Intelligent e-commerce supply chain logistics tracking analysis method and system

Abstract

This application provides a smart e-commerce supply chain logistics tracking and analysis method and system, belonging to the field of smart e-commerce technology. The method includes: analyzing the timeliness deviation distribution of each path based on initial path optimization conditions; determining the adjustment direction of orders regarding transit paths based on the timeliness deviation distribution of each path; determining a node combination scheme for fragmented distribution optimization; identifying progressively increasing transit nodes based on path optimization conditions; statistically analyzing the changing trend of dwell time fragmentation by calculating the proportion of dwell time at each level; evaluating the effect of node combination on improving timeliness achievement rate; obtaining a preliminary evaluation result on the effectiveness of node combination; statistically analyzing the dwell time distribution of each node after optimization based on the target node combination scheme; analyzing the coverage of the window start and end times and dwell time; and determining the delivery time based on the matching degree between the coverage and the promised window.

Description

Technical Field

[0001] This invention relates to the field of smart e-commerce technology, and in particular to a smart e-commerce supply chain logistics tracking and analysis method and system. Background Technology

[0002] In smart e-commerce logistics, supply chain tracking and timeliness management directly determine platform reputation and user experience. Timeliness achievement rate has become a core indicator affecting repurchase rate and complaint volume. Traditional logistics design always pursues the fewest transit nodes to compress total transportation time, believing that the shorter the route and the fewer the stops, the easier it is to deliver within the promised time. However, the actual distribution of promised delivery time windows for orders is extremely uneven. Some orders require next-day delivery, while others can be as lenient as three days, and even the time windows for the same batch of orders may differ by several hours. This dispersed nature of time windows leads to an implicit contradiction: simply reducing transit nodes, while shortening the theoretical total time, can easily cause a large number of orders to miss their delivery window. For example, a next-day delivery order, if delivered directly, might arrive at the terminal at 3 AM, but the terminal doesn't start delivery until 9 AM, resulting in the order waiting for six hours and still being considered overdue. Conversely, if a nighttime transit is added, allowing the goods to arrive at the terminal around 10 PM, it can catch the first wave of deliveries the next morning, successfully hitting the time window.

[0003] A deeper problem lies in the fact that logistics tracking systems can only record the actual arrival and departure times of orders at each node, as well as the specific dwell time at each node. However, they cannot dynamically determine whether adding or removing a transit node will allow more orders to fall within their respective commitment windows based on the fragmented distribution of these dwell times. Existing systems often only focus on whether the total duration meets the target, completely ignoring the impact of the fragmented nature of the dwell time after being divided by different nodes on time window matching. This results in route selection being limited to fixed rules, making it impossible to flexibly adjust to the time window distribution of order groups.

[0004] Therefore, how to obtain the arrival time and dwell time of each transit node in real time during the logistics tracking process, and dynamically select the node combination path that can maximize the timeliness achievement rate based on the fragmentation of the dwell time and the actual distribution of the order commitment time window, has become a key issue in improving the timeliness management level of smart e-commerce logistics. Summary of the Invention

[0005] On the one hand, this invention provides a smart e-commerce supply chain logistics tracking and analysis method, including:

[0006] By obtaining the order arrival time and dwell time of each transit node through the data interface of the logistics tracking platform, the start and end time of the promised delivery window for each order is determined, and the dwell fragmentation distribution of each order is determined based on the start and end time of the promised delivery window to obtain the initial path optimization conditions.

[0007] Based on the initial route optimization conditions, analyze the timeliness deviation distribution of each transit route, determine the adjustment direction of orders regarding transit routes based on the timeliness deviation distribution, and determine the node combination scheme for fragmented distribution optimization.

[0008] The order flow of each path is simulated by the node combination scheme to obtain whether each order hits the promised delivery time window. The hit results are compared with historical timeliness data to evaluate the improvement of the timeliness achievement rate. The path optimization conditions are determined based on the improvement.

[0009] Based on the path optimization conditions, progressively increasing transit nodes are identified, the dwell time ratio of each level of node is statistically analyzed, the changing trend of dwell fragmentation is analyzed, the effect of node combination on improving timeliness is evaluated, and a preliminary evaluation result of the effectiveness of node combination is obtained.

[0010] Candidate schemes are screened based on the preliminary evaluation results of the effectiveness of the node combination, the arrival time and the start and end time of the promised delivery window are obtained from the actual logistics tracking data, the degree of consistency between the order hitting the promised delivery time window and the actual operation is evaluated, and the target node combination scheme is determined based on the degree of consistency.

[0011] Based on the statistical optimization of the target node combination scheme and the distribution of dwell time of each node, the coverage of the promised delivery window start and end time and dwell time is analyzed, and the delivery timeliness is determined according to the degree of matching between the coverage and the promised window.

[0012] Furthermore, the process involves obtaining the order arrival time and dwell time at each transit node through the data interface of the logistics tracking platform, determining the start and end times of the promised delivery window for each order, and determining the fragmented distribution of dwell time for each order based on the promised delivery window start and end times to obtain initial path optimization conditions, including:

[0013] The arrival time of orders at each transit node is obtained through the data interface of the logistics tracking platform, the dwell time of each order at the transit node is recorded, and the start and end times of the corresponding promised delivery window are queried according to the order type.

[0014] The order dwell time percentage at each transit node is calculated based on the start and end times of the promised delivery window. Nodes with a dwell time percentage exceeding a preset threshold are marked as long-dwelling nodes. The ratio of the number of long-dwelling nodes to the total number of nodes is calculated to obtain the fragmented distribution characteristics of the dwell time.

[0015] The order grouping vector is constructed using the fragmented distribution characteristics of the stay. The window width and the proportion of stay time are used as input dimensions. The orders are grouped by K-means clustering algorithm. The initial path optimization conditions corresponding to orders with different window widths are determined according to the distance between the cluster center and the order.

[0016] Furthermore, based on the initial path optimization conditions, the timeliness deviation distribution of each transit path is analyzed. Based on this timeliness deviation distribution, the adjustment direction of orders regarding transit paths is determined, and a node combination scheme for fragmented distribution optimization is determined, including:

[0017] Based on the initial route selection conditions, historical delivery data for each transit route is obtained, the difference between the actual delivery time and the promised delivery time for each order on each route is calculated, the distribution frequency of the difference in different time periods is statistically analyzed, the order backlog during the time period when the difference exceeds a preset threshold is recorded, and the timeliness deviation distribution of each route is obtained.

