A flower rental delivery system based on big data

Abstract

This invention relates to the field of big data processing and analysis technology, and in particular to a big data-based flower rental delivery system. The system includes: real-time monitoring of multi-source environmental data and biomechanical indicators to accurately quantify flower freshness, providing a reliable basis for delivery decisions; dynamic prediction of flower survival time and risk level classification to prioritize rental demand and effectively reduce the risk of return losses; optimization of delivery routes by integrating urgency and real-time traffic data to minimize flower freshness loss while ensuring timeliness; dynamic adjustment of route weights by responding to vehicle environmental anomalies and traffic events in real time to ensure delivery reliability in complex scenarios; and continuous optimization of model parameters through freshness feedback and execution data to form an adaptive improvement mechanism and improve the long-term efficiency of the delivery system.

Description

Technical Field

[0001] This invention relates to the field of big data processing and analysis technology, and in particular to a big data-based flower rental delivery system. Background Technology

[0002] Current flower delivery management still faces the following problems: In traditional flower rental, the freshness assessment of flowers often relies on manual inspection, which is highly subjective and lacks timeliness, making it difficult to accurately reflect the actual condition of the flowers in real time; the lack of an effective survival time prediction mechanism leads to premature wilting of flowers due to improper storage during the rental period, increasing the return damage rate and operating costs; uncertainties such as urban traffic congestion and bumpy roads make delivery time uncontrollable, resulting in a high rate of flower loss in transit; delivery vehicles may encounter various environmental disturbances (such as sudden temperature changes and vibrations) during travel, and traditional systems lack effective response mechanisms, leading to flower damage. Furthermore, traditional systems often lack continuous optimization mechanisms after deployment, making it difficult to adapt to market changes and the diversity of customer needs. Therefore, this invention proposes a flower rental delivery system based on big data. Summary of the Invention

[0003] The purpose of this invention is to solve the problems in the background art by proposing a big data-based flower rental delivery system.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] A big data-based flower rental delivery system includes: an environmental data freshness analysis module, a survival time prediction and grading module, a dynamic path preservation decision module, a path anti-disturbance optimization and adjustment module, and a path decision iterative update module.

[0006] Environmental data freshness analysis module: Real-time collection of multi-source environmental data and flower physiological indicators from rental terminals to generate flower freshness feature vectors;

[0007] Survival duration prediction and grading module: Dynamically predicts the equivalent survival duration of flowers based on flower freshness feature vectors, and generates grading labels for the urgency of rental demand;

[0008] Dynamic route preservation decision module: Integrates urgency labels with real-time urban traffic heat maps to construct a preservation weighted route decision model;

[0009] Path disturbance rejection optimization and adjustment module: Based on the status feedback of delivery vehicles, dynamically adjust the delivery strategy and generate disturbance rejection optimized paths;

[0010] Route decision iterative update module: Iteratively updates the decision model parameters based on rental terminal freshness feedback and route execution data.

[0011] Furthermore, the environmental data freshness analysis module collects multi-source environmental data and flower physiological indicators from the rental terminal in real time, and the process of generating flower freshness feature vectors includes:

[0012] Real-time data collection from the sensor array deployed at the rental terminal, and simultaneous acquisition of flower image recognition data;

[0013] The time-series data streams generated by the multi-source sensors of the rental terminal are analyzed, and non-steady-state environmental pulses are separated as environmental stress factors through signal processing technology; among them, non-steady-state environmental pulses include temperature change curves, cumulative irradiance, and vibration spectrum peaks;

[0014] Biomechanical response indicators were quantified through petal edge morphology analysis and stem mechanical deformation analysis.

[0015] The extracted environmental stress factors and biomechanical response indices are aligned within a spatiotemporal grid to construct a stress-response correlation tensor.

[0016] Based on the constructed stress-response correlation tensor, tensor decomposition technology is used to extract cross-modal coupling features; with freshness decay acceleration as the core indicator, the cross-modal coupling features are weighted and fused to generate a flower freshness feature vector containing geographic location tags and timestamps.

[0017] To address the noise interference generated during sensor acquisition, an adversarial generative network is used to simulate the sensor noise distribution characteristics; through adversarial training in the feature space, perturbation components with low correlation to flower freshness decay are filtered out.

[0018] The optimized flower freshness feature vectors are encapsulated according to the rental terminal ID to form a data packet with associated weights; the final output contains a set of flower freshness feature vectors for all rental terminals. .

