Anchorage space-time utilization efficiency evaluation method based on time geography

Abstract

The application discloses an anchorage space-time utilization efficiency evaluation method based on time geography, comprising the following steps: S1, obtaining original AIS data covering multiple target anchorages of a target port and preprocessing; then extracting anchoring state data of each target anchorage; S2, based on the anchoring state data, extracting habitual anchorage of each target anchorage respectively; S3, calculating static characteristic indexes of space-time utilization efficiency of each target anchorage respectively; S4, calculating dynamic stability indexes of space-time utilization efficiency of each target anchorage respectively; S5, standardizing each index obtained from S3 and S4, and then realizing operation type evaluation through clustering analysis to obtain operation types of each target anchorage. The method promotes anchorage research to a micro space unit level and breaks through the limitation of static load; through unsupervised clustering, the anchorage is divided into four operation types, which provides direct data support for the port and shipping management department to implement differentiated and accurate regulation and control.

Description

Technical Field

[0001] This invention belongs to the field of port management and maritime traffic engineering technology, specifically relating to a method for evaluating the spatiotemporal utilization efficiency of anchorages based on time geography. Background Technology

[0002] As a critical buffer node in the port logistics chain, the operational efficiency of anchorages directly impacts the overall port capacity and supply chain stability. With the continuous growth of global maritime trade, anchorage resources are becoming increasingly scarce, making the scientific and accurate assessment of anchorage utilization a pressing technical issue for port management.

[0003] Currently, the assessment of anchorage utilization mainly follows two technical paradigms:

[0004] The first approach is based on a static capacity assessment paradigm. This paradigm calculates the theoretical design capacity of the anchorage by taking into account physical parameters such as anchorage area, safe distances between vessels, and water depth. This method focuses on the static configuration of the anchorage's hardware resources.

[0005] The second approach is a time-cumulative assessment paradigm. This paradigm calculates the anchorage occupancy rate by dividing the time a ship spends in the anchorage by the total duration of the statistical period, thereby reflecting the load intensity of the anchorage.

[0006] However, the aforementioned existing technological paradigm has the following significant drawbacks:

[0007] (1) The analysis granularity is too coarse and there is a "black box" effect: the existing technology treats the entire anchorage as a homogeneous whole, ignoring the differences in utilization patterns between different spatial locations within the anchorage, resulting in a lack of spatial targeting in management decisions.

[0008] (2) The indicators are too singular and cannot assess the dynamic process: Existing technologies focus too much on static load indicators and lack quantitative characterization of the dynamic utilization process of anchorage resources (such as the frequency and rhythm of ship entry and exit).

[0009] (3) It is impossible to distinguish between anchorages with similar surface loads but different internal conditions: two anchorages may have the same time occupancy rate, but one is in a state of high-efficiency turnover while the other is on the verge of rigid congestion. Existing technology cannot distinguish between these two different operational health conditions.

[0010] (4) Lack of multi-dimensional automated classification and diagnosis capabilities: Existing technologies are unable to scientifically classify and automatically diagnose anchorage operation modes based on multi-dimensional dynamic characteristics.

[0011] In summary, there is an urgent need for a method that can comprehensively diagnose the utilization efficiency and stability of anchorages from a microscopic spatiotemporal dynamic perspective. Summary of the Invention

[0012] To address the problems existing in the prior art, this invention provides a comprehensive diagnostic method for the spatiotemporal utilization efficiency of anchorages based on time geography, which can realize the automated classification and evaluation of anchorage operation modes.

[0013] Therefore, the present invention adopts the following technical solution:

[0014] An anchorage spatiotemporal utilization efficiency evaluation method based on time geography includes the following steps:

[0015] S1, acquire raw AIS data covering multiple target anchorages of the target port; preprocess the raw AIS data; use a density-based dwell point identification algorithm to extract anchorage status data of each target anchorage from the preprocessed AIS data; the anchorage status data is the data of the trajectory segment of the ship moving at a low speed continuously within the target anchorage waters and within the ship's anchorage area for a preset duration.

[0016] S2, based on the anchorage status data, extract the habitual anchorage of each target anchorage, specifically including:

[0017] S21, divide the target anchorage into multiple grid cells of equal area, and for each grid cell, calculate its cumulative anchorage time within the statistical period;

[0018] S22, calculate the global statistics and the sum of local anchorage times for each grid cell;

[0019] S23, calculate the Getis-Ord Gi* statistic and identify anchorage hotspots, and use these hotspots as candidate habitual anchorages;

[0020] S24, calculate the directional distribution ellipse of each anchorage hotspot area, determine the typical anchorage radius, and generate a circular buffer zone for each candidate habitual anchorage as the final spatial range of that candidate habitual anchorage.

[0021] S25, filter the candidate habitual anchors to obtain the final set of habitual anchors;

[0022] S3, calculate the static characteristic indicators of the spatiotemporal utilization efficiency of each target anchorage, including the anchorage time occupancy rate. Spatial Gini coefficient and average anchorage time ;

[0023] S4, calculate the dynamic stability index of the spatiotemporal utilization efficiency of each target anchorage, including the average state transition rate. and spatiotemporal efficiency heterogeneity ;

[0024] S5 standardizes the indicators obtained from S3 and S4, and then uses cluster analysis to evaluate the operation type and obtain the operation type of each target anchorage.

[0025] In step S1 above:

[0026] The preprocessing includes deleting duplicate records, removing invalid data whose latitude and longitude exceed the target anchorage waters or whose speed is abnormal, and segmenting the trajectory based on the vessel identification code;

[0027] The method for extracting the anchorage status data of each target anchorage is as follows:

[0028] Set speed threshold Distance threshold and time threshold The speed threshold, distance threshold, and time threshold are set according to the distribution of vessel types and water conditions at the target anchorage;

[0029] For consecutive trajectory points of the same vessel, if the following conditions are met within a preset time window:

[0030] The ship's average speed or track point speed is less than the speed threshold. ;

[0031] The spatial movement distance, maximum distribution radius, or positional dispersion of the trajectory points are less than a distance threshold. ;

[0032] The duration of the trajectory segment is greater than the time threshold. ;

[0033] The data for the corresponding trajectory segment will then be determined as the anchoring status data.