[0018] The congested nodes in the route are identified by the timeliness deviation distribution. The ratio of the order processing volume of the node to the designed capacity of the node is calculated to determine the capacity saturation. Based on the capacity saturation, it is determined whether to bypass the node or add a parallel transfer node to obtain the adjustment direction of the transfer route.

[0019] Based on the adjustment direction, a candidate set of node combinations is constructed. The candidate set is iteratively optimized using a genetic algorithm. The order flow time and node dwell time dispersion between nodes are used as inputs to the fitness function. The timeliness achievement rate under different node combinations is calculated, and the combination with the highest timeliness achievement rate is selected as the node combination scheme for fragmented distribution optimization.

[0020] Furthermore, by simulating the order flow of each path through the node combination scheme, it is determined whether each order hits the promised delivery time window. The hit results are compared with historical timeliness data to evaluate the improvement in timeliness achievement rate. Based on the improvement, path optimization conditions are determined, including:

[0021] The node combination scheme is used to construct a simulation environment for order flow, generate an order arrival sequence based on historical order arrival patterns, simulate the queuing and processing of orders at each transit node, and record the actual delivery time of each order.

[0022] The actual delivery time is matched with the start and end time of the promised delivery window of the order, and the proportion of orders that hit the window is counted as the simulated timeliness achievement rate. This is compared with the actual timeliness achievement rate in the same period of history, and the difference between the two is calculated as the timeliness improvement rate.

[0023] The routes are sorted according to the improvement in timeliness, and the routes with the highest improvement are selected. The route selection criteria are determined based on the range of the transportation characteristic parameters of the routes.

[0024] Furthermore, based on the path optimization conditions, progressively increasing transit nodes are identified, the dwell time ratio of each node is statistically analyzed, the changing trend of dwell fragmentation is analyzed, and the effect of node combination on improving timeliness is evaluated, resulting in a preliminary assessment of the effectiveness of node combination, including:

[0025] Based on the route optimization conditions, transfer nodes are added step by step from the original route. The ratio of the dwell time at each node to the total delivery time is calculated to obtain the dwell time percentage data of each level of node.

[0026] The ratio of the standard deviation to the mean of the dwell time percentage data is used as the coefficient of variation. By comparing the changes in the coefficient of variation under different numbers of nodes, the trend of dwell time fragmentation is determined.

[0027] Based on the correlation coefficient between the changing trend and the actual timeliness achievement rate, a preliminary assessment result of the effectiveness of the node combination is obtained.

[0028] Furthermore, candidate solutions are screened based on the preliminary evaluation results of the effectiveness of the node combination, the arrival time and the start and end times of the promised delivery window are obtained from the actual logistics tracking data, the degree of consistency between the order hitting the promised delivery time window and the actual operation is evaluated, and the target node combination solution is determined based on the degree of consistency, including:

[0029] The candidate solutions are ranked based on the preliminary evaluation results of the effectiveness of the node combination, and the solutions whose preliminary evaluation results exceed a preset threshold are selected as the filtered candidate solutions.

[0030] Obtain the arrival time record of each order at each node from the actual logistics tracking data, extract the start and end times of the promised delivery window for the order, and calculate the proportion of orders whose actual delivery time falls within the start and end time range of the window.

[0031] The ratio is compared with the timeliness achievement rate obtained from the previous simulation. The candidate schemes are then sorted a second time according to the magnitude of the deviation. The scheme with the highest consistency and the smallest deviation is selected to determine the target node combination scheme.

[0032] Furthermore, based on the statistical optimization of the dwell time distribution of each node according to the target node combination scheme, the coverage range of the promised delivery window start and end times and dwell time is analyzed. The delivery timeliness is determined based on the matching degree between the coverage range and the promised window, including:

[0033] Based on the optimized target node combination scheme, the dwell time of each node is statistically analyzed, and the frequency distribution of dwell time is calculated by time period.

[0034] The dwell time at each node is sequentially added up to the order start time to calculate the estimated delivery time;

[0035] The percentage of orders whose expected delivery time falls within the promised delivery window's start and end time range is statistically analyzed, and this percentage is used as a delivery timeliness indicator.

[0036] Secondly, this invention discloses a smart e-commerce supply chain logistics tracking and analysis system, the system comprising:

[0037] The data acquisition and window determination module is used to acquire the order arrival time and dwell time of each transit node through the data interface of the logistics tracking platform, determine the start and end time of the promised delivery window for each order, determine the dwell fragmentation distribution of each order based on the promised delivery window start and end time, and obtain the initial path optimization conditions.

[0038] The fragmentation analysis and adjustment direction determination module is used to analyze the timeliness deviation distribution of each transit path based on the initial path optimization conditions, determine the adjustment direction of the order with respect to the transit path based on the timeliness deviation distribution, and determine the node combination scheme for fragmentation distribution optimization.

[0039] The circulation simulation and timeliness assessment module is used to simulate the order circulation of each path through the node combination scheme, obtain whether each order hits the promised delivery time window, compare the hit results with historical timeliness data, evaluate the improvement of timeliness achievement rate, and determine the path optimization conditions based on the improvement.

[0040] The node combination effect evaluation module is used to identify the progressively increasing transit nodes according to the path optimization conditions, count the proportion of dwell time at each level of node, analyze the changing trend of dwell fragmentation, evaluate the effect of node combination on improving timeliness, and obtain a preliminary evaluation result of the effectiveness of node combination.

[0041] The scheme screening and matching degree evaluation module is used to screen candidate schemes based on the preliminary evaluation results of the effectiveness of the node combination, obtain the arrival time and the start and end time of the promised delivery window in the actual logistics tracking data, evaluate the matching degree between the order hitting the promised delivery time window and the actual operation, and determine the target node combination scheme based on the matching degree.

[0042] The timeliness determination module is used to statistically optimize the dwell time distribution of each node according to the target node combination scheme, analyze the coverage of the start and end time of the promised delivery window and the dwell time, and determine the delivery timeliness according to the degree of matching between the coverage and the promised window.

