Intelligent express pickup and charging method and device, equipment and storage medium

Abstract

The invention relates to the field of logistics, and discloses an express intelligent weight taking and charging method, device and equipment and a storage medium, and the method is used for realizing intelligent weight taking and dynamic charging of a to-be-charged package. The method comprises the following steps: acquiring multi-dimensional data of a to-be-charged package, and extracting initial features of the multi-dimensional data; based on the initial features, generating comprehensive features for representing the charging attributes of the parcels by using a pre-trained charging attribute feature generation model; acquiring a real-time service parameter, matching the comprehensive feature and the real-time service parameter with a preset charging rule, and determining a dynamic charging rule matched with the to-be-charged package; and calculating the predicted freight of the to-be-charged parcel based on the dynamic charging rule and the comprehensive feature.

Description

Technical Field

[0001] This invention relates to the field of logistics, and in particular to a smart weighing and billing method, apparatus, equipment and storage medium for express delivery. Background Technology

[0002] With the continued explosive growth of the e-commerce industry and the rapid penetration of new retail models, my country's express delivery volume has maintained high-speed growth for many consecutive years. Operational efficiency and cost control in last-mile pickup, transit transportation, and final delivery have become the core focus of industry competition. Weighting and billing, as crucial initial steps in the express delivery business, directly affect logistics companies' revenue accounting, cost control, and user experience; their accuracy and efficiency are of great significance to the industry's development.

[0003] The current methods for weighing and billing in the express delivery industry still have many pain points that urgently need to be addressed. At the data collection level, traditional methods rely heavily on manual weighing and volume estimation, which is not only inefficient and difficult to adapt to the massive parcel processing needs during peak hours, but also prone to distortion of weight and volume data due to human error. Even if some outlets have introduced simple weighing equipment, it can only collect single weight data and lacks the ability to collect multi-dimensional data such as volume and transportation route, resulting in insufficient data integrity.

[0004] In terms of utilizing billing features, existing billing methods mostly formulate fixed rules based on a single dimension of weight or volume, such as billing by weight-based tiered rates. This ignores the correlation between features of different dimensions (such as the weight-volume ratio of heavy and bulky goods, and the correlation between the distance of the transportation route and the transshipment cost), resulting in one-sided feature representation and difficulty in accurately reflecting the actual billing attributes of the package.

[0005] In terms of rule adaptability, billing rules are mostly static and preset, which cannot respond to real-time changes in business dynamics, such as seasonal fluctuations in transport capacity, fuel price adjustments, and updates to remote area surcharges. This can easily lead to a disconnect between billing rules and actual business scenarios, resulting in unreasonable billing and user disputes.

[0006] Furthermore, the existing solution lacks an effective closed-loop optimization mechanism. When there is a discrepancy between the predicted freight cost and the actual freight cost, it is impossible to accurately pinpoint the cause of the discrepancy, nor can it use the difference data to optimize the billing model and rule system in reverse. This leads to the continuous accumulation of errors and makes it difficult to achieve a continuous improvement in billing accuracy.

[0007] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0008] This invention provides a method, apparatus, device, and storage medium for intelligent weighing and billing of express parcels, which can realize intelligent weighing and dynamic billing of parcels to be billed.

[0009] The first aspect of this invention provides an intelligent weighing and billing method for express delivery, comprising: acquiring multidimensional data of a package to be billed, and extracting initial features from the multidimensional data, the initial features including weight value, volume value, and route identification information; based on the initial features, generating comprehensive features for characterizing the billing attributes of the package using a pre-trained billing attribute feature generation model; acquiring real-time business parameters, matching the comprehensive features and the real-time business parameters with a pre-designed billing rule to determine a dynamic billing rule matching the package to be billed; calculating the predicted shipping cost of the package to be billed based on the dynamic billing rule and the comprehensive features; collecting packages to be billed where the difference between the actual shipping cost and the predicted shipping cost exceeds a preset difference threshold as difference samples, and using the difference samples to optimize the pre-trained billing attribute feature generation model to obtain an optimized billing attribute feature generation model.