[0019] Furthermore, the survival duration prediction and grading module dynamically predicts the equivalent survival duration of flowers based on flower freshness feature vectors, and generates rental demand urgency grading labels through the following process:

[0020] Obtain the feature vector set of flower freshness ;

[0021] Transfer learning is used to map freshness decay to equivalent survival time. Using transfer learning techniques, a historical lease termination damage rate sample database is loaded; based on this database, a cross-category freshness decay mapping field is constructed; and the obtained flower freshness feature vector set is... Project the data onto the constructed cross-category freshness decay mapping field; through the mapping relationship, convert the freshness decay characteristics into equivalent survival time. The predicted value; simultaneously, based on the flower freshness feature vector set Association weights in This is further converted into confidence weights. Among them, confidence weight Used to calculate the confidence interval of equivalent survival time ;

[0022] Collect data on return damage rates from historical rental orders to reflect the survival rate of flowers during the actual rental process; calculate the remaining rental period for each order. confidence interval of equivalent survival time The overlap ratio of the lower limit, i.e., the urgency coefficient. When the calculated urgency coefficient When the remaining time is less than the standard value for flower survival, it indicates that the remaining time of the order is insufficient to support the predicted survival time of the flowers, at which point a freshness crisis warning sign is activated; based on the urgency coefficient... The value and status of the warning indicator are used to classify the urgency level into three levels: when the urgency coefficient is... Less than the preset urgency threshold When a warning sign is activated, the situation is deemed a crisis and marked as emergency; when the urgency coefficient... Less than the preset urgency threshold And greater than or equal to the preset urgency threshold When the urgency coefficient is high, it is judged as a high-risk state and marked as high priority; when the urgency coefficient is high, it is marked as high priority. Greater than or equal to the preset urgency threshold If the condition is as described above, it is considered a safe state and marked as normal level.

[0023] The urgency level labels are injected into the corresponding order data bodies to generate a set of rental orders with urgency labels. .

[0024] Furthermore, the dynamic route preservation decision module integrates urgency tags with real-time urban traffic heat maps to construct a preservation weighted route decision model. The process includes:

[0025] Get rental order collection Each order in this collection has been labeled with an urgency level, such as urgent, high priority, or normal, based on the predicted survival time of the flowers.

[0026] Analyze the rental order set The urgency label in the data is converted into a transportation time penalty coefficient α.

[0027] Based on the flower category information in the order, query the corresponding turbulence damage sensitivity β;

[0028] The system integrates real-time urban traffic heat map data, which reflects the real-time traffic conditions of the road network; it abstracts the dynamic parameters of the real-time urban traffic heat map into a set of quantifiable and operable traffic influencing factors; and it encodes the travel time and bump index of road network nodes into a spatiotemporal loss matrix.

[0029] For each order, a freshness loss function is defined, which comprehensively analyzes the impact of transportation time penalty coefficient, bump damage sensitivity, travel time and bump index on the freshness preservation effect of flowers;

[0030] The objective function is to minimize the total freshness loss of all orders, while also constraining vehicle capacity and time windows. A Pareto evolutionary algorithm is used to solve the problem, and an initial path set is output through algorithm iteration. .

[0031] Furthermore, the process by which the route disturbance rejection optimization adjustment module dynamically adjusts the delivery strategy and generates disturbance rejection optimized routes based on the delivery vehicle status feedback includes:

[0032] Receive initial path set After receiving the initial path set, the vehicle environment offset detection operation is performed: the seismic temperature and humidity data is compared with the preset category preservation threshold corridor in real time; during continuous monitoring, when the seismic temperature and humidity data of N consecutive sampling points deviates from the preset corridor, it indicates that the internal environment of the vehicle can no longer meet the preservation requirements of the goods, and the vehicle is marked as having an abnormal microenvironment.

[0033] For vehicles marked as having abnormal vehicle microenvironment, the correlation coefficient of their undelivered orders is adjusted: the transportation time penalty coefficient α for undelivered orders is multiplied according to the urgency of the orders; for urgent orders, For other order levels, Based on the updated transit time penalty coefficient Reconstruct the preservation loss function;

[0034] While reconstructing the preservation loss function, a traffic incident early warning stream is integrated; combined with traffic incident early warning data, the initial path set is... A comprehensive analysis of the path was conducted to identify potential interruption points within the path;

[0035] After identifying path interruption risk points, dynamic rearrangement of path nodes is triggered under the constraint of the reconstructed preservation loss function. Simultaneously, based on real-time traffic conditions and delivery needs, interruption risk points are dynamically removed, and optimal detour segments are inserted, ultimately generating an optimized path set that is resistant to environmental disturbances. .