[0043] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0044] This invention discloses a smart e-commerce supply chain logistics tracking and analysis method. It acquires order arrival times and dwell times at each transit node from a logistics tracking platform through a node acquisition module, determines the start and end times of the promised delivery window, and analyzes the fragmented distribution of dwell time to form initial route optimization conditions. Based on the initial conditions, it evaluates the timeliness deviation distribution of each route, determines the direction of transit route adjustments, and generates node combination schemes. By simulating order flow and comparing it with historical timeliness data, it evaluates the improvement in timeliness achievement rate to optimize the route optimization conditions. It further identifies progressively increasing transit nodes, analyzes the changing trend of dwell time fragmentation, and verifies the consistency of orders hitting the promised window using actual logistics tracking data, selecting and determining target node combination schemes. Finally, it statistically analyzes the matching degree between the optimized dwell time distribution and the promised window to accurately determine the delivery time. This invention effectively solves the timeliness deviation problem caused by fragmented dwell time in logistics transit, achieving a significant improvement in route optimization and timeliness achievement rate through data-driven node combination optimization and simulation verification. Attached Figure Description

[0045] Figure 1 This is a flowchart of a smart e-commerce supply chain logistics tracking and analysis method according to the present invention.

[0046] Figure 2 This is a schematic diagram of a smart e-commerce supply chain logistics tracking and analysis method according to the present invention.

[0047] Figure 3 This is a schematic diagram of the structure of a smart e-commerce supply chain logistics tracking and analysis system according to the present invention. Detailed Implementation

[0048] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0049] like Figures 1-3 This embodiment of a smart e-commerce supply chain logistics tracking and analysis method and system may specifically include:

[0050] S101. Obtain the order arrival time and dwell time of each transit node through the data interface of the logistics tracking platform, determine the start and end time of the order's promised delivery window, determine the dwell fragmentation distribution of each order based on the order's promised delivery window, and obtain the initial path optimization conditions.

[0051] The arrival times of orders at each transit node are obtained through the data interface of the logistics tracking platform. The dwell time of each order at each transit node is recorded. The start and end times of the promised delivery window are queried according to the order type, and the time difference between the start and end times of the window is obtained. The dwell time percentage of each order at each node is calculated based on the time difference. If the dwell time percentage exceeds a preset threshold, the node is marked as a long-dwelling node. The ratio of the number of long-dwelling nodes to the total number of nodes is calculated to obtain the dwell fragmentation distribution characteristics. The dwell fragmentation distribution characteristics are used to construct an order grouping vector. The window width and dwell time percentage are used as input dimensions. The K-means clustering algorithm is used to group the orders. The initial path optimization conditions for orders with different window widths are determined based on the distance between the cluster center and the order.

[0052] In one embodiment, the logistics tracking platform records order flow information in real time using scanning devices deployed at various transit nodes. When an order package passes through a sorting center, delivery station, or transfer hub, the scanning device automatically reads the barcode information on the electronic waybill and uploads the arrival time to a central database. The arrival time is stored in a unified timestamp format, accurate to the second.

[0053] Specifically, the dwell time is obtained directly from platform data, which includes the departure and arrival times of each node. When an order leaves node A and arrives at node B, the difference between the departure time of node A and the arrival time of node B is the dwell time at node A. The promised delivery window start and end times are automatically determined based on the delivery service type selected by the user when placing the order. For next-day delivery orders, the window starts at 9:00 AM the following day and ends at 9:00 PM the following day.

[0054] It should be noted that the fragmented distribution of dwell time reflects the degree of dispersion of order time consumption along the entire delivery route. When the dwell time at a certain transit node exceeds a preset threshold, it indicates that there is a time backlog at that node, which will lead to a reduction in the processing time of subsequent nodes.

[0055] For example, if an order spends 8 hours at a regional sorting center, accounting for 40% of the total delivery time, this excessive concentration of dwell time will cause fragmentation of the delivery rhythm. By statistically analyzing the ratio of the number of long-stay nodes to the total number of nodes, we can quantify the uniformity of time distribution along the entire path. The higher the ratio, the more uneven the time distribution and the more severe the fragmentation.

[0056] In one possible implementation, the K-means clustering algorithm uses window width and dwell time percentage as a two-dimensional input vector to automatically group massive orders. The window width refers to the start and end time difference of the promised delivery window, and the dwell time percentage is a sequence of ratios of dwell time at each node to the total delivery time, with the average of this sequence used as the second dimension. The algorithm first randomly initializes K cluster centers, then iteratively calculates the Euclidean distance from each order to each cluster center, assigning the order to the nearest cluster, and then recalculates the center position of each cluster until convergence.

[0057] Preferably, when determining the optimal path based on the distance between the cluster center and the order, orders with shorter distances use a direct path, while orders with longer distances have additional transit nodes. The timeliness requirements of different types of orders are matched by adjusting the number of transits.

[0058] S102. Based on the initial path optimization conditions, the timeliness deviation distribution of each path is obtained. Based on the timeliness deviation distribution of each path, the adjustment direction of the order regarding the transit path is determined, and the node combination scheme for fragmented distribution optimization is determined.

[0059] Historical delivery data for each transit path is obtained based on the initial path optimization conditions. The difference between the actual delivery time and the promised delivery time for each order on each path is calculated. The frequency of the difference distribution in different time periods is statistically analyzed. If the difference exceeds a preset threshold, the order backlog for that time period is recorded, resulting in the timeliness deviation distribution for each path. The timeliness deviation distribution is used to identify congested nodes in the path. The capacity saturation is determined by calculating the ratio of the node's order processing volume to its designed capacity. Based on the capacity saturation, it is determined whether orders should bypass the node or add parallel transit nodes, thus obtaining the adjustment direction for the transit path. A candidate set of node combinations is constructed based on the adjustment direction. The candidate set is iteratively optimized using a genetic algorithm. The timeliness achievement rate under different node combinations is calculated using the dispersion of order flow time and node dwell time as inputs to the fitness function. The combination with the highest timeliness achievement rate is selected. The transit node sequence of orders is redistributed through the combination. The change in order backlog at each node after adjustment is monitored. If the backlog decreases and the dwell time distribution becomes more uniform, the combination is adopted, determining the node combination scheme for fragmented distribution optimization.