[0034] In step S21 above:

[0035] The grid unit is The area of ​​each grid cell is determined comprehensively based on the total area of ​​the anchorage, the type of anchorage, and the required calculation accuracy.

[0036] For each grid cell Statistics on its statistical period Cumulative anchorage time The calculation formula is:

[0037] ,

[0038] in, For grid cells The total number of anchoring incidents that occurred within the area. For the first occurrence within this grid cell The duration of each anchoring event is obtained from the anchoring status data.

[0039] In step S22 above:

[0040] Define grid cells and Spatial weights are The spatial weights are calculated using any one of the following methods: inverse distance weighting, K-nearest neighbor weighting, or fixed distance bandwidth.

[0041] The global statistics include the arithmetic mean of the cumulative anchorage duration of all grid cells. and standard deviation The calculation formulas are as follows:

[0042] ,

[0043] ,

[0044] The total duration of the local anchorage For grid cells The weighted sum of the cumulative anchorage times of all other grid cells is calculated using the following formula:

[0045] ,

[0046] in, For grid cells In the statistical period The cumulative anchorage time within the area.

[0047] In step S23 above:

[0048] For each grid cell Its Getis-Ord Gi* statistic The calculation formula is:

[0049] ,

[0050] For each grid cell The corresponding significance probability P-value The calculation formula is:

[0051] ,

[0052] in, The cumulative distribution function of the standard normal distribution;

[0053] Set significance level The value of is obtained, and the corresponding critical Z value is acquired; when Greater than the critical Z value and Less than the significance level When the value is [value], the grid cell The anchorage duration within its spatial neighborhood is significantly higher than the global average.

[0054] When Greater than the critical Z value Less than the significance level Adjacent grid cells that are both spatially connected are aggregated into the same mooring hotspot region. Each mooring hotspot region corresponds to a candidate habitual anchorage, thus obtaining all candidate habitual anchorages in each target anchorage.

[0055] In step S24 above:

[0056] The center of the directional distribution ellipse is used as the estimated center of the corresponding candidate habitual anchor position, and the coordinates of the estimated center are... The calculation formula is:

[0057] , ,

[0058] in, This represents the total number of grid cells within the anchorage hotspot area. For each grid cell within the hotspot area The center coordinates;

[0059] Typical anchoring radius The determination process is as follows:

[0060] Calculate the reference radius :

[0061] ,

[0062] in, To design the overall length of the ship, For water depth, the corresponding information is obtained from the port design specifications based on the designed vessel type for each anchorage. and Range of values;

[0063] Typical anchorage radius exist Based on the theoretical range, increase by 20%;

[0064] The circular buffer zone is centered on the estimated center and extends to the typical anchorage radius. Generates the radius.

[0065] In step S3 above:

[0066] The anchorage time occupancy rate The calculation formula is:

[0067] ,

[0068] in, This represents the total duration of all anchoring events occurring in the anchorage within the statistical period. For the first The actual anchoring duration of this anchoring event This represents the total number of anchoring events that occurred in the anchorage during the statistical period. For the design capacity of the anchorage, This represents the total duration of the statistical period.

[0069] The spatial Gini coefficient The calculation process is as follows:

[0070] First, the cumulative anchorage durations of the grid cells obtained in S21 are sorted from smallest to largest, resulting in an ordered sequence of the cumulative anchorage durations of each grid cell. ,in, The cumulative anchorage duration is the minimum anchorage duration for the grid cell. The cumulative anchorage time is the longest anchorage time for the grid cell with the longest anchorage time. The first number after sorting by anchorage duration in ascending order. Anchorage duration for each grid cell;

[0071] Then, the sorted sequence is substituted into the spatial Gini coefficient formula for calculation:

[0072] ,

[0073] The average anchorage duration The calculation formula is:

[0074] .

[0075] Step S4 above includes the following steps:

[0076] Constructing a state sequence: Collecting each habitual anchor point within a statistical period All anchoring events within the timeline are recorded and arranged in chronological order; events with overlapping times or very short intervals are merged to form a continuous state sequence; by traversing the entire timeline, the state of the habitual anchorage at each moment is marked: 0 indicates that it is idle and not occupied; 1 indicates that it is occupied and a ship is anchored.

[0077] The number of state changes in the state sequence is counted, including the number of events from "0→1" and "1→0", and recorded as the state transition count. ;in, For the habitual anchorage numbering, , The total number of habitual anchorages within the target anchorage;

[0078] Calculate the state transition frequency :

[0079] ,

[0080] Calculate the average state transition rate :

[0081] ,

[0082] Computational spatiotemporal efficiency heterogeneity :

[0083] .

[0084] Step S5 above includes the following steps:

[0085] Z-score standardization is performed on each metric to transform them to the same order of magnitude, resulting in a standardized metric. Five-dimensional data set The number of target anchorages, ;

[0086] Unsupervised clustering algorithm is used to standardize the data. Cluster analysis was performed on the five-dimensional data to... The target anchor points are aggregated into four categories, and the cluster centers of each category are obtained, represented as a five-dimensional vector. ,in, This represents the anchorage time occupancy rate component. For the spatial Gini coefficient component, This is the average anchorage duration component. For the average state transition rate component, For spatiotemporal efficiency heterogeneity components;

[0087] According to the five-dimensional vector The target anchorage is determined to be either a high-pressure saturation type, a high-efficiency circulation type, a silent reserve type, or a batch processing operation type.

[0088] In step S5 above:

[0089] If the components of the five-dimensional vector of the cluster center satisfy: , , , , If so, all target anchorages within this category are determined to be high-pressure saturation type. These anchorages operate under high load for a long time, with space occupancy highly concentrated in a few grids. Ships stay for long periods, the frequency of anchorage state transitions is low, and the states of each anchorage are highly similar, lacking flexible allocation space. The characteristics of this type of anchorage are high load, long stay, low turnover, and high homogeneity, facing extremely high congestion risks and operational pressures.