[0010] Optionally, in a first implementation of the first aspect of the present invention, the step of acquiring multidimensional data of the package to be billed and extracting initial features from the multidimensional data, the initial features including weight value, volume value and route identification information, includes: acquiring multidimensional data of the package to be billed, the multidimensional data including real-time weight, three-dimensional dimensions and current transportation stage information; performing calibration, noise reduction and format standardization processing on the multidimensional data to obtain preprocessed data; and extracting initial features from the preprocessed data, the initial features including weight value, volume value and route identification information.

[0011] Optionally, in a second implementation of the first aspect of the present invention, the step of generating comprehensive features for characterizing package billing attributes based on the initial features using a pre-trained billing attribute feature generation model includes: vectorizing the weight value, the volume value, and the path identification information to obtain a weight feature vector, a volume feature vector, and a path feature vector; performing nonlinear transformations and feature enhancements on the weight feature vector, the volume feature vector, and the path feature vector respectively through the pre-trained billing attribute feature generation model to obtain an enhanced weight feature vector, an enhanced volume feature vector, and an enhanced path feature vector; and aggregating the enhanced weight feature vector, the enhanced volume feature vector, and the enhanced path feature vector to generate comprehensive features for characterizing package billing attributes.

[0012] Optionally, in a third implementation of the first aspect of the present invention, the step of performing nonlinear transformations and feature enhancements on the weight feature vector, the volume feature vector, and the path feature vector using a pre-trained billing attribute feature generation model to obtain enhanced weight feature vectors, enhanced volume feature vectors, and enhanced path feature vectors includes: constructing and training a billing attribute feature generation model based on a neural network, wherein the billing attribute feature generation model includes an input layer, a parallel fully connected network layer, a feature enhancement module layer based on an attention mechanism, and an output layer; mapping the weight feature vector, the volume feature vector, and the path feature vector into high-dimensional weight features, high-dimensional volume features, and high-dimensional path features through the input layer, and performing nonlinear transformations on the high-dimensional weight features, the high-dimensional volume features, and the high-dimensional path features through the parallel fully connected network layer to obtain various features after nonlinear transformation; performing feature enhancements on the various features after nonlinear transformation through the feature enhancement module layer to obtain enhanced weight feature vectors, enhanced volume feature vectors, and enhanced path feature vectors, and outputting the enhanced weight feature vectors, the enhanced volume feature vectors, and the enhanced path feature vectors through the output layer.

[0013] Optionally, in a fourth implementation of the first aspect of the present invention, the step of obtaining real-time business parameters and matching the comprehensive features and the real-time business parameters with pre-designed billing rules to determine a dynamic billing rule matching the package to be billed includes: constructing a billing rule query knowledge base, the billing rule query knowledge base including multi-dimensional pre-designed billing rules; obtaining real-time business parameters and generating a query request based on the comprehensive features and the real-time business parameters; and performing a matching query in the billing rule query knowledge base based on the query request to determine a dynamic billing rule matching the package to be billed.

[0014] Optionally, in a fifth implementation of the first aspect of the present invention, the step of calculating the predicted shipping cost of the package to be billed based on the dynamic billing rules and the comprehensive features includes: determining a rate formula based on the dynamic billing rules; substituting the weight and volume values ​​of the comprehensive features into the rate formula to calculate the basic shipping cost of the package to be billed; and adjusting the basic shipping cost based on the timeliness of the route identification information of the comprehensive features to obtain the predicted shipping cost.