[0036] Compared with existing technologies, the beneficial effects of this invention are as follows: By constructing a dynamic freshness monitoring system through multi-dimensional environmental perception and biomechanical response quantification, it can identify environmental stress on flowers, providing reliable data support for subsequent decision-making and effectively avoiding flower quality deterioration caused by environmental out-of-control conditions; Based on cross-category freshness decay mapping using transfer learning, it realizes dynamic prediction and risk classification of flower survival time, enabling early identification of crisis orders and triggering early warning mechanisms, significantly reducing the risk of unplanned lease terminations and improving the level of refined management in the leasing business; By integrating urgency labels and real-time traffic heat maps, it constructs a freshness-weighted path optimization model, which can minimize the freshness loss of all orders under the condition of meeting vehicle capacity and time window constraints, improving delivery efficiency while ensuring the stability of flower quality; Through vehicle environmental deviation detection and traffic event early warning, it dynamically adjusts delivery strategies and generates anti-disturbance optimized paths, effectively responding to the impact of emergencies on the delivery process and ensuring the reliable execution of delivery tasks in complex environments; By establishing a closed-loop optimization mechanism for freshness feedback and path execution data, it continuously iterates the decision model parameters, enabling the system to adapt to changes in the operating environment, forming a virtuous cycle of continuous improvement, and ensuring the cost-effectiveness of delivery services in the long term. Attached Figure Description

[0037] Figure 1 This is a block diagram of a big data-based flower rental delivery system proposed in this invention. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] Reference Figure 1 A big data-based flower rental delivery system includes an environmental data freshness analysis module, a survival time prediction and grading module, a dynamic path preservation decision module, a path anti-disturbance optimization and adjustment module, and a path decision iterative update module.

[0040] Environmental data freshness analysis module: Real-time collection of multi-source environmental data and flower physiological indicators from rental terminals to generate flower freshness feature vectors;

[0041] Survival duration prediction and grading module: Dynamically predicts the equivalent survival duration of flowers based on flower freshness feature vectors, and generates grading labels for the urgency of rental demand;

[0042] Dynamic route preservation decision module: Integrates urgency labels with real-time urban traffic heat maps to construct a preservation weighted route decision model;

[0043] Path disturbance rejection optimization and adjustment module: Based on the status feedback of delivery vehicles, dynamically adjust the delivery strategy and generate disturbance rejection optimized paths;

[0044] Route decision iterative update module: Iteratively updates the decision model parameters based on rental terminal freshness feedback and route execution data.

[0045] It should be further explained that, in the specific implementation process, the environmental data freshness analysis module collects multi-source environmental data and flower physiological indicators from the rental terminal in real time, and the process of generating flower freshness feature vectors includes:

[0046] Real-time data collection from the sensor array deployed at the rental terminal, including environmental parameters such as temperature, humidity, light intensity, and vibration, and simultaneous acquisition of flower image recognition data, to comprehensively perceive the microenvironment in which the flowers are located and their growth status;

[0047] The time-series data streams generated by the multi-source sensors of the rental terminal were analyzed, and non-steady-state environmental pulses were separated as environmental stress factors through signal processing technology. Among them, non-steady-state environmental pulses include temperature change curves (reflecting rapid temperature changes), cumulative light radiation (measuring the cumulative effect of light on flowers), and vibration spectrum peaks (capturing the potential impact of vibration on flowers). These pulse factors are directly related to the environmental stress of flowers.

[0048] Biomechanical response indicators are quantified through petal edge morphological analysis and stem mechanical deformation analysis: image processing technology is used to perform morphological analysis on petal edges and calculate the petal curling entropy value to quantify the morphological changes of petals caused by environmental stress; at the same time, stem mechanical deformation analysis is used to evaluate the fatigue coefficient of stems under environmental stress, reflecting the mechanical response characteristics of flowers; these biomechanical response indicators are important basis for assessing the decline in flower freshness.