[0090] If the components of the five-dimensional vector of the cluster center satisfy: , , , , Then all target anchorages within this category are determined to be of the high-efficiency circulation type. Although this type of anchorage has a high load, due to the relatively balanced space utilization, fast ship turnaround speed, and frequent anchorage status changes, the status of each anchorage is highly similar. The characteristics of this type of anchorage are high load, high turnover, and balanced use. The resource utilization shows a healthy saturation characteristic, rather than rigid congestion. It is in a healthy state of efficient operation and is the optimal operating state.

[0091] If the components of the five-dimensional vector of the cluster center satisfy: , , , , If all target anchorages within this category are classified as silent reserve anchorages, these anchorages have low overall load, sparse space occupancy, short vessel dwell time, inactive anchorage status transitions, and highly similar statuses across anchorages, exhibiting a pulse-like usage pattern of general idleness and occasional activation. These anchorages are characterized by low load, low activity, and high homogeneity, and are in a state of waiting to be activated. The main problem with these anchorages is the conflict between resource idleness and emergency support functions.

[0092] If the components of the five-dimensional vector of the cluster center satisfy: , , If all target anchorages within this category are determined to be batch operation type, the overall load of this type of anchorage is high, but the internal space and time utilization is extremely uneven, the occupation is concentrated in some areas, some anchorage status changes frequently while others are almost immobile; the characteristics of this type of anchorage are high load, strong imbalance, segmented use, and highly correlated with the batch arrival of cargo and concentrated operation mode. The anchorage serves specific ship types that arrive in batches, exhibiting "tidal" usage characteristics.

[0093] Compared with the prior art, the present invention has the following beneficial effects:

[0094] 1. This invention transforms the concept of "state transition" in time geography into a computable dynamic stability index (average state transition rate, spatiotemporal efficiency heterogeneity), breaking through the limitations of traditional methods that only focus on static loads, and can reveal the inherent dynamic rhythm of anchorage utilization.

[0095] 2. Spatial Refinement: This invention proposes the concept and extraction method of "habitual anchorage", which advances anchorage research from the traditional "overall black box" to the "micro-spatial unit" level, providing a stable spatial reference for refined management.

[0096] 3. Comprehensiveness of diagnosis: This invention constructs a five-dimensional indicator system that integrates "intensity-distribution-efficiency-activity-balance", which can accurately distinguish between different operating states that traditional methods cannot distinguish, such as "high load and high efficiency turnover" and "high load and rigid congestion".

[0097] 4. Intuitive Decision Making: This invention classifies anchorages into four operational types—high-pressure saturation, high-efficiency circulation, silent reserve, and batch processing—through unsupervised clustering. This transforms complex operational data into intuitive operational types, providing direct data support for port and shipping management departments to implement differentiated and precise control measures. Attached Figure Description

[0098] Figure 1 This is a flowchart of the anchorage space-time utilization efficiency evaluation method according to an embodiment of the present invention. Detailed Implementation

[0099] The technical solution of the invention will be clearly and completely described below with reference to the accompanying drawings and embodiments. Obviously, the following embodiments are only some embodiments of the present invention. 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.

[0100] Example

[0101] like Figure 1 As shown, the present invention provides a method for evaluating the spatiotemporal utilization efficiency of anchorages based on time geography, comprising the following steps:

[0102] S1, Data Preprocessing and Anchorage Status Data Extraction:

[0103] First, raw AIS (Automatic Identification System) data covering multiple target anchorages at the target port is acquired. The following preprocessing operations are performed on the raw AIS data: duplicate records are deleted, invalid data with latitude and longitude exceeding the target anchorage waters or abnormal speeds are removed, and the trajectory is segmented based on the vessel identification code. In this embodiment, the number of target anchorages is... , .

[0104] Next, a density-based stop point identification algorithm (SPID) is used to extract anchorage status data for each target anchorage from the preprocessed AIS data. This anchorage status data refers to the trajectory segment data of a vessel moving continuously at low speed within the target anchorage area for a preset duration. One vessel anchorage status data point corresponds to one anchorage event.

[0105] Specifically, setting a speed threshold Distance threshold and time threshold Among them, for consecutive trajectory points of the same vessel, if the following conditions are met within a preset time window:

[0106] 1. The ship's average speed or speed at the trajectory point is less than the speed threshold. ;

[0107] 2. The spatial movement distance, maximum distribution radius, or positional dispersion of the trajectory points is less than the distance threshold. ;

[0108] 3. The duration of the trajectory segment exceeds the time threshold. ;

[0109] The data for the corresponding trajectory segment will then be determined as the anchoring status data.

[0110] The speed threshold, distance threshold, and time threshold can be set according to the distribution of vessel types and water conditions in the target water area. In this embodiment, the distance threshold... Set to 0.2 nautical miles, time threshold Set to 1 hour.

[0111] S2, Based on the anchorage status data, extract the habitual anchorage of each target anchorage:

[0112] In actual anchorage operations, vessels are not evenly distributed throughout the entire anchorage area, but rather tend to cluster and anchor in specific locations. These locations typically possess advantages such as suitable water depth, good seabed conditions, and proximity to waterways or berths, reflecting the spatial usage preferences of ship operators and port dispatchers. This invention defines such stable anchorage areas with a statistically significant spatial clustering tendency as "habitual anchorages."

[0113] The habitual anchorage is the core microscopic analysis unit of this invention. Its significance lies in the fact that traditional anchorage assessment methods treat the entire anchorage as a homogeneous whole, failing to reveal differences in internal space utilization. In contrast, the habitual anchorage elevates the anchorage from a macroscopic "black box" to the microscopic spatial unit level, making subsequent dynamic state transformation analysis of each anchorage possible. Each habitual anchorage corresponds to an independent analysis unit in the subsequent dynamic analysis.

[0114] Specifically, the following steps are included:

[0115] S21, Divide the spatial grid and calculate the cumulative anchorage time:

[0116] Each target anchorage is divided into The grid consists of grid cells of equal area. The area of ​​each grid cell is determined comprehensively based on the total area of ​​the anchorage, the anchorage type, and the required calculation accuracy. In this embodiment, referring to the typical ship length of 200 meters at the target port, and considering the data analysis accuracy, half the typical ship length is used as the side length of the grid cell, i.e., the preferred area of ​​the grid cell is 100m². 100m.