[0015] Optionally, in a sixth implementation of the first aspect of the present invention, after calculating the predicted shipping cost of the package to be billed, the method further includes collecting packages to be billed whose actual shipping cost and predicted shipping cost differ by more than a preset difference threshold as difference samples, and using the difference samples to optimize the pre-trained billing attribute feature generation model to obtain an optimized billing attribute feature generation model. The step of collecting packages to be billed whose actual shipping cost and predicted shipping cost differ by more than a preset difference threshold as difference samples and using the difference samples to optimize the pre-trained billing attribute feature generation model to obtain the optimized billing attribute feature generation model includes: collecting multiple packages to be billed whose actual shipping cost and predicted shipping cost differ by more than a preset difference threshold as sample packages, and recording the corresponding data for each sample package. The associated data is used to construct a difference sample set using the associated data corresponding to each sample package. The associated data includes initial features, comprehensive features, dynamic billing rules, and actual freight costs. Using the initial features in the difference sample set as input and the billing attributes included in the corresponding actual freight costs as optimization targets, the pre-trained billing attribute feature generation model is retrained in a supervised manner to obtain an optimized billing attribute feature generation model. The key factors in the difference sample set that cause deviations in rule matching or rate calculation are analyzed, and based on the analysis results, at least one of the following operations is performed on the preset billing rule query knowledge base: correcting the calculation parameters of existing rules, supplementing new constraint rules, or adjusting the priority weights between rules, to generate an optimized billing rule query knowledge base.

[0016] A second aspect of the present invention provides an intelligent weighing and billing device for express delivery, comprising: an extraction module for acquiring multidimensional data of a package to be billed and extracting initial features from the multidimensional data, the initial features including weight value, volume value and path identification information; a generation module for generating comprehensive features characterizing the billing attributes of the package based on the initial features using a pre-trained billing attribute feature generation model; and a matching module for acquiring real-time business parameters and matching the comprehensive features and the real-time business parameters with pre-designed billing rules to determine a dynamic billing rule matching the package to be billed.

[0017] A third aspect of the present invention provides an intelligent weighing and billing device for express delivery, comprising: a memory and at least one processor, wherein the memory stores computer-readable instructions, and the memory and the at least one processor are interconnected via a circuit; the at least one processor invokes the computer-readable instructions in the memory to cause the intelligent weighing and billing device for express delivery to perform the various steps of the intelligent weighing and billing method for express delivery as described above.

[0018] A fourth aspect of the present invention provides a computer-readable storage medium storing computer-readable instructions that, when executed on a computer, cause the computer to perform the steps of the express delivery intelligent weighing and billing method described above.

[0019] The technical solution provided by this invention collects multi-dimensional core data such as package weight, volume, and route to extract initial features, providing a comprehensive foundation for billing and avoiding billing deviations caused by a single data dimension. Furthermore, based on a pre-trained billing attribute feature generation model, the scattered initial features are transformed into comprehensive features that accurately characterize billing attributes, breaking the limitation of traditional billing relying solely on a single physical attribute. Simultaneously, dynamic matching is performed by combining real-time business parameters with a multi-dimensional billing rule knowledge base, allowing billing rules to adapt to real-time business scenarios and solving the rigidity problem of fixed rules. In addition, freight prediction is calculated using comprehensive features and dynamic rules, taking into account both package attributes and actual business conditions. Furthermore, continuous optimization of the model by collecting difference samples achieves iterative improvements in billing accuracy. This not only improves the accuracy and efficiency of freight calculation but also enables rapid response to business changes, effectively reducing disputes and operating costs caused by billing errors. Attached Figure Description

[0020] Figure 1 This is a first flowchart of the intelligent weighing and billing method for express delivery provided in an embodiment of the present invention; Figure 2 This is a second flowchart of the intelligent weighing and billing method for express delivery provided in an embodiment of the present invention; Figure 3 This is a third flowchart of the intelligent weighing and billing method for express delivery provided in an embodiment of the present invention; Figure 4 This is a fourth flowchart of the intelligent weighing and billing method for express delivery provided in an embodiment of the present invention; Figure 5 This is a fifth flowchart of the intelligent weighing and billing method for express delivery provided in an embodiment of the present invention; Figure 6 This is a sixth flowchart of the intelligent weighing and billing method for express delivery provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the intelligent weighing and billing device for express delivery provided in an embodiment of the present invention; Figure 8 This is a schematic diagram of the intelligent weighing and billing device for express delivery provided in an embodiment of the present invention. Detailed Implementation

[0021] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0022] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 The first embodiment of the intelligent weighing and billing method for express delivery in this invention includes: S101. Obtain multidimensional data of the package to be billed and extract the initial features of the multidimensional data, including weight value, volume value and path identification information.