[0060] In one embodiment, the actual delivery time records of all orders on each transit route are obtained by retrieving historical delivery data from the logistics tracking platform for the past three months. The timeliness deviation is obtained by calculating the difference between the actual delivery time and the promised delivery time; a positive difference indicates delayed delivery, while a negative difference indicates early delivery.

[0061] Specifically, the frequency of timeliness deviations for each path is statistically analyzed hourly to form a 24-hour deviation distribution map. When more than 30% of the orders have a deviation value exceeding a preset threshold of 2 hours within a certain period, the order backlog for that period is recorded, which is the difference between the total number of orders to be processed within that period and the processing capacity of the bottleneck node in the path.

[0062] It should be noted that congested nodes are identified based on abnormal peak values ​​in the timeliness deviation distribution. When a transit node experiences consecutive deviation peaks within a specific time period, and the peak height exceeds 1.5 times the average deviation, the node is marked as a congested node. The capacity saturation is calculated by dividing the node's order processing volume by its designed capacity. Order processing volume refers to the number of orders actually processed by the node per unit time, and designed capacity refers to the node's maximum processing capacity under standard operating conditions. A capacity saturation exceeding 0.8 indicates the node is operating under high load, while exceeding 1.0 indicates order backlog. Based on the capacity saturation range, corresponding adjustment directions are determined: when saturation is between 0.8 and 1.0, temporary increases in processing resources are considered; when exceeding 1.0, orders need to bypass the node or additional parallel transit nodes are added for diversion. Based on these adjustment directions, when constructing a candidate set of node combinations, each candidate scheme includes permutations and combinations of existing and newly added nodes.

[0063] For example, the implementation process of the genetic algorithm in node combination optimization begins with population initialization. The initial population contains 50 individuals, each representing a node combination scheme, with permutation encoding representing the selected nodes and their order. The fitness function is designed as a weighted combination of timeliness achievement rate and cost, where the weight of timeliness achievement rate is 0.7 and the weight of cost is 0.3. The inputs to the fitness function include two key variables: order flow time between nodes and node dwell time dispersion. The order flow time between nodes is obtained by dividing the geographical distance between adjacent nodes by the transportation speed, and the dwell time dispersion is measured by the standard deviation of the dwell time of each node; the smaller the standard deviation, the more uniform the time distribution. The cost is obtained by multiplying the number of activated nodes by the unit cost of 50. The selection operation uses a roulette wheel method, where individuals with higher fitness have a greater probability of being selected. The crossover operation uses single-point crossover, randomly selecting a crossover point and exchanging gene segments of two parent individuals. The mutation operation randomly flips a gene position with a probability of 0.01. After 100 generations of evolution, the individual with the highest fitness is selected as the optimal node combination scheme.

[0064] In one possible implementation, the timeliness achievement rate in the fitness function is obtained through simulation. For each node combination scheme, the delivery process of 1000 orders under that combination is simulated, and the proportion of orders delivered within the promised time window is statistically analyzed. The smaller the dispersion of node dwell time, the more balanced the waiting time of orders across nodes, avoiding cascading delays caused by excessive backlog at a single node.

[0065] Preferably, the calculation of the timeliness achievement rate also takes into account the weight difference of different time periods. The achievement rate weight is set to 1.5 during peak periods, 1.0 during off-peak periods, and 0.8 during low-peak periods. This differentiated weight setting more realistically reflects the timeliness requirements in actual operation.

[0066] Understandably, when monitoring the effectiveness of adjustments, two key indicators are considered: the trend of order backlog and the uniformity of dwell time distribution. When the order backlog continues to decrease and the coefficient of variation of dwell time at each node is less than 0.3, the node combination scheme is considered to have achieved the goal of fragmented distribution optimization, where the coefficient of variation CV = σ / μ, σ is the standard deviation, and μ is the mean.

[0067] S103. Simulate the order flow of each path through the node combination scheme, simulate the order hit commitment delivery time window, compare the order hit commitment delivery time window with historical timeliness data, evaluate the improvement of timeliness achievement rate, and determine the path optimization conditions based on the improvement.

[0068] A simulation environment for order flow is constructed using a node combination scheme. The processing capacity and transportation time parameters of each node are read, and an order arrival sequence is generated based on historical order arrival patterns. The queuing and processing of orders at each transit node are simulated, and the actual delivery time of each order is recorded to obtain order flow time-series data. This order flow time-series data is matched with the promised delivery window of each order. If the actual delivery time falls within the start and end time range of the window, it is marked as a hit. The hit rate of all orders is calculated as the simulated timeliness achievement rate, and the actual timeliness achievement rate of the same period in history is obtained as benchmark data. The difference between the simulated timeliness achievement rate and the historical timeliness achievement rate is calculated based on the benchmark data to determine the percentage improvement in timeliness. If the improvement exceeds a preset threshold, the transportation characteristic parameters of that path are recorded to evaluate the degree of timeliness improvement for different paths. Paths are ranked according to the degree of timeliness improvement, and the paths with the highest improvement are selected. Based on the range of the transportation characteristic parameters of the paths, optimal path selection conditions are determined.

[0069] In one embodiment, the simulation environment is constructed based on the principle of discrete event simulation, simulating the dynamic flow of orders in the logistics network through an event-driven mechanism. The simulation environment includes four core components: node attribute configuration, transportation network topology, order generator, and event scheduler. These components work together to achieve a digital reproduction of real logistics scenarios.

[0070] It is important to note that the generation pattern of order arrival sequences directly affects the accuracy of the simulation results. Analysis of three months of historical order data revealed a clear time-based pattern in order arrivals: 35% of orders were received during the morning peak (8-10 AM), 28% during the evening peak (7-9 PM), and the remaining time periods showed a relatively even distribution. Based on this pattern, a segmented modeling method was used to construct the order generator. During peak hours, the order arrival interval follows a negative exponential distribution with a mean of 30 seconds; during off-peak hours, the mean is adjusted to 90 seconds; and during low-peak hours, it is 180 seconds. Each order's attributes include order number, order placement time, promised delivery window, cargo weight, and delivery priority. The promised delivery window is determined based on the user's selected service type: next-day delivery is from 9:00 AM to 9:00 PM the following day, standard delivery is within 72 hours, and economy delivery is 5-7 business days. The order generator continuously generates order events according to the set arrival pattern and adds them to the event queue for processing.