[0117] For each grid cell Statistics on its statistical period Cumulative anchorage time The calculation formula is:

[0118] ,

[0119] in, For grid cells The total number of anchoring incidents that occurred within the area. For the first occurrence within this grid cell The duration of each mooring event (in hours) is obtained from the mooring status data.

[0120] S22, calculate the global statistics and the sum of local anchorage times for each grid cell, as follows:

[0121] (1) Calculate spatial weights:

[0122] To characterize the spatial proximity relationships between mesh cells, mesh cells are defined. and Spatial weights are The spatial weights reflect the strength of spatial interactions between grid cells. This embodiment uses the inverse distance weighting method to calculate the spatial weights; the closer the grid cells are, the greater the weight. The calculation formula is:

[0123] ,

[0124] in, For grid cells and The Euclidean distance between them; For distance attenuation parameters, in this embodiment =1; when When, define =1.

[0125] Spatial weights can also be calculated using other methods, such as K-nearest neighbor weights and fixed-distance bandwidth.

[0126] (2) Calculate the global statistics:

[0127] The global statistics include the arithmetic mean of the cumulative anchorage duration of all grid cells. and standard deviation The calculation formulas are as follows:

[0128] ,

[0129] .

[0130] (3) Calculate the total duration of partial anchorage:

[0131] For each grid cell Define the total local anchorage time. For grid cells The formula for calculating the total local anchorage time is: (This is a weighted sum of the cumulative anchorage times of all other grid cells.)

[0132] ,

[0133] in, For grid cells In the statistical period The cumulative anchorage time within the area.

[0134] Reflects the grid cells The intensity of anchoring activities in and around the area.

[0135] S23, Calculate Getis-Ord Gi* statistics and identify anchorage hotspots:

[0136] For each grid cell Its Getis-Ord Gi* statistic Based on the sum of the above local anchoring times The calculation formula is as follows:

[0137] ,

[0138] The Getis-Ord Gi* statistic follows a standard normal distribution, and its value... This is the Z score.

[0139] For each grid cell The corresponding significance probability value The calculation formula is:

[0140] ,

[0141] in, This is the cumulative distribution function of the standard normal distribution.

[0142] Set significance level The value of Z is obtained, and the corresponding critical Z value is acquired. In this embodiment, the significance level is... The value is 0.05, and the corresponding critical Z value is 1.96.

[0143] when Greater than the critical Z value ( >1.96) and Less than the significance level value( When <0.05), it indicates that the mesh element is... The anchorage duration within its spatial neighborhood is significantly higher than the global average.

[0144] Will >1.96 Adjacent grid cells with a density <0.05 and spatial connectivity are aggregated into the same mooring hotspot region. Each mooring hotspot region corresponds to a candidate habitual anchorage, thereby obtaining all candidate habitual anchorages in each target anchorage.

[0145] S24, Calculate the directional distribution ellipse of each anchorage hotspot area and determine the typical anchorage radius. Generate a circular buffer:

[0146] For each anchoring hotspot region, its orientation distribution ellipse is calculated, and the center of the ellipse is used as the estimated center of the candidate habitual anchorage corresponding to that hotspot region. The center coordinates of the orientation distribution ellipse are... The calculation formula is:

[0147] , ,

[0148] in, This represents the total number of grid cells within the anchorage hotspot area. For each grid cell within the hotspot area The center coordinates.

[0149] Typical anchoring radius This is a key parameter for defining the spatial range of habitual anchorages. First, the reference radius is calculated according to the standard formula in the "General Design Code for Seaports" (JTJ 211-99) for single-anchored vessels under wind and wave conditions ≤ 7. :

[0150] ,

[0151] in, The overall length of the designed ship (in meters). Water depth (unit: meters). The corresponding depth is obtained from the port design specifications based on the designed vessel type for each anchorage. and Range of values.

[0152] Based on this, taking into account operational safety margins and specific port requirements (such as vessel waiting patterns, dangerous goods handling requirements, and emergency response needs), the final typical anchorage radius was determined. , making exist The theoretical range is appropriately increased by 20% to ensure a safety margin.

[0153] Using the estimated center of each candidate habitual anchorage as the center and the typical mooring radius as the center, A circular buffer is generated for the radius, serving as the final spatial extent of the candidate habitual anchor.

[0154] S25, filter the candidate habitual anchors to obtain the final set of habitual anchors.

[0155] In real-world data, temporary hotspots may exist due to a few isolated mooring events. To eliminate such noise interference with the analysis, a filtering threshold is set: only candidate habitual anchorages with five or more mooring events are retained. This filtering threshold can be adaptively adjusted based on the amount of data and the frequency of anchorage use.

[0156] After filtering, the final set of habitual anchorages within each target anchorage is obtained, with each target anchorage including... A habitual anchorage ( Depending on the specific circumstances of different target anchorages (and the values ​​may vary in actual operation), each customary anchorage corresponds to a circular spatial unit, the center of which reflects the spatial preference of the ship's anchoring, and its radius reflects the typical range of anchoring activities of the ship.

[0157] S3, calculate the static characteristic indicators of the spatiotemporal utilization efficiency of each target anchorage:

[0158] Static characteristic indicators are used to describe the utilization status of anchorage resources from a "quantitative" dimension, including anchorage time occupancy rate. Spatial Gini coefficient and average anchorage time Three indicators.

[0159] (1) Anchorage time occupancy rate :

[0160] The anchorage time occupancy rate This refers to the ratio of the total duration of all anchoring events within an anchorage to the theoretically maximum available total duration of the anchorage within a statistical period. It is used to characterize the overall time utilization intensity of the anchorage. The calculation formula is:

[0161] ,

[0162] in, This represents the total duration of all anchoring events occurring in the anchorage within the statistical period. For the first The actual anchoring duration of this anchoring event (in hours). This represents the total number of anchoring events that occurred in the anchorage during the statistical period. Design capacity of the anchorage (unit: vessels). This represents the total duration of the statistical period (in hours). All data is obtained from the anchorage status data.