[0023] It is understood that the executing entity of this invention can be an intelligent weighing and billing device for express delivery, or it can be a terminal or a server; no specific limitation is made here. This embodiment of the invention will be described using a server as an example.

[0024] In this embodiment, the core data of the package to be billed is quickly collected by the intelligent weighing equipment at the express delivery outlet (such as a terminal that integrates weighing and volume measurement functions). This includes the actual weight data of the package, three-dimensional dimension data, and the current transportation node information (such as the pickup point or transit center). From the collected data, the key initial features that directly affect the billing are extracted, namely the accurate weight value, the calculated volume value, and the path identification information marking the transportation route. This forms a standardized set of initial features, providing concise and core data support for the subsequent feature generation.

[0025] S102. Based on the initial features, use the pre-trained billing attribute feature generation model to generate comprehensive features to characterize the package billing attributes.

[0026] In this embodiment, the extracted initial features are first input into a pre-trained billing attribute feature generation model. This model has been trained on a massive amount of express delivery billing samples and has the ability to deeply mine feature associations. The model first performs internal transformation processing on initial features such as weight, volume, and route to mine potential associations between features (such as the weight-to-volume ratio and the association between route and transportation cost). Then, it enhances the representational ability of the features through nonlinear transformation, highlighting information that is key to billing. Finally, the processed features are fused and integrated to generate a comprehensive feature that fully and accurately reflects the billing attributes of the package. This feature not only includes physical attribute information but also incorporates implicit billing association information, laying an accurate foundation for subsequent rule matching and freight calculation.

[0027] S103. Obtain real-time business parameters, match the comprehensive features and real-time business parameters with the pre-designed billing rules, and determine the dynamic billing rules that match the package to be billed.

[0028] In this embodiment, dynamic parameters of the express delivery business are captured in real time, including the current transportation capacity, peak and off-peak season price adjustment coefficients, and special area surcharge standards. Combining the generated comprehensive features with real-time business parameters, targeted billing rule query conditions are constructed, covering multi-dimensional information such as package physical attributes, transportation routes, and real-time business scenarios. Subsequently, a rapid retrieval and matching process is performed in a pre-set billing rule query knowledge base. This knowledge base stores pre-designed billing rules for multiple scenarios. Through keyword matching and semantic association analysis, the dynamic billing rule most suitable for the current package and business scenario is selected, ensuring that the billing rule aligns with real-time business needs and avoiding unreasonable billing issues caused by fixed rules.

[0029] S104. Based on dynamic billing rules and comprehensive characteristics, calculate the predicted shipping cost of the parcel to be billed.

[0030] In this embodiment, based on the matched dynamic billing rules, the core rate parameters and calculation logic are extracted to form a standardized freight calculation template. The core billing data, such as weight and volume, contained in the comprehensive features are substituted into the corresponding rate formula according to the requirements of the calculation template to obtain the basic freight cost of the package. Then, combining the route identification information in the comprehensive features, and considering factors such as transportation distance, number of transit nodes, and timeliness requirements, the basic freight cost is adjusted according to specific scenarios (such as long-distance transportation surcharges and express delivery surcharges), ultimately generating an accurate predicted freight cost.

[0031] S105. Collect packages to be billed where the difference between actual and predicted shipping costs exceeds a preset difference threshold as difference samples, and use the difference samples to optimize the pre-trained billing attribute feature generation model to obtain an optimized billing attribute feature generation model.