[0071] Specifically, the processing capacity parameters of each node are calibrated based on actual operational data. The standard processing capacity of the regional sorting center is 8,000 pieces per hour, the delivery station is 500 pieces per hour, and the last-mile delivery point is 200 pieces per hour. The node processing time has a non-linear relationship with the current load rate. When the load rate L is below 0.7, the processing time remains at the standard time T0; when it exceeds 0.7, the processing time T = T0 * (1 + 0.15 * floor((L - 0.7) / 0.1)), where T0 is the standard processing time, L is the current load rate, and floor is the floor function.

[0072] In one possible implementation, order queuing and processing follows a first-in, first-out (FIFO) principle. When an order arrives at a node, the current queue length of that node is checked first. If the queue is not full, the order is added to the waiting queue; if the queue is full, an overflow handling mechanism is triggered, forwarding the order to a backup node or a temporary buffer for processing. Nodes process orders in the queue one by one according to their processing capacity, and after processing, the order is sent to the next node. The duration of the transportation process is determined based on the distance between nodes and the mode of transportation; the speed is set at 60 km / h for road transportation, 80 km / h for rail transportation, and 800 km / h for air transportation.

[0073] Preferably, time window matching is achieved by comparing the actual delivery time of an order with the promised window. If the actual delivery time falls within the window range, the order is marked as a hit, and the ratio of the number of hit orders to the total number of orders is the simulated timeliness achievement rate.

[0074] For example, the benchmark data is obtained by extracting the actual timeliness achievement rate of the same route during the same period from historical operation records, and selecting the average value of the most recent 30 days as the comparison benchmark, excluding the influence of special circumstances such as abnormal weather and holidays.

[0075] Furthermore, the improvement in timeliness is calculated using a relative improvement rate. If the simulated timeliness achievement rate is 85% and the historical benchmark is 75%, then the improvement is approximately (85%-75%) / 75% ≈ 13.3%. When the improvement exceeds a preset threshold of 10%, it indicates that the node combination scheme has a significant improvement effect. The transportation characteristic parameters include the average number of transfers, the average dwell time at each node, the combination of transportation modes, and the total route length. These parameters are used together for the comparison and evaluation of route schemes.

[0076] Understandably, by ranking the timeliness improvements of different routes, the top 20% of routes are selected as the preferred route set. The distribution of transportation characteristic parameters of these routes is analyzed, and statistical analysis methods are used to calculate the mean and variance of each parameter. Clustering algorithms are then used to identify similar feature groups, and common features are extracted as the basis for determining route selection criteria. Route selection criteria refer to a route with an average of 3-4 transfers, an average stop time at nodes not exceeding 4 hours, and at least one segment of rail transport.

[0077] For example, when a path meets the above conditions, the path meets the preferred condition.

[0078] S104. Identify the progressively increasing transit nodes based on the route optimization conditions, and analyze the changing trend of the fragmentation of the stay by statistically analyzing the proportion of the stay time at each level. Evaluate the effect of node combination on improving the timeliness achievement rate and obtain preliminary evaluation results on the effectiveness of node combination.

[0079] Based on the route optimization criteria, transit nodes are added level by level from the original route. The route structure after each added node is recorded, and the ratio of dwell time at each node to the total delivery time is calculated to obtain the dwell time percentage data for each level of node. The ratio of the standard deviation to the mean of the dwell time percentage data is used as the coefficient of variation. If the coefficient of variation exceeds a preset threshold, it is determined to be highly fragmented. By comparing the changes in the coefficient of variation under different numbers of nodes, the trend of dwell fragmentation is determined. Based on the trend, the correlation coefficient between the trend and the actual timeliness achievement rate is calculated. When the correlation coefficient is positive and exceeds the preset threshold, it indicates that the node combination has an effect on improving the timeliness achievement rate, and a preliminary evaluation result on the effectiveness of the node combination is obtained.

[0080] In one embodiment, when identifying additional transit nodes based on route optimization conditions, starting from the original two-node direct route, geographical location and capacity distribution information are obtained from the system database, and regional sorting centers, secondary transfer centers, and delivery stations are added sequentially according to this information.

[0081] Specifically, each additional node is considered a new level. The dwell time of all nodes on the path is recalculated, and the percentage of dwell time at each node relative to the total delivery time is calculated to form a dwell time percentage dataset for that level.

[0082] It should be noted that the coefficient of variation (CV), as a relative indicator of data dispersion, is used in the logistics field to assess the uniformity of time distribution. The CV is calculated by dividing the standard deviation of the dwell time percentage data by its mean, specifically as CV = SD / Mean, where SD is the standard deviation and Mean is the mean. When the dwell time percentages of each node are similar, the standard deviation is small, and the CV is close to 0, indicating a uniform time distribution. When the dwell time percentages of some nodes are particularly high or low, the standard deviation increases, and the CV increases accordingly. In practical applications, based on statistical experience, a CV below 0.3 indicates a relatively uniform distribution, 0.3 to 0.6 indicates moderate fragmentation, and above 0.6 indicates high fragmentation, where the degree of fragmentation refers to the unevenness of the time distribution. By comparing the CVs under different numbers of nodes, it was found that as the number of nodes increases progressively, the CV shows a significant decreasing trend, reflecting the improvement in the degree of fragmentation.

[0083] Preferably, when the number of nodes increases to more than 50, the rate of decrease in the coefficient of variation is less than 5% and tends to level off. At this point, continuing to add nodes has limited effect on improving the degree of fragmentation, but will instead increase management complexity and operating costs.