[0163] This ratio reflects the proportion of time the anchorage is occupied during the statistical period. The closer the value is to 1, the higher the time utilization intensity of the anchorage; The closer the value is to 0, the lower the time utilization intensity of the anchorage.

[0164] (2) Spatial Gini coefficient :

[0165] The spatial Gini coefficient This is used to quantify the uneven distribution of anchorage duration among different spatial units within the anchorage waters. The closer the value is to 0, the more balanced the space utilization; the closer it is to 1, the more concentrated the space utilization.

[0166] The calculation process is as follows:

[0167] First, the cumulative anchorage durations of the grid cells obtained in S21 are sorted from smallest to largest, resulting in an ordered sequence of the cumulative anchorage durations of each grid cell. ,in, The cumulative anchorage duration is the minimum anchorage duration for the grid cell. The cumulative anchorage time is the longest anchorage time for the grid cell with the longest anchorage time. The first number after sorting by anchorage duration in ascending order. Anchorage duration for each grid cell.

[0168] Then, the sorted sequence is substituted into the spatial Gini coefficient formula for calculation:

[0169] ,

[0170] The numerator of this formula calculates the sum of anchorage times for each grid cell, weighted according to its sorting position, with weighting coefficients... This gives greater weight to grid cells that are ranked lower (with longer mooring times); the denominator is the product of the total number of grid cells and the total mooring time, used for normalization.

[0171] (3) Average anchorage duration :

[0172] The average anchorage duration This refers to the average duration of each anchoring event in the anchorage within a statistical period, used to reflect the turnaround efficiency of the anchorage. The calculation formula is:

[0173] ,

[0174] This indicator is measured in hours. Generally, it refers to the average anchorage duration. The shorter the anchorage, the higher the turnaround efficiency and the stronger the ship mobility. The longer the anchorage, the longer the ship stays at anchor, which may indicate problems such as poor scheduling or port congestion.

[0175] S4, calculate the dynamic stability index of the spatiotemporal utilization efficiency of each target anchorage:

[0176] The dynamic stability index is one of the core innovations that distinguishes this invention from existing technologies. Based on the state transition theory of time geography, this invention constructs an average state transition rate by analyzing the switching frequency and rhythm of each habitual anchorage between the "idle" and "occupied" states. and spatiotemporal efficiency heterogeneity Two dynamic indicators describe the utilization activity and stability of anchorage resources from a "quality" dimension. Details are as follows:

[0177] S41, Construct the idle-occupied time-series state sequence, as follows:

[0178] Collect each habitual anchor point in the statistical period All anchoring event records are arranged chronologically. Events with overlapping times or very short intervals are merged to form a continuous state sequence. For example, if the time interval between two anchoring events is less than a preset threshold (e.g., 30 minutes), they are considered as the same continuous occupancy. By traversing the entire timeline, the state of the habitual anchorage at each moment is marked: 0 indicates idle (unoccupied), and 1 indicates occupied (a ship is anchored). The number of state changes in this binary state sequence, i.e., the number of events from "0→1" (idle → occupied) and "1→0" (occupied → idle), is recorded as the state transition count. .in, For the habitual anchorage numbering, , This represents the total number of habitual anchorages within the target anchorage.

[0179] S42, Calculate the average state transition rate :

[0180] The average state transition rate is used to characterize the overall dynamic activity of the anchorage, that is, the frequency of ships entering and leaving the anchorage.

[0181] For the number is The habitual anchorage, its state transition frequency Defined as the number of state transitions per unit time, the calculation formula is:

[0182] ,

[0183] in, The total duration of the statistical period (unit: days). This represents the number of state transitions of the habitual anchor within the statistical period. The unit is "times / day", which indicates the average number of state transitions that occur per day.

[0184] The average state transition rate of the anchorage is obtained by taking the arithmetic mean of the state transition frequencies of all habitual anchorages within the anchorage:

[0185] ,

[0186] The higher the value, the more frequently ships enter and leave the anchorage, the faster the resource turnover, and the higher the dynamic activity. The lower the value, the more stable the anchorage status, the slower the change, and the lower the dynamic activity.

[0187] S43, Calculating spatiotemporal efficiency heterogeneity :

[0188] Spatiotemporal efficiency heterogeneity is used to characterize the degree of difference in the rhythm of state transitions among habitual anchorages within an anchorage, i.e. the balance of resource utilization.

[0189] First, calculate the deviation of the state transition frequency of each habitual anchorage from the average value. And square it, we get The purpose of squaring is to eliminate the mutual cancellation of positive and negative deviations and to amplify the effect of larger deviations.

[0190] Then, sum the squared deviations of all habitual anchorages and divide by the total number of habitual anchorages to obtain the variance:

[0191] ,

[0192] The larger the value, the more likely it is that there are both high-frequency switching (flexible use) and low-frequency switching (rigid occupation) habitual anchorages within the anchorage, indicating a significant imbalance in resource utilization patterns. The smaller the value, the more consistent the utilization rhythm of each habitual anchorage, and the more coordinated the overall operation of the anchorage.

[0193] S5 standardizes the indicators obtained from S3 and S4, and then uses cluster analysis to evaluate the operational type, obtaining the operational type of each target anchorage, including:

[0194] S51, Data Standardization:

[0195] S3 obtained the anchorage time occupancy rate for each target anchorage. Spatial Gini coefficient and average anchorage time The average state transition rate of each target anchorage obtained from S4 and spatiotemporal efficiency heterogeneity They have different dimensions and orders of magnitude. For example, Expressed as a percentage (0-100%). Expressed in hours, while Expressed as times / day.

[0196] If cluster analysis is performed directly, the larger-scale indicators will dominate the clustering results, leading to analytical distortion. Therefore, we first perform Z-score standardization on each indicator to transform them to the same order of magnitude, obtaining standardized results. Five-dimensional data set.

