image segmentation algorithm to obtain multiple image blocks; The image patches are classified according to the decision tree classification algorithm and based on the multivariate features of each image patch to determine the category of each image patch; The region growing algorithm is used to merge image patches of the same category to obtain road regions and vehicle regions under the same monitoring image; The traffic congestion situation is obtained by jointly analyzing the spatial and state relationships of the road area and the vehicle area.

7. The ferry capacity resource allocation method based on multi-source data according to claim 6, characterized in that, The process of jointly analyzing the spatial and state relationships between the road area and the vehicle area to obtain the traffic congestion situation includes: The Hough transform algorithm is used in the road area to identify and obtain lane line position information and traffic sign information; Instantaneous traffic flow is counted in the vehicle area, and the real-time speed of the vehicle is calculated based on the displacement and time difference of the vehicle between consecutive frames. Using the K-Means algorithm, multi-dimensional feature clustering analysis is performed on the lane line position information, the traffic sign information, the traffic flow, and the vehicle speed to form multiple data clusters; Based on the average feature value within each data cluster, each data cluster is divided and mapped to different congestion levels to obtain the traffic congestion situation.

8. A ferry capacity resource allocation system based on multi-source data, characterized in that, The system is adapted to implement the method as described in any one of claims 1 to 7, wherein the system comprises: The information acquisition module is used to acquire historical ferry reservation data and collect external behavioral factors; The profile generation module is used to perform pattern recognition on the historical reservation data to divide it into multiple basic adjustment periods and construct a spatiotemporal demand profile. The resource allocation module is used to process the external behavioral factors according to the spatiotemporal demand profile to obtain the user travel sensitivity weight corresponding to the basic adjustment period; based on the historical reservation data, it obtains the total number of historically bookable tickets and the number of historically successfully booked tickets for each of the basic adjustment periods; it calculates the historical booking success rate based on the total number of historically bookable tickets and the number of historically successfully booked tickets to obtain the booking success rate weight that represents the historical supply and demand of the spatiotemporal demand profile; and it calculates and determines the first-stage capacity adjustment coefficient corresponding to the basic adjustment period based on the user travel sensitivity weight, the booking success rate weight, and preset demand adjustment coefficients and supply and demand adjustment coefficients. The fare generation module is used to determine the guiding fare for each of the basic adjustment periods of the ferry based on the capacity adjustment coefficient of the first stage.

9. An electronic device, characterized in that, include: processor; as well as A memory having computer-readable instructions stored thereon for controlling the processor to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program is configured to perform the method described in any one of claims 1 to 7 when executed.

Images