[0084] In one possible implementation, the correlation coefficient is calculated using the Pearson correlation coefficient method. First, the on-time completion rate sequence refers to the proportion of logistics tasks completed on time under different numbers of nodes, and its data comes from simulation experiment results. Correlation analysis is then performed between the coefficient of variation sequences under different numbers of nodes and the corresponding on-time completion rate sequences to obtain the degree of linear correlation between the two. The specific calculation formula is as follows: , where x i The coefficient of variation is the sequence value, y i μ is the timeliness achievement rate sequence value. x and μ y These are the means of the two sequences, respectively.

[0085] For example, the degree of fragmentation is measured by the coefficient of variation, and the on-time delivery rate is the proportion of orders delivered on time. Since a smaller coefficient of variation indicates a lower degree of fragmentation, and a higher on-time delivery rate indicates a better delivery performance, when the coefficient of variation and the on-time delivery rate show a negative correlation, i.e., r is negative and |r|>0.6, it indicates that reducing the degree of fragmentation can effectively improve the on-time delivery rate. The node combination scheme has a significant effect on improving delivery timeliness, forming a preliminary assessment result of the effectiveness of the node combination.

[0086] S105. Screen candidate schemes based on the preliminary evaluation results of the effectiveness of node combination, obtain the arrival time and window start and end time in the actual logistics tracking data, evaluate the consistency between the order hit the promised delivery time window and the actual operation, and determine the target node combination scheme based on the consistency.

[0087] Candidate schemes are ranked based on the preliminary evaluation results of node combination effectiveness. Schemes with preliminary evaluation results exceeding a preset threshold are selected. The arrival time records of each order at each node are obtained from actual logistics tracking data. The start and end times of the promised delivery window for each order are extracted to obtain the actual operational dataset. Based on the actual operational dataset, it is calculated whether the actual delivery time of the order falls within the window start and end time range. The proportion of orders hitting the window is counted to the total number of orders. This proportion is compared with the timeliness achievement rate obtained from previous simulations. If the deviation is within a preset threshold range, the consistency is considered satisfactory. The candidate schemes are then ranked a second time using the consistency score. The scheme with the highest consistency score and the smallest deviation from the simulation prediction is selected to determine the target node combination scheme.

[0088] In one embodiment, the preliminary evaluation results include a comprehensive score for each candidate solution across three dimensions: expected improvement in timeliness, node utilization, and cost increment. The comprehensive score is calculated using a weighted average, with the expected improvement in timeliness having a weight of 0.5, node utilization at 0.3, and cost increment at 0.2. The data for each dimension are normalized and then summed. Solutions with a comprehensive score greater than 0.7 are selected for subsequent practical verification. Actual logistics tracking data is obtained through the logistics information interface of the e-commerce platform. First, the start and end times of the promised delivery window are read from the order master data table. Then, each tracking record contains four fields: order number, node code, scan time, and operation type. The arrival time is extracted from records with the operation type "arrived," forming a complete actual operation dataset.

[0089] It should be noted that the calculation of the consistency rate involves comparing and verifying the actual operating results with the preliminary simulation predictions. By statistically analyzing the actual delivery status of all orders within a month, the actual timeliness achievement rate is calculated, which is the percentage of orders delivered within the promised time window. The actual timeliness achievement rate is compared with the predicted achievement rate obtained from the simulation; the difference reflects the accuracy of the model's prediction. If the absolute value of the difference is less than 10%, the consistency rate is considered satisfactory; less than 5% is considered good, 5%-10% is considered moderate, and more than 10% is considered unsatisfactory. The consistency rate not only considers the deviation of the overall achievement rate but also performs detailed analysis by time period. The deviation of the achievement rate for the morning peak, evening peak, and off-peak periods is calculated separately, and a weighted average is used to obtain the comprehensive consistency rate index. The weight settings are dynamically adjusted according to the order volume proportion of each time period, with higher order volume periods receiving higher weights. The specific calculation is as follows: Where D is the overall deviation, d1, d2, and d3 are the deviations of the three time periods, and w1, w2, and w3 are the corresponding weights (w1+w2+w3=1, calculated based on the order volume ratio).

[0090] Preferably, the comparison process refers to the process of comparing and analyzing the actual delivery route with the planned route. This process also introduces an abnormal order identification mechanism, which identifies and removes abnormal delayed orders by judging whether the order delay time exceeds a preset threshold T (T is 2 hours) and whether it is affected by uncontrollable factors such as extreme weather or traffic control, thereby improving the accuracy of the consistency assessment.

[0091] For example, when determining the target node combination scheme, multiple candidate schemes are first generated based on the node attribute matching algorithm and ranked from high to low according to the matching score; then, a cost-benefit analysis is performed on the top three candidate schemes, including calculating the incremental operating cost C (unit: yuan) and the timeliness improvement rate E (unit: percentage) of each scheme. By comparing the C / E ratio, the scheme with the smallest ratio and a matching degree of not less than 80% is selected as the final implementation scheme to achieve a balance between timeliness improvement and cost control.

[0092] S106. Based on the statistical optimization of the target node combination scheme, analyze the distribution of dwell time of each node, analyze the coverage of the window start and end time and dwell time, and determine the delivery time based on the matching degree between the coverage and the promised window.

[0093] Based on the optimized destination node combination scheme, the dwell time of each node is statistically analyzed. The frequency distribution of the dwell time is calculated by dividing the time into hourly periods, resulting in dwell time distribution data. Using this dwell time distribution data, the dwell time of each node on the path is summed sequentially and added to the order start time to calculate the estimated delivery time. If the estimated delivery time falls within the start and end time range of the commitment window, it is considered a match. The matching ratio is then used as a delivery timeliness indicator.

[0094] In one embodiment, dividing the day into hourly time periods means dividing a 24-hour day into 24 time periods, and counting the frequency of dwell time at each transit node within each time period to form a frequency distribution table of dwell time. The average dwell time T at each transit node is calculated using this table, where T is equal to the sum of the times of each dwell time multiplied by the corresponding frequencies and divided by the total frequency. This value is used for subsequent calculations of the actual dwell time and the estimated delivery time.