[0197] S52 uses an unsupervised clustering algorithm for cluster analysis:

[0198] In this invention, the unsupervised clustering algorithm can be any one of K-means clustering, hierarchical clustering, and DBSCAN clustering. In this embodiment, the K-means unsupervised clustering algorithm is used to cluster the standardized... Cluster analysis was performed on the five-dimensional data to... The target anchor points are aggregated into K classes, and the cluster centers of each class are obtained. Each cluster center is represented as a five-dimensional vector. ,in, This represents the anchorage time occupancy rate component. For the spatial Gini coefficient component, This is the average anchorage duration component. For the average state transition rate component, For the spatiotemporal efficiency heterogeneity component; specifically:

[0199] Randomly select K initial cluster centers;

[0200] Calculate the Euclidean distance from the five-dimensional data of each target anchorage to each cluster center, and assign it to the category of the nearest cluster center;

[0201] Recalculate the cluster center for each category (the corresponding five-dimensional vector contains the average value of each indicator of all target anchorages within the category).

[0202] Repeat the above steps until the cluster centers no longer change or the preset number of iterations is reached.

[0203] In this embodiment, K=4, which is confirmed by the following method:

[0204] (1) Algorithm metrics evaluation:

[0205] The optimal number of clusters K is determined by combining the elbow rule and the silhouette coefficient method. In this embodiment, the elbow rule shows that the sum of squared errors (SSE) curve within groups shows a clear inflection point when K=4; similarly, the silhouette coefficient method shows that the average silhouette coefficient is the largest when K=4, indicating that the four-class division has the best clustering effect at the data level.

[0206] (2) Multi-scenario verification:

[0207] Cross-validation was conducted using multiple sets of anchorage area and anchoring behavior data across various scenarios. Specifically, multiple anchorage sample sets were selected from different port areas, and cluster analysis was performed on each set, comparing three schemes with K=3, K=4, and K=5. Validation results show that:

[0208] When K=3, the differences in indicators within the category increase significantly, making it impossible to distinguish between the two completely different operating states of "high load-high turnover" and "high load-low turnover", thus losing the diagnostic significance of this invention;

[0209] When K=5, multiple sets of data show that a certain category contains only a single anchorage, and the difference between this category and the neighboring categories in the mean of each indicator is small. This is an over-segmentation, which lacks statistical support and practical significance in operation and management.

[0210] When K=4, multiple sets of data exhibit a stable four-group structure, with consistent internal indicator characteristics within each category, clear distinctions between categories, and good reproducibility of classification results across different datasets.

[0211] (3) Hierarchical clustering cross-validation:

[0212] Ward hierarchical clustering was used to cluster the same standardized dataset, and the correlation coefficient for the same phenotype was calculated. The results show that the dendrogram of hierarchical clustering has the largest inter-class distance at the four-class partition (i.e., the distance increase is the largest when the four classes are merged in the dendrogram). Its grouping results are basically consistent with the K-means clustering results. The correlation coefficient for the same phenotype is 0.6515, which is greater than the acceptable threshold of 0.6, proving that the four-class partition has good structural stability.

[0213] Based on the analysis of the above three aspects, this invention determines the optimal number of clusters K=4.

[0214] S53, based on the cluster analysis results, implements an operational type assessment that reflects the spatiotemporal utilization efficiency of the anchorage, specifically based on the five-dimensional vector corresponding to the cluster centers. The target anchorage is determined to be either a high-pressure saturation type, a high-efficiency circulation type, a silent reserve type, or a batch processing operation type.

[0215] Since the mean of each element after Z-score standardization is 0, if a component in the five-dimensional vector is greater than 0, the corresponding indicator is higher than the sample average level and is denoted as "high"; if it is less than 0, it is lower than the average level and is denoted as "low". It should be noted that, based on tests with multiple sets of data, the four clustering results obtained from different sets of data have the same data characteristics and correspond to the same four operational types. The actual operational characteristics of the four types of anchorages are clearly distinguishable and consistent with port management experience. Specifically:

[0216] (1) High-pressure saturation type: high ,high ,high ,Low ,Low

[0217] If the components of the five-dimensional vector of the cluster center satisfy: , , , , If so, all target anchorages within this category are determined to be of the high-pressure saturation type.

[0218] The anchorage has been operating under high load for a long time (high) Space occupancy is highly concentrated in a few grids (high) ), long ship stay time (high) The anchor position has a low state transition frequency (low) The states of each anchorage are highly similar (low) (This) lacks flexibility in resource allocation.

[0219] With high load, long stay, low turnover, and high homogeneity, this type of anchorage faces extremely high congestion risks and operational pressures.

[0220] (2) High-efficiency cyclic type: high ,Low ,Low ,high ,Low

[0221] If the components of the five-dimensional vector of the cluster center satisfy: , , , , If so, all target anchorages within this category are determined to be of the efficient cyclic type.

[0222] Although the anchorage has a high load (high) However, due to the relatively balanced use of space (low ), fast ship turnaround speed (low) Frequent anchorage status changes (high) The states of each anchorage are highly similar (low) ).

[0223] High load, high turnover, and balanced use indicate that resource utilization is healthy and saturated, rather than rigid and congested. It is in a healthy state of efficient operation, which is the optimal operating state.

[0224] (3) Silent reserve type: low ,Low ,Low ,Low ,Low

[0225] If the components of the five-dimensional vector of the cluster center satisfy: , , , , If so, all target anchorages within this category are determined to be silent reserve type.

[0226] The overall load of the anchorage is low (low) ), sparse space occupancy (low) ), short ship stay time (low) Anchorage status transition is inactive (low) The status of each anchorage is highly similar (low) It exhibits a pulse-like usage pattern of being generally idle and occasionally activated.

[0227] With low load, low activity, and high homogeneity, the anchorages are in a state of inactivity. The main problem with this type of anchorage is the conflict between idle resources and emergency support functions.

[0228] (4) Batch processing operation type: high ,high ,high

[0229] If the components of the five-dimensional vector of the cluster center satisfy: , , If so, then all target anchors within this category are determined to be batch processing operations.

[0230] The overall load of the anchorage is high ( High), but the internal spacetime utilization is extremely uneven, occupying localized concentrated areas ( High), some anchor positions change status frequently while others remain almost stationary ( high).