[0032] In this embodiment, the difference between predicted and actual shipping costs is periodically analyzed, and packages with differences exceeding a preset threshold are selected as difference samples. Complete associated data for these difference samples is collected, including initial features, comprehensive features, matching dynamic billing rules, actual shipping costs, and annotations of the reasons for the differences. Using these difference samples as training data, and taking the initial features as input, the pre-trained billing attribute feature generation model is iteratively retrained with the actual billing attributes corresponding to the actual shipping costs as the optimization target. The model's internal parameters and feature processing logic are adjusted. By continuously incorporating difference samples for optimization, the accuracy of the comprehensive features generated by the model is continuously improved, making subsequent billing rule matching and shipping cost calculation more closely aligned with actual business scenarios.

[0033] This embodiment provides an intelligent weighing and billing method for express delivery. It collects multi-dimensional core data such as package weight, volume, and route to extract initial features, providing a comprehensive foundation for billing and avoiding billing errors caused by single data dimensions. Furthermore, based on a pre-trained billing attribute feature generation model, it transforms scattered initial features into comprehensive features that accurately characterize billing attributes, breaking the limitation of traditional billing relying solely on a single physical attribute. Simultaneously, it dynamically matches real-time business parameters with a multi-dimensional billing rule knowledge base, allowing billing rules to adapt to real-time business scenarios and solving the rigidity problem of fixed rules. In addition, it calculates and predicts freight costs through comprehensive features and dynamic rules, taking into account both package attributes and actual business conditions. Moreover, by collecting difference samples to continuously optimize the model, it achieves iterative improvements in billing accuracy. This not only improves the accuracy and efficiency of freight calculation but also enables rapid response to business changes, effectively reducing disputes and operating costs caused by billing errors.

[0034] Please see Figure 2 The second embodiment of the intelligent weighing and billing method for express delivery in this invention includes: S201. Obtain multidimensional data of the package to be billed, including real-time weight, three-dimensional dimensions and current transportation stage information.

[0035] In this embodiment, core data of parcels awaiting billing are collected in real time using integrated intelligent devices (such as automatic weighing and volume measurement machines) deployed at express delivery outlets. High-precision weighing sensors are used to obtain the actual weight of the parcel; a laser ranging module scans the three-dimensional dimensions of the parcel to automatically calculate its volume; and the device, in conjunction with the express delivery business system, obtains information on the parcel's current transportation stage (such as pickup or transit). During the data collection process, the device automatically records the data collection timestamp and device number for easy data traceability and anomaly investigation. For parcels with unusual shapes, the device activates an adaptive collection mode, scanning multiple times and averaging the data to ensure the accuracy of size and weight data, ultimately forming multi-dimensional raw data containing three types of core information.

[0036] S202. Perform calibration, denoising, and format standardization on the multidimensional data to obtain preprocessed data.

[0037] In this embodiment, the collected multidimensional raw data undergoes three steps of processing. The first step is calibration, which corrects sensor system errors by comparing the data with standard weights and volume blocks, ensuring the accuracy of weight and volume data. The second step is noise reduction, employing statistical filtering algorithms to eliminate outliers caused by equipment vibration and light interference (such as instantaneous weight fluctuations or volume data exceeding reasonable ranges), retaining only the true and valid data. The third step is format standardization, converting data from different units to a preset standard (e.g., weight units are standardized to kg, volume units to m³, and transportation stage information to standardized codes). Simultaneously, data fields are named in a standardized manner to ensure the data format meets the requirements of subsequent processing, ultimately outputting clean and standardized preprocessed data.

[0038] S203. Extract initial features from the preprocessed data. The initial features include weight value, volume value and path identification information.

[0039] In this embodiment, the calibrated weight value is extracted as the weight value feature, and the calculated volume value is extracted as the volume value feature. The transportation stage information is parsed, and combined with the package's origin and destination information, a unique route identifier is generated (such as the code corresponding to "pickup point A - transit center B - delivery point C"). This identifier accurately reflects the package's transportation route. During the extraction process, the feature data undergoes secondary verification to ensure that the weight and volume values ​​are within a reasonable business range and that the route identifier information has no coding errors.