[0095] Specifically, the cumulative calculation starts from the order placement time and sequentially adds the actual dwell time at each transit node. The actual dwell time is obtained by calculating the average value from historical data; for example, the historical average dwell time at the first node is 3 hours, the second is 5 hours, and the third is 2 hours. If the order is placed at 10:00 AM, the preliminary estimated time after accumulation is 8:00 PM that day. The estimated delivery time is combined with the actual delivery patterns in business operations. Delivery patterns refer to delivery time patterns based on historical data, and the time window limit for last-mile delivery refers to the deliverable period from 8:00 AM to 6:00 PM daily. The process is that if the preliminary estimate exceeds the window, it is postponed to the beginning of the next window.

[0096] It should be noted that the start and end times of the commitment window are determined based on the delivery service selected by the user. The window for next-day delivery is from 9:00 to 21:00 the next day. When the estimated delivery time falls within this range, the order is marked as a match.

[0097] Preferably, the test orders are derived from the historical order data mentioned above, and simulated orders are generated through random sampling for testing. The number of matches for all test orders is counted, and the matching ratio is calculated as a quantitative indicator of delivery timeliness. A higher matching ratio indicates a stronger timeliness guarantee capability of the node combination scheme.

[0098] This invention provides a smart e-commerce supply chain logistics tracking and analysis system, mainly comprising:

[0099] The data acquisition and window determination module is used to acquire the order arrival time and dwell time of each transit node through the data interface of the logistics tracking platform, determine the start and end time of the order's promised delivery window, determine the dwell fragmentation distribution of each order based on the start and end time of the order's promised delivery window, and obtain the initial path optimization conditions.

[0100] The fragmentation analysis and adjustment direction determination module is used to analyze the timeliness deviation distribution of each path based on the initial path optimization conditions, determine the adjustment direction of the order regarding the transit path based on the timeliness deviation distribution of each path, and determine the node combination scheme for fragmentation distribution optimization.

[0101] The circulation simulation and timeliness assessment module is used to simulate the order circulation of each path through node combination schemes, simulate the order hit commitment delivery time window, compare the order hit commitment delivery time window with historical timeliness data, evaluate the improvement of timeliness achievement rate, and determine the path optimization conditions based on the improvement.

[0102] The node combination effect evaluation module is used to identify the progressively increasing transit nodes based on the route optimization conditions, and to analyze the changing trend of the fragmentation of the stay by statistically analyzing the proportion of the stay time at each level, evaluate the effect of node combination on improving the timeliness achievement rate, and obtain preliminary evaluation results on the effectiveness of node combination.

[0103] The scheme screening and matching evaluation module is used to screen candidate schemes based on the preliminary evaluation results of the effectiveness of node combination, obtain the arrival time and window start and end time in the actual logistics tracking data, evaluate the matching degree between the order hit the promised delivery time window and the actual operation, and determine the target node combination scheme based on the matching degree.

[0104] The timeliness determination module is used to statistically analyze the distribution of dwell time of each node after optimization based on the target node combination scheme, analyze the coverage of the window start and end time and dwell time, and determine the delivery timeliness based on the degree of matching between the coverage and the promised window.

[0105] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

Claims

1. A smart e-commerce supply chain logistics tracking and analysis method, characterized in that, include: By obtaining the order arrival time and dwell time of each transit node through the data interface of the logistics tracking platform, the start and end time of the promised delivery window for each order is determined, and the dwell fragmentation distribution of each order is determined based on the start and end time of the promised delivery window to obtain the initial path optimization conditions. Based on the initial route optimization conditions, analyze the timeliness deviation distribution of each transit route, determine the adjustment direction of orders regarding transit routes based on the timeliness deviation distribution, and determine the node combination scheme for fragmented distribution optimization. The order flow of each path is simulated by the node combination scheme to obtain whether each order hits the promised delivery time window. The hit results are compared with historical timeliness data to evaluate the improvement of the timeliness achievement rate. The path optimization conditions are determined based on the improvement. Based on the path optimization conditions, progressively increasing transit nodes are identified, the dwell time ratio of each level of node is statistically analyzed, the changing trend of dwell fragmentation is analyzed, the effect of node combination on improving timeliness is evaluated, and a preliminary evaluation result of the effectiveness of node combination is obtained. Candidate schemes are screened based on the preliminary evaluation results of the effectiveness of the node combination, the arrival time and the start and end time of the promised delivery window are obtained from the actual logistics tracking data, the degree of consistency between the order hitting the promised delivery time window and the actual operation is evaluated, and the target node combination scheme is determined based on the degree of consistency. Based on the statistical optimization of the target node combination scheme and the distribution of dwell time of each node, the coverage of the promised delivery window start and end time and dwell time is analyzed, and the delivery timeliness is determined according to the degree of matching between the coverage and the promised window.

2. The intelligent e-commerce supply chain logistics tracking and analysis method according to claim 1, characterized in that, The process involves obtaining the order arrival time and dwell time at each transit node through the data interface of the logistics tracking platform, determining the start and end times of the promised delivery window for each order, determining the fragmented distribution of dwell time for each order based on the promised delivery window start and end times, and obtaining initial route optimization conditions, including: The arrival time of orders at each transit node is obtained through the data interface of the logistics tracking platform, the dwell time of each order at the transit node is recorded, and the start and end times of the corresponding promised delivery window are queried according to the order type. The order dwell time percentage at each transit node is calculated based on the start and end times of the promised delivery window. Nodes with a dwell time percentage exceeding a preset threshold are marked as long-dwelling nodes. The ratio of the number of long-dwelling nodes to the total number of nodes is calculated to obtain the fragmented distribution characteristics of the dwell time. The order grouping vector is constructed using the fragmented distribution characteristics of the stay. The window width and the proportion of stay time are used as input dimensions. The orders are grouped by K-means clustering algorithm. The initial path optimization conditions corresponding to orders with different window widths are determined according to the distance between the cluster center and the order.