[0231] High load, strong imbalance, and segmented use: This type of anchorage is highly correlated with the arrival of cargo in batches and concentrated operation mode. The anchorage serves specific ship types that arrive in batches, exhibiting a "tidal" usage characteristic.

[0232] Compared to evaluation methods that rely solely on time occupancy rate as a single indicator, this invention introduces a dynamic stability index, enabling the differentiation of anchorages with the same static load but different intrinsic operating states. Specifically:

[0233] (1) It reveals the phenomenon of "load-activity" decoupling:

[0234] This invention utilizes time occupancy rate (Static load) and mean state transition rate The joint analysis of (dynamic activity) revealed a "decoupling" phenomenon: high anchorage time occupancy does not necessarily lead to low dynamic activity.

[0235] Take two typical anchorages as examples:

[0236] Anchorage A (high-efficiency circulation type) maintained a high utilization rate of 86.09% and achieved the highest average state transition rate of 1.6505 times per day, indicating that efficient scheduling and management and a moderate average mooring time (37.5 hours) can achieve "healthy circulation" under high load.

[0237] Anchorage B (high-pressure saturation type) has a similar load (84.6%), but its average state transition rate is only 0.5222 times per day and its average anchorage time is as long as 83.6 hours, which is in a state of "rigid congestion".

[0238] High static load does not necessarily lead to congestion. By optimizing operational efficiency (such as timely vessel arrivals and dynamic anchorage allocation), high resource mobility and flexibility can be maintained even with high utilization rates. Two anchorages with similar time occupancy rates can be compared using the average state transition rate of this invention. It can reveal their essential differences and has high... The value of the anchorage indicates an efficient turnover state, while having low The value of the anchorage indicates a low activity state.

[0239] (2) Accurately and quantitatively characterize internal imbalances:

[0240] Through spatiotemporal efficiency heterogeneity With spatial Gini coefficient Combined analysis can identify structural imbalances in anchorage usage. With Gao The combination precisely reflects the operational mode of "high-intensity occupation of some anchorages and long-term idleness of the rest." For example, batch-processing type anchorages... =0.6293 and The coupling characteristic of 0.2235 accurately captures its "tidal" usage characteristic.

Claims

1. A time-geography-based anchorage space-time utilization efficiency evaluation method, characterized in that, Includes the following steps: S1, acquire raw AIS data covering multiple target anchorages of the target port; preprocess the raw AIS data; use a density-based dwell point identification algorithm to extract anchorage status data of each target anchorage from the preprocessed AIS data; the anchorage status data is the data of the trajectory segment of the ship moving at a low speed continuously within the target anchorage waters and within the ship's anchorage area for a preset duration. S2, based on the anchorage status data, extract the habitual anchorage of each target anchorage, specifically including: S21, divide the target anchorage into multiple grid cells of equal area, and for each grid cell, calculate its cumulative anchorage time within the statistical period; S22, calculate the global statistics and the sum of local anchorage times for each grid cell; S23, calculate the Getis-Ord Gi* statistic and identify anchorage hotspots, and use these hotspots as candidate habitual anchorages; S24, calculate the directional distribution ellipse of each anchorage hotspot area, determine the typical anchorage radius, and generate a circular buffer zone for each candidate habitual anchorage as the final spatial range of that candidate habitual anchorage. S25, filter the candidate habitual anchors to obtain the final set of habitual anchors; S3, respectively calculate the static feature index of the space-time utilization efficiency of each target anchorage, including anchorage time occupancy rate , space Gini coefficient and average anchoring time ; S4, calculate the dynamic stability index of the spatiotemporal utilization efficiency of each target anchorage, including the average state transition rate. and spatiotemporal efficiency heterogeneity ; S5 standardizes the indicators obtained from S3 and S4, and then uses cluster analysis to evaluate the operation type and obtain the operation type of each target anchorage.

2. The method for evaluating the spatiotemporal utilization efficiency of anchorages according to claim 1, characterized in that, In S1: The preprocessing includes deleting duplicate records, removing invalid data whose latitude and longitude exceed the target anchorage waters or whose speed is abnormal, and segmenting the trajectory based on the vessel identification code; The method for extracting the anchorage status data of each target anchorage is as follows: Set speed threshold Distance threshold and time threshold The speed threshold, distance threshold, and time threshold are set according to the distribution of vessel types and water conditions at the target anchorage; For consecutive trajectory points of the same vessel, if the following conditions are met within a preset time window: The ship's average speed or track point speed is less than the speed threshold. ; The spatial movement distance, maximum distribution radius, or positional dispersion of the trajectory points are less than a distance threshold. ; The duration of the trajectory segment is greater than the time threshold. ; The data for the corresponding trajectory segment will then be determined as the anchoring status data.

3. The method for evaluating the spatiotemporal utilization efficiency of anchorages according to claim 2, characterized in that, In S21: The grid unit is The area of ​​each grid cell is determined comprehensively based on the total area of ​​the anchorage, the type of anchorage, and the required calculation accuracy. For each grid cell Statistics on its statistical period Cumulative anchorage time The calculation formula is: ， in, For grid cells The total number of anchoring incidents that occurred within the area. For the first occurrence within this grid cell The duration of each anchoring event is obtained from the anchoring status data.

4. The method for evaluating the spatiotemporal utilization efficiency of anchorages according to claim 3, characterized in that, In S22: Define grid cells and Spatial weights are The spatial weights are calculated using any one of the following methods: inverse distance weighting, K-nearest neighbor weighting, or fixed distance bandwidth. The global statistics include the arithmetic mean of the cumulative anchorage duration of all grid cells. and standard deviation The calculation formulas are as follows: ， ， The total duration of the local anchorage For grid cells The weighted sum of the cumulative anchorage times of all other grid cells is calculated using the following formula: ， in, For grid cells In the statistical period The cumulative anchorage time within the area.