[0049] The extracted environmental stress factors and biomechanical response indices are aligned within a spatiotemporal grid, thereby constructing a stress-response correlation tensor (including correlation weights). This tensor not only contains the relationships between the various factors, but also uses correlation weights. This reflects the strength of its impact and provides a rich information basis for subsequent analysis;

[0050] Based on the constructed stress-response correlation tensor, tensor decomposition technology is used to extract cross-modal coupling features, which can comprehensively reflect the complex relationship between environmental stress and the physiological response of flowers. Taking the freshness decay acceleration (freshness change rate per unit time) as the core indicator, the cross-modal coupling features are weighted and fused to generate a flower freshness feature vector containing geographic location tags and timestamps, providing strong support for the dynamic monitoring of flower freshness. The weighted fusion formula is: Freshness feature value = 0.4 × Temperature-curling coupling degree + 0.3 × Light-fatigue coupling degree + 0.3 × Vibration-morphology coupling degree.

[0051] To address the noise interference generated during sensor acquisition, an adversarial generative network is used to simulate the sensor noise distribution characteristics. Through adversarial training in the feature space, perturbation components with low correlation to flower freshness decay are filtered out, thereby improving the accuracy and reliability of the feature vectors.

[0052] The optimized flower freshness feature vector is encapsulated according to the rental terminal ID to form a data packet with associated weights. Each data packet not only contains the flower's freshness feature information under the current environment, but also includes associated weights. This reflects the contribution of each factor to freshness decay; the final output contains a set of flower freshness feature vectors from all rental terminals. This provides data support for the refined management and freshness maintenance of the flower rental business.

[0053] It should be further explained that, in the specific implementation process, the survival time prediction and grading module dynamically predicts the equivalent survival time of flowers based on the flower freshness feature vector, and generates a grading label for the urgency of rental demand, including the following steps:

[0054] Obtain the feature vector set of flower freshness ;

[0055] Transfer learning is used to map freshness decay to equivalent survival time. Using transfer learning techniques, a historical return-of-lease damage rate sample database was loaded, which records the return-of-lease damage of different flower varieties under different freshness conditions. Based on the historical return-of-lease damage rate sample database, a cross-category freshness decay mapping field was constructed, which can reflect the complex relationship between flower freshness decay and equivalent survival time. The obtained flower freshness feature vector set was then used. Project the data onto the constructed cross-category freshness decay mapping field; through the mapping relationship, convert the freshness decay characteristics into equivalent survival time. Predicted values:

[0056] ,

[0057] In the formula, This is the predicted mean for freshness degradation. The standard deviation of freshness degradation prediction. It is the inverse cumulative distribution function of the standard normal distribution. For confidence levels (e.g., 95% confidence intervals) =0.05); at the same time, based on the flower freshness feature vector set Association weights in This is further converted into confidence weights. :

[0058] ,

[0059] In the formula, An adjustment factor (e.g., 0.1) calibrated using historical lease termination data is used to calibrate the matching degree between the association weights and the actual survival status; confidence weights. Used to calculate the confidence interval of equivalent survival time (Non-point estimation) to reflect the uncertainty of the prediction results;

[0060] Collect data on return damage rates from historical rental orders to reflect the survival rate of flowers during the actual rental process; calculate the remaining rental period for each order. confidence interval of equivalent survival time The overlap ratio of the lower limit, i.e., the urgency coefficient:

[0061] ,

[0062] In the formula, This is the urgency coefficient, which measures the urgency relationship between the remaining time of the order and the predicted survival time of the flowers; The remaining lease term for the order; This is the lower bound of the confidence interval for equivalent survival time; This is an urgency adjustment factor (calibrated based on historical lease termination data, e.g., 0.2); when the calculated urgency factor... When the remaining time is less than the standard survival value for flowers, it indicates that the remaining time of the order is insufficient to support the predicted survival time of the flowers. At this time, a freshness crisis warning sign is activated, prompting managers to take emergency measures; based on the urgency coefficient... The value and status of the warning indicator are used to classify the urgency level into three levels: when the urgency coefficient is... Less than the preset urgency threshold When the urgency coefficient (e.g., 0.7) or warning sign is activated, it is considered a crisis and marked as emergency, indicating that a response is required within 4 hours to prevent damage to the flowers or lease cancellation; when the urgency coefficient is... Less than the preset urgency threshold (e.g., 1.2) and greater than or equal to the preset urgency threshold. When the urgency coefficient is 0.7, it is considered a high-risk state and marked as high priority, indicating that it requires immediate response to ensure the survival and quality of the flowers during the remaining lease period; when the urgency coefficient is... Greater than or equal to the preset urgency threshold If it is as shown in 1.2, it is considered a safe state and marked as normal level, indicating that the treatment can be delayed until the next day, and the flowers have a relatively long survival time in the current environment;

[0063] The urgency level labels are injected into the corresponding order data bodies to generate a set of rental orders with urgency labels. This set contains basic information about orders and clarifies the urgency of each order, providing strong support for priority management and resource allocation in the leasing business.