[0040] In this embodiment, the real-time weight, three-dimensional dimensions, and current transportation stage information of the package are collected comprehensively first to ensure the integrity of multi-dimensional data and avoid billing deviations caused by missing data. Then, through a preprocessing process of calibration, noise reduction, and format standardization, abnormal data caused by equipment acquisition errors and environmental interference are effectively corrected, and the data format of different acquisition devices is unified, which greatly improves the data quality. Finally, the weight value, volume value, and path identification information are accurately extracted from the preprocessed data as initial features, providing a clean and standardized data source for the subsequent generation of comprehensive features.

[0041] Please see Figure 3 A third embodiment of the intelligent weighing and billing method for express delivery in this invention includes: S301. Vectorize the weight value, volume value and path identification information to obtain the weight feature vector, volume feature vector and path feature vector.

[0042] In this embodiment, an encoding method adapted to different feature types is adopted to transform the three initial features into standardized data vectors. For continuous numerical features such as weight and volume values, the Min-Max normalization method is used to map the values ​​to the [0,1] interval, eliminating the influence of dimensions and generating fixed-dimensional weight and volume feature vectors. For discrete classification features such as route identification information, a one-hot encoding method is used to map different transportation routes into binary vectors, ensuring that each route identifier has a unique vector representation and generating a route feature vector.

[0043] S302. The pre-trained billing attribute feature generation model is used to perform nonlinear transformation and feature enhancement on the weight feature vector, volume feature vector and path feature vector respectively to obtain the enhanced weight feature vector, enhanced volume feature vector and enhanced path feature vector.

[0044] In this embodiment, a pre-trained billing attribute feature generation model is used to perform nonlinear transformations and feature enhancements on the weight feature vector, volume feature vector, and path feature vector, respectively, to obtain enhanced weight feature vector, enhanced volume feature vector, and enhanced path feature vector. Specifically, this includes: constructing and training a billing attribute feature generation model based on a neural network, which includes an input layer, a parallel fully connected network layer, an attention-based feature enhancement module layer, and an output layer; mapping the weight feature vector, volume feature vector, and path feature vector into high-dimensional weight features, high-dimensional volume features, and high-dimensional path features through the input layer, and performing nonlinear transformations on the high-dimensional weight features, high-dimensional volume features, and high-dimensional path features through the parallel fully connected network layer, respectively, to obtain various features after nonlinear transformation; enhancing the various features after nonlinear transformation through the feature enhancement module layer to obtain enhanced weight feature vector, enhanced volume feature vector, and enhanced path feature vector; and outputting the enhanced weight feature vector, enhanced volume feature vector, and enhanced path feature vector through the output layer.

[0045] In this embodiment, a neural network model is built based on a deep learning framework, defining a three-layer core architecture: the input layer receives three types of feature vectors, and the output dimension is consistent with the feature vector dimension; the intermediate layer contains three parallel fully connected network layers (corresponding to weight, volume, and path features, respectively), each containing multiple hidden units, and introducing non-linearity through the ReLU activation function; an attention mechanism feature enhancement layer is set before the output to focus on key features. During the training phase, a massive amount of express delivery billing samples (including initial feature vectors and corresponding real billing attribute labels) are collected, and the training set, validation set, and test set are divided in a 7:2:1 ratio. Gradient descent is used to optimize the model parameters, and the matching degree between feature representations and real billing attributes is used as the evaluation index. Hyperparameters such as the number of network layers and the number of hidden units are continuously adjusted until the model's evaluation index on the test set reaches the preset standard, completing pre-training and solidifying the model parameters.

[0046] In this embodiment, after receiving feature vectors of weight, volume, and path from the model input layer, these vectors are first mapped to high-dimensional feature vectors. This dimensionality increase expands the feature expression space, allowing the features to carry more potential information. Subsequently, the high-dimensional feature vectors are input into corresponding parallel fully connected network layers. Each network layer performs layer-by-layer operations and non-linear transformations on the features. Through matrix multiplication and activation functions, the linear correlation between features is broken, uncovering deep patterns within the features (such as the implicit correlation between different weight ranges and rates). During the transformation process, the parallel architecture ensures that the processing of the three types of features does not interfere with each other, while the activation function effectively solves the gradient vanishing problem, ensuring the effective extraction of deep features. Finally, the output consists of various feature vectors that have undergone deep processing and possess strong expressive power.