5. The method for evaluating the spatiotemporal utilization efficiency of anchorages according to claim 4, characterized in that, In S23: For each grid cell Its Getis-Ord Gi* statistic The calculation formula is: ， For each grid cell The corresponding significance probability value The calculation formula is: ， in, The cumulative distribution function of the standard normal distribution; Set significance level The value of is obtained, and the corresponding critical Z value is acquired; when Greater than the critical Z value and Less than the significance level When the value is [value], the grid cell The anchorage duration within its spatial neighborhood is significantly higher than the global average. When Greater than the critical Z value Less than the significance level Adjacent grid cells that are both spatially connected are aggregated into the same mooring hotspot region. Each mooring hotspot region corresponds to a candidate habitual anchorage, thus obtaining all candidate habitual anchorages in each target anchorage.

6. The method for evaluating the spatiotemporal utilization efficiency of anchorages according to claim 5, characterized in that, In S24: The center of the directional distribution ellipse is used as the estimated center of the corresponding candidate habitual anchor position, and the coordinates of the estimated center are... The calculation formula is: ， ， in, This represents the total number of grid cells within the anchorage hotspot area. For each grid cell within the hotspot area The center coordinates; Typical anchoring radius The determination process is as follows: Calculate the reference radius : ， in, To design the overall length of the ship, For water depth, the corresponding information is obtained from the port design specifications based on the designed vessel type for each anchorage. and Range of values; Typical anchorage radius exist Based on the theoretical range, increase by 20%; The circular buffer zone is centered on the estimated center and extends to the typical anchorage radius. Generates the radius.

7. The method for evaluating the spatiotemporal utilization efficiency of anchorages according to claim 6, characterized in that, In S3: The anchorage time occupancy rate The calculation formula is: ， in, This represents the total duration of all anchoring events occurring in the anchorage within the statistical period. For the first The actual anchoring duration of this anchoring event This represents the total number of anchoring events that occurred in the anchorage during the statistical period. For the design capacity of the anchorage, The total duration of the statistical period; The spatial Gini coefficient The calculation process is as follows: First, the cumulative anchorage durations of the grid cells obtained in S21 are sorted from smallest to largest, resulting in an ordered sequence of the cumulative anchorage durations of each grid cell. ,in, The cumulative anchorage duration is the minimum anchorage duration for the grid cell. The cumulative anchorage time is the longest anchorage time for the grid cell with the longest anchorage time. The first number after sorting by anchorage duration in ascending order. Anchorage duration for each grid cell; Then, the sorted sequence is substituted into the spatial Gini coefficient formula for calculation: ， The average anchorage duration The calculation formula is: 。 8. The method for evaluating the spatiotemporal utilization efficiency of anchorages according to claim 7, characterized in that, S4 includes the following steps: Constructing a state sequence: Collecting each habitual anchor point within a statistical period All anchoring events within the timeline are recorded and arranged in chronological order; events with overlapping times or very short intervals are merged to form a continuous state sequence; by traversing the entire timeline, the state of the habitual anchorage at each moment is marked: 0 indicates that it is idle and not occupied; 1 indicates that it is occupied and a ship is anchored. The number of state changes in the state sequence is counted, including the number of events from "0→1" and "1→0", and recorded as the state transition count. ;in, For the habitual anchorage numbering, , The total number of habitual anchorages within the target anchorage; Calculate the state transition frequency : ， Calculate the average state transition rate : ， Computational spatiotemporal efficiency heterogeneity : 。 9. The method for evaluating the spatiotemporal utilization efficiency of anchorages according to claim 8, characterized in that, S5 includes the following steps: Z-score standardization is performed on each metric to transform them to the same order of magnitude, resulting in a standardized metric. Five-dimensional data set The number of target anchorages, ; Unsupervised clustering algorithm is used to standardize the data. Cluster analysis was performed on the five-dimensional data to... The target anchor points are aggregated into four categories, and the cluster centers of each category are obtained, represented as a five-dimensional vector. ,in, This represents the anchorage time occupancy rate component. For the spatial Gini coefficient component, This is the average anchorage duration component. For the average state transition rate component, For spatiotemporal efficiency heterogeneity components; According to the five-dimensional vector The target anchorage is determined to be either a high-pressure saturation type, a high-efficiency circulation type, a silent reserve type, or a batch processing operation type.

10. The method for evaluating the spatiotemporal utilization efficiency of anchorages according to claim 9, characterized in that, In S5: If the components of the five-dimensional vector of the cluster center satisfy: , , , , If so, all target anchorages within this category are determined to be high-pressure saturation type. These anchorages operate under high load for a long time, with space occupancy highly concentrated in a few grids. Ships stay for long periods, the frequency of anchorage state transitions is low, and the states of each anchorage are highly similar, lacking flexible allocation space. The characteristics of this type of anchorage are high load, long stay, low turnover, and high homogeneity, facing extremely high congestion risks and operational pressures. If the components of the five-dimensional vector of the cluster center satisfy: , , , , Then all target anchorages within this category are determined to be of the high-efficiency circulation type. Although this type of anchorage has a high load, due to the relatively balanced space utilization, fast ship turnaround speed, and frequent anchorage status changes, the status of each anchorage is highly similar. The characteristics of this type of anchorage are high load, high turnover, and balanced use. The resource utilization shows a healthy saturation characteristic, rather than rigid congestion. It is in a healthy state of efficient operation and is the optimal operating state. If the components of the five-dimensional vector of the cluster center satisfy: , , , , If all target anchorages within this category are classified as silent reserve anchorages, these anchorages have low overall load, sparse space occupancy, short vessel dwell time, inactive anchorage status transitions, and highly similar statuses across anchorages, exhibiting a pulse-like usage pattern of general idleness and occasional activation. These anchorages are characterized by low load, low activity, and high homogeneity, and are in a state of waiting to be activated. The main problem with these anchorages is the conflict between resource idleness and emergency support functions. If the components of the five-dimensional vector of the cluster center satisfy: , , If all target anchorages within this category are determined to be batch operation type, the overall load of this type of anchorage is high, but the internal space and time utilization is extremely uneven, the occupation is concentrated in some areas, some anchorage status changes frequently while others are almost immobile; the characteristics of this type of anchorage are high load, strong imbalance, segmented use, and highly correlated with the batch arrival of cargo and concentrated operation mode. The anchorage serves specific ship types that arrive in batches, exhibiting "tidal" usage characteristics.

Images