[0064] It should be further explained that, in the specific implementation process, the dynamic route preservation decision module integrates urgency labels with real-time urban traffic heat maps to construct a preservation weighted route decision model, which includes the following steps:

[0065] Get rental order collection Each order in this collection has been labeled with an urgency level, such as urgent, high priority, or normal, based on the predicted survival time of the flowers.

[0066] Analyze the rental order set The urgency label in the order is converted into a delivery time penalty coefficient α. The specific conversion rules are as follows: the delivery time penalty coefficient α for urgent orders is assigned a value of 3, indicating that the order is extremely sensitive to delivery time; the delivery time penalty coefficient α for high-priority orders is assigned a value of 1.5, indicating that the order is relatively sensitive to delivery time; and the delivery time penalty coefficient α for regular orders is assigned a value of 1, indicating that the order is moderately sensitive to delivery time.

[0067] Based on the type of flowers in the order, the corresponding sensitivity to bump damage (β) is determined. For example, cut flowers are extremely sensitive to bumps due to their fragile stems, so β ​​is set to 2.0; while potted plants are less sensitive to bumps due to their stable root systems, so β ​​is set to 0.5.

[0068] This method integrates real-time urban traffic heatmap data, which reflects the real-time traffic conditions of the road network. The dynamic parameters of the urban real-time traffic heatmap, including the coefficient of variation of average vehicle speed per road segment, the time gradient of congestion index, and the topological density of bumpy road segments, are abstracted into a set of quantifiable traffic influencing factors. In the formula, As the first traffic impact factor, As the second traffic impact factor, Let k be the k-th traffic impact factor, where k is the number of traffic impact factors, for example... The coefficient of variation of average vehicle speed, This represents the time gradient of the congestion index, where these factors dynamically reflect the road network's capacity and bumpiness.

[0069] The travel time t and bump index b of the road network nodes are encoded into a spatiotemporal loss matrix:

[0070] ,

[0071] In the formula, Let i be the travel time from road network node i to j; Let b be the bump index from road network node i to j, where the process of obtaining the bump index b is as follows:

[0072] By accessing the real-time vibration time-series stream from the triaxial accelerometer built into the delivery vehicle, parameters including vertical vibration energy are extracted. (Calculation of main frequency band energy using power spectral density) and successive impact events (Pulse count exceeding the threshold acceleration); from the flower freshness feature vector set Extracting the stress fatigue coefficient of the stem Normalize it to a biomechanical damage coefficient : Define the effective turbulence index: Based on the vehicle's historical trajectory, the vibration sequence is mapped to road network nodes to generate a b-distribution. The weighting coefficients for traffic impact factors; This is the bias term (default setting is the zero vector);

[0073] In the process of constructing the spatiotemporal loss matrix, traffic impact factors are used as a basis. The impact on travel time and bump index is expressed as follows: The travel time t and bump index b of each road network node are represented as... For linear combinations of k, if k=3, then: ;

[0074] At the same time, the traffic impact factors will be set. Each factor is bound to an adjustable parameter layer of the decision model according to its physical meaning, so as to facilitate subsequent optimization;

[0075] Define a freshness loss function for each order: The function comprehensively analyzes the effects of transportation time penalty coefficient α, bump damage sensitivity β, travel time t, and bump index b on the preservation effect of flowers; by minimizing the preservation loss function, the transportation route can be optimized and the loss of flowers during transportation can be reduced.

[0076] The objective function is to minimize the total freshness loss of all orders, while also constraining vehicle capacity and time windows. A Pareto evolutionary algorithm is used to solve the problem, and an initial path set is output through algorithm iteration. The specific process is as follows:

[0077] Set the target function:

[0078] ,

[0079] In the formula, For the first The set of edges of a path, For single order indexing, Total number of orders across the entire domain; The safe passage time threshold (e.g., 1.5 times the average time). The penalty coefficient for violations within the time window (e.g., 5); The objective function represents minimizing the total freshness loss of all orders; the vehicle capacity constraint states that the total amount of goods in all orders must not exceed the maximum carrying capacity of the vehicle, specifically:

[0080] ,

[0081] In the formula, For the first The volume or weight of the goods in an order indicates the vehicle space or load capacity required for that order. The maximum capacity limit of the vehicle ensures that the total amount of goods delivered in a single trip does not exceed the vehicle's physical carrying capacity; the time window constraint indicates the first... The service time of each order (the total time from the origin to the destination) must not exceed its allowed time window, specifically:

[0082] ,

[0083] In the formula, For the first The earliest available service time for each order (i.e., the earliest time the vehicle arrives at the origin of the order) is set to prevent vehicles from arriving too early and causing excessively long waiting times (which may be related to the freshness of the flowers). For the first The latest completion time for each order (i.e. the time when the vehicle must leave the destination of the order) is set to ensure that the order is completed within the time limit required by the customer, so as to avoid late fees or loss of flower freshness;

[0084] Initialize the population and generate G initial paths, each containing a randomly selected sequence of nodes; calculate the freshness loss function and time window violation penalty for each path; define a multi-objective fitness function: F = [total freshness loss, total delivery time]; classify paths according to Pareto fronts (e.g., the first front is the optimal solution set); use tournament selection to filter parent individuals, retaining representative solutions in the Pareto front; perform crossover operations by retaining path segments using sequential crossover, and perform node exchange mutation operations by swapping two random nodes, randomly shifting the delivery time within a safety window to achieve time window perturbation mutation; repeat selection, crossover, and mutation operations until convergence condition (e.g., no improvement for 50 consecutive generations), and output the initial path set in the Pareto front. The path set includes a distribution map of preservation costs for each path, which visually demonstrates the impact of different paths on the preservation effect of flowers.

[0085] It should be further explained that, in the specific implementation process, the route disturbance rejection optimization adjustment module dynamically adjusts the delivery strategy and generates disturbance rejection optimized routes based on the delivery vehicle status feedback.

[0086] Receive initial path set After receiving the initial path set, the vehicle environment offset detection operation is performed: the seismic temperature and humidity data is compared with the preset category preservation threshold corridor in real time; during continuous monitoring, when the seismic temperature and humidity data of N consecutive (e.g., 3) sampling points deviate from the preset corridor, it indicates that the internal environment of the vehicle can no longer meet the preservation requirements of the goods, and the vehicle is marked as having an abnormal microenvironment.

[0087] For vehicles marked as having abnormal vehicle microenvironment, the correlation coefficient of their undelivered orders is adjusted: the transportation time penalty coefficient α for undelivered orders is multiplied according to the urgency of the orders; for urgent orders, For other order levels, Understandably, adjusting the transportation time penalty coefficient α can more accurately reflect the weight of freshness loss for orders with different levels of urgency under abnormal conditions, providing key parameters for subsequent reconstruction of the freshness loss function; based on the updated transportation time penalty coefficient... Reconstruct the preservation loss function:

[0088] ,

[0089] In the formula, The reconstructed preservation loss function, The time-sensitive decay constant, The threshold for bump damage is defined; among them, the freshness loss function is an important indicator for measuring the degree of quality damage caused by environmental and other factors during the delivery process, and its accuracy directly affects the effectiveness of route optimization; by adjusting the transportation time penalty coefficient... Incorporating functions, making It can more accurately reflect the preservation loss of different orders under the current vehicle environment, thus providing more realistic constraints for route optimization;

[0090] While reconstructing the preservation loss function, a traffic incident early warning stream is integrated, which includes information on various traffic incidents such as accidents and traffic control measures. Combined with the traffic incident early warning data, the initial path set is... A comprehensive analysis of the path was conducted to identify potential points of interruption within the path:

[0091] Define the impact factors of traffic events on route traffic, and transform events such as accidents and traffic control into calculable numerical indicators: In the formula, This represents the overall impact of the traffic incident (dimensionless; the larger the value, the more severe the impact). The severity of the event (e.g., accident = 3, control = 2, temporary construction = 1, no event = 0). The proportion of the event's impact duration to the total path duration (e.g., if the control lasts for 2 hours and the total path duration is 10 hours, then...). ); The event type weighting coefficient (reflecting the difference in the impact of event type on delivery); for the initial route set Each path in Calculate the risk score of its nodes or road segments. :

[0092] ,

[0093] In the formula, For the edge The combined impact of related traffic incidents; For the edge Historical interruption frequency; For example, a risk amplification factor. Enhance the real-time impact of events. Reduce the lag of historical frequencies; set risk assessment thresholds. If the risk score Greater than the risk assessment threshold Then mark the edge. As a risk point of disruption; it is understandable that traffic incidents will directly affect the traffic conditions of delivery routes. Identifying risk points in advance can provide forward-looking information for route optimization, avoid delivery delays caused by traffic disruptions, and further ensure delivery efficiency and goods quality.