3. The intelligent e-commerce supply chain logistics tracking and analysis method according to claim 1, characterized in that, Based on the initial route optimization conditions, the timeliness deviation distribution of each transit route is analyzed. Based on this timeliness deviation distribution, the adjustment direction of orders regarding transit routes is determined, and a node combination scheme for fragmented distribution optimization is determined, including: Based on the initial route selection conditions, historical delivery data for each transit route is obtained, the difference between the actual delivery time and the promised delivery time for each order on each route is calculated, the distribution frequency of the difference in different time periods is statistically analyzed, the order backlog during the time period when the difference exceeds a preset threshold is recorded, and the timeliness deviation distribution of each route is obtained. The congested nodes in the route are identified by the timeliness deviation distribution. The ratio of the order processing volume of the node to the designed capacity of the node is calculated to determine the capacity saturation. Based on the capacity saturation, it is determined whether to bypass the node or add a parallel transfer node to obtain the adjustment direction of the transfer route. Based on the adjustment direction, a candidate set of node combinations is constructed. The candidate set is iteratively optimized using a genetic algorithm. The order flow time and node dwell time dispersion between nodes are used as inputs to the fitness function. The timeliness achievement rate under different node combinations is calculated, and the combination with the highest timeliness achievement rate is selected as the node combination scheme for fragmented distribution optimization.

4. The intelligent e-commerce supply chain logistics tracking and analysis method according to claim 1, characterized in that, The node combination scheme is used to simulate the order flow of each path, and to determine whether each order hits the promised delivery time window. The hit results are compared with historical timeliness data to evaluate the improvement in timeliness achievement rate. Based on the improvement, path optimization conditions are determined, including: The node combination scheme is used to construct a simulation environment for order flow, generate an order arrival sequence based on historical order arrival patterns, simulate the queuing and processing of orders at each transit node, and record the actual delivery time of each order. The actual delivery time is matched with the start and end time of the promised delivery window of the order, and the proportion of orders that hit the window is counted as the simulated timeliness achievement rate. This is compared with the actual timeliness achievement rate in the same period of history, and the difference between the two is calculated as the timeliness improvement rate. The routes are sorted according to the improvement in timeliness, and the routes with the highest improvement are selected. The route selection criteria are determined based on the range of the transportation characteristic parameters of the routes.

5. The intelligent e-commerce supply chain logistics tracking and analysis method according to claim 1, characterized in that, Based on the path optimization conditions, progressively increasing transit nodes are identified, the dwell time ratio of each node is statistically analyzed, the changing trend of dwell fragmentation is examined, and the effect of node combination on improving timeliness is evaluated. Preliminary evaluation results of the effectiveness of node combination are obtained, including: Based on the route optimization conditions, transfer nodes are added step by step from the original route. The ratio of the dwell time at each node to the total delivery time is calculated to obtain the dwell time percentage data of each level of node. The ratio of the standard deviation to the mean of the dwell time percentage data is used as the coefficient of variation. By comparing the changes in the coefficient of variation under different numbers of nodes, the trend of dwell time fragmentation is determined. Based on the correlation coefficient between the changing trend and the actual timeliness achievement rate, a preliminary assessment result of the effectiveness of the node combination is obtained.

6. The intelligent e-commerce supply chain logistics tracking and analysis method according to claim 1, characterized in that, Candidate solutions are screened based on the preliminary evaluation results of the effectiveness of the node combination. The arrival time and promised delivery window start and end times from the actual logistics tracking data are obtained. The degree of consistency between the order hitting the promised delivery time window and the actual operation is evaluated. Based on the degree of consistency, the target node combination solution is determined, including: The candidate solutions are ranked based on the preliminary evaluation results of the effectiveness of the node combination, and the solutions whose preliminary evaluation results exceed a preset threshold are selected as the filtered candidate solutions. Obtain the arrival time record of each order at each node from the actual logistics tracking data, extract the start and end times of the promised delivery window for the order, and calculate the proportion of orders whose actual delivery time falls within the start and end time range of the window. The ratio is compared with the timeliness achievement rate obtained from the previous simulation. The candidate schemes are then sorted a second time according to the magnitude of the deviation. The scheme with the highest consistency and the smallest deviation is selected to determine the target node combination scheme.

7. The intelligent e-commerce supply chain logistics tracking and analysis method according to claim 1, characterized in that, Based on the statistical optimization of the target node combination scheme, the distribution of dwell time for each node is statistically analyzed. The coverage range of the promised delivery window start and end times and dwell time is analyzed. The delivery timeliness is determined based on the degree of matching between the coverage range and the promised window, including: Based on the optimized target node combination scheme, the dwell time of each node is statistically analyzed, and the frequency distribution of dwell time is calculated by time period. The dwell time at each node is sequentially added up to the order start time to calculate the estimated delivery time; The percentage of orders whose expected delivery time falls within the promised delivery window's start and end time range is statistically analyzed, and this percentage is used as a delivery timeliness indicator.

8. A smart e-commerce supply chain logistics tracking and analysis system, characterized in that, The system includes: The data acquisition and window determination module is used to acquire the order arrival time and dwell time of each transit node through the data interface of the logistics tracking platform, determine the start and end time of the promised delivery window for each order, determine the dwell fragmentation distribution of each order based on the promised delivery window start and end time, and obtain the initial path optimization conditions. The fragmentation analysis and adjustment direction determination module is used to analyze the timeliness deviation distribution of each transit path based on the initial path optimization conditions, determine the adjustment direction of the order with respect to the transit path based on the timeliness deviation distribution, and determine the node combination scheme for fragmentation distribution optimization. The circulation simulation and timeliness assessment module is used to simulate the order circulation of each path through the node combination scheme, obtain whether each order hits the promised delivery time window, compare the hit results with historical timeliness data, evaluate the improvement of timeliness achievement rate, and determine the path optimization conditions based on the improvement. The node combination effect evaluation module is used to identify the progressively increasing transit nodes according to the path optimization conditions, count the proportion of dwell time at each level of node, analyze the changing trend of dwell fragmentation, evaluate the effect of node combination on improving timeliness, and obtain a preliminary evaluation result of the effectiveness of node combination. The scheme screening and matching degree evaluation module is used to screen candidate schemes based on the preliminary evaluation results of the effectiveness of the node combination, obtain the arrival time and the start and end time of the promised delivery window in the actual logistics tracking data, evaluate the matching degree between the order hitting the promised delivery time window and the actual operation, and determine the target node combination scheme based on the matching degree. The timeliness determination module is used to statistically optimize the dwell time distribution of each node according to the target node combination scheme, analyze the coverage of the start and end time of the promised delivery window and the dwell time, and determine the delivery timeliness according to the degree of matching between the coverage and the promised window.

Images