[0094] After identifying potential route disruption points, dynamic rearrangement of path nodes is triggered under the constraints of the reconstructed preservation loss function. Simultaneously, based on real-time traffic conditions and delivery needs, potential disruption points are dynamically removed (avoiding delivery vehicles passing through areas prone to traffic disruption), and optimal detour segments are inserted to ensure vehicles can complete delivery tasks with the shortest time and lowest preservation loss. This ultimately generates an optimized path set that is resistant to environmental disturbances. :

[0095] ,

[0096] In the formula, Candidate paths, A set of feasible paths; Total path time (including real-time traffic impact); The reconstructed preservation loss function (integrating travel time and bump index); These are travel time and freshness loss weighting factors, respectively. It is understandable that this route set can effectively cope with disturbances such as abnormal vehicle environment and traffic incidents, ensuring the efficiency of the delivery process and the stability of cargo quality.

[0097] Furthermore, the formulas mentioned above are all dimensionless calculations, derived from software simulation using a large amount of collected data to approximate the real situation. The proportionality coefficients in the formulas and the preset thresholds in the analysis process are set by those skilled in the art based on the actual situation or obtained through large-scale data simulation. The magnitude of the proportionality coefficient is a specific value obtained by quantifying each parameter to facilitate subsequent comparison. The magnitude of the proportionality coefficient depends on the amount of sample data and the processing coefficients initially set by those skilled in the art for each set of sample data. As long as it does not affect the proportional relationship between the parameter and the quantified value, it is acceptable.

[0098] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. The focus of each embodiment is on its differences from other embodiments. In particular, the apparatus embodiments are described simply because they are fundamentally based on the method embodiments; relevant details can be found in the descriptions of the method embodiments.

[0099] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0100] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0101] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0102] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0103] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0104] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0105] In conclusion, the above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A flower rental delivery system based on big data, characterized in that: It includes an environmental data freshness analysis module, a survival time prediction and grading module, a dynamic path preservation decision module, a path anti-disturbance optimization and adjustment module, and a path decision iterative update module; Environmental data freshness analysis module: This module collects multi-source environmental data and flower physiological indicators from rental terminals in real time, generating a flower freshness feature vector. The process of generating the flower freshness feature vector includes: Real-time data collection from the sensor array deployed at the rental terminal, and simultaneous acquisition of flower image recognition data; The time-series data streams generated by the multi-source sensors of the rental terminal are analyzed, and non-steady-state environmental pulses are separated as environmental stress factors through signal processing technology; among them, non-steady-state environmental pulses include temperature change curves, cumulative irradiance, and vibration spectrum peaks; Biomechanical response indicators were quantified through petal edge morphology analysis and stem mechanical deformation analysis. The extracted environmental stress factors and biomechanical response indices are aligned within a spatiotemporal grid to construct a stress-response correlation tensor. Based on the constructed stress-response correlation tensor, tensor decomposition technology is used to extract cross-modal coupling features; with freshness decay acceleration as the core indicator, the cross-modal coupling features are weighted and fused to generate a flower freshness feature vector containing geographic location tags and timestamps. To address the noise interference generated during sensor acquisition, an adversarial generative network is used to simulate the sensor noise distribution characteristics; through adversarial training in the feature space, perturbation components with low correlation to flower freshness decay are filtered out. The optimized flower freshness feature vectors are encapsulated according to the rental terminal ID to form a data packet with associated weights; the final output contains a set of flower freshness feature vectors for all rental terminals. ; Survival duration prediction and grading module: Dynamically predicts the equivalent survival duration of flowers based on flower freshness feature vectors, and generates grading labels for the urgency of rental demand; Dynamic route preservation decision module: Integrates urgency labels with real-time urban traffic heat maps to construct a preservation weighted route decision model; Path disturbance rejection optimization adjustment module: Based on the status feedback of delivery vehicles, dynamically adjust the delivery strategy and generate disturbance rejection optimized paths; Route decision iterative update module: Iteratively updates the decision model parameters based on rental terminal freshness feedback and route execution data.

2. The big data-based flower rental delivery system according to claim 1, characterized in that: The survival duration prediction and grading module dynamically predicts the equivalent survival duration of flowers based on flower freshness feature vectors, and generates rental demand urgency grading labels through the following process: Obtain the feature vector set of flower freshness ; Transfer learning is used to map freshness decay to equivalent survival time. Using transfer learning techniques, a historical lease termination damage rate sample database is loaded; based on this database, a cross-category freshness decay mapping field is constructed; and the obtained flower freshness feature vector set is... Project the data onto the constructed cross-category freshness decay mapping field; through the mapping relationship, convert the freshness decay characteristics into equivalent survival time. The predicted value; simultaneously, based on the flower freshness feature vector set Association weights in This is further converted into confidence weights. Among them, confidence weight Used to calculate the confidence interval of equivalent survival time ; Collect data on return damage rates from historical rental orders to reflect the survival rate of flowers during the actual rental process; calculate the remaining rental period for each order. confidence interval of equivalent survival time The overlap ratio of the lower limit, i.e., the urgency coefficient. When the calculated urgency coefficient When the remaining time is less than the standard value for flower survival, it indicates that the remaining time of the order is insufficient to support the predicted survival time of the flowers, at which point a freshness crisis warning sign is activated; based on the urgency coefficient... The value and status of the warning indicator are used to classify the urgency level into three levels: when the urgency coefficient is... Less than the preset urgency threshold When a warning sign is activated, the situation is deemed a crisis and marked as emergency; when the urgency coefficient... Less than the preset urgency threshold And greater than or equal to the preset urgency threshold When the urgency coefficient is high, it is judged as a high-risk state and marked as high priority; when the urgency coefficient is high, it is marked as high priority. Greater than or equal to the preset urgency threshold If the condition is as described above, it is considered a safe state and marked as normal level. The urgency level labels are injected into the corresponding order data bodies to generate a set of rental orders with urgency labels. .

3. The delivery system for flower rental based on big data according to claim 2, characterized in that: The dynamic route preservation decision module integrates urgency tags with real-time urban traffic heat maps to construct a preservation weighted route decision model. The process includes: Get rental order collection Each order in this collection has been labeled with an urgency level, such as urgent, high priority, or normal, based on the predicted survival time of the flowers. Analyze the rental order set The urgency label in the data is converted into a transportation time penalty coefficient α. Based on the flower category information in the order, query the corresponding turbulence damage sensitivity β; The system integrates real-time urban traffic heat map data, which reflects the real-time traffic conditions of the road network; it abstracts the dynamic parameters of the real-time urban traffic heat map into a set of quantifiable and operable traffic influencing factors; and it encodes the travel time and bump index of road network nodes into a spatiotemporal loss matrix. For each order, a freshness loss function is defined, which comprehensively analyzes the impact of transportation time penalty coefficient, bump damage sensitivity, travel time and bump index on the freshness preservation effect of flowers; The objective function is to minimize the total freshness loss of all orders, while also constraining vehicle capacity and time windows. A Pareto evolutionary algorithm is used to solve the problem, and an initial path set is output through algorithm iteration. .

4. The big data-based flower rental delivery system according to claim 1, characterized in that: The route disturbance rejection optimization adjustment module dynamically adjusts the delivery strategy and generates disturbance rejection optimized routes based on the delivery vehicle status feedback. The process includes: Receive initial path set After receiving the initial path set, the vehicle environment offset detection operation is performed: the seismic temperature and humidity data is compared with the preset category preservation threshold corridor in real time; during continuous monitoring, when the seismic temperature and humidity data of N consecutive sampling points deviates from the preset corridor, it indicates that the internal environment of the vehicle can no longer meet the preservation requirements of the goods, and the vehicle is marked as having an abnormal microenvironment. For vehicles marked as having abnormal vehicle microenvironment, the correlation coefficient of their undelivered orders is adjusted: the transportation time penalty coefficient α for undelivered orders is multiplied according to the urgency of the orders; for urgent orders, For other order levels, Based on the updated transit time penalty coefficient Reconstruct the preservation loss function; While reconstructing the preservation loss function, a traffic incident early warning stream is integrated; combined with traffic incident early warning data, the initial path set is... A comprehensive analysis of the path was conducted to identify potential interruption points within the path; After identifying path interruption risk points, dynamic rearrangement of path nodes is triggered under the constraint of the reconstructed preservation loss function. Simultaneously, based on real-time traffic conditions and delivery needs, interruption risk points are dynamically removed, and optimal detour segments are inserted, ultimately generating an optimized path set that is resistant to environmental disturbances. .

Images