[0047] In this embodiment, the three types of feature vectors, after nonlinear transformation, are input into the attention mechanism feature enhancement layer. This layer automatically allocates attention resources by calculating the correlation weights between various features and billing attributes. For core billing features such as weight and volume, higher weights are assigned to strengthen their contribution to the overall feature set. For key information affecting transportation costs in the route features (such as transportation distance and number of transfers), the focus is placed on increasing their weights. For secondary information (such as irrelevant features implied by the surface material of the package), lower weights are assigned to weaken it. Through weighted summation, the various feature vectors are enhanced to highlight key information and filter out interfering information, ultimately generating enhanced weight feature vectors, enhanced volume feature vectors, and enhanced route feature vectors.

[0048] S303. Aggregate the enhanced weight feature vector, enhanced volume feature vector, and enhanced path feature vector to generate a comprehensive feature for characterizing the package billing attributes.

[0049] In this embodiment, differentiated weights are assigned to three types of enhanced vectors based on historical billing data. Weight and volume features have higher weights, while the weight of path features is dynamically adjusted according to the business scenario (e.g., the weight of long-distance transportation paths is increased). By concatenating vectors and weighted summing, the three types of feature vectors are merged into a high-dimensional comprehensive feature vector. This vector retains the core information of each type of feature and integrates the correlation between features, comprehensively representing the billing attributes of the package (including physical attributes, transportation scenario attributes, etc.).

[0050] In this embodiment, the three initial features—weight, volume, and path identifier—are first vectorized and encoded, transforming unstructured and heterogeneous features into standardized vectors that the model can process, laying the foundation for subsequent feature processing. Then, a pre-trained model is used to perform nonlinear transformations and feature enhancements on various vectors, deeply mining the potential value of individual features while strengthening the correlation information between features, avoiding the limitations of directly using traditional features. Finally, the enhanced features are aggregated to generate comprehensive features that can fully and accurately represent the billing attributes of the package, including the core information of physical attributes and incorporating implicit billing-related features.

[0051] Please see Figure 4 A fourth embodiment of the intelligent weighing and billing method for express delivery in this invention includes: S401. Construct a billing rule query knowledge base, which includes multi-dimensional pre-designed billing rules.

[0052] In this embodiment, various billing scenarios in the express delivery industry are analyzed, and a multi-dimensional billing rule system is constructed based on weight, volume, route, timeliness, and surcharges. Each dimension contains specific pre-designed billing rules. For example, the weight dimension covers tiered rates for different weight ranges, the route dimension includes basic rates and transit surcharges for different transportation routes, and the timeliness dimension includes surcharge rules for different timeliness requirements such as standard and express delivery. All rules are entered into a knowledge base in a standardized format, and keyword tags (such as 1-3kg, inter-provincial transportation, peak season) are added to each rule. At the same time, an association index between rules is established (such as the priority determination of weight rules and volume rules). The billing rule query knowledge base supports dynamic updates, allowing new rules to be added, old rules to be modified, or rule priorities to be adjusted at any time according to business adjustments, ensuring the timeliness and comprehensiveness of the rule system.

[0053] S402. Obtain real-time business parameters and generate a query request based on comprehensive features and real-time business parameters.

[0054] In this embodiment, the system connects in real-time to the express delivery business management platform to capture various real-time business parameters, including the current peak / off-peak season indicators, the capacity pressure coefficient of the transportation route, fuel surcharge standards, and surcharge rules for special areas (such as remote areas). Combining the previously generated comprehensive parcel characteristics, core query elements (such as weight range, volume category, route type, and delivery time requirements) are extracted. These core elements are then integrated with the real-time business parameters to generate a structured query request in a preset format. The query request includes multiple keyword fields (such as "weight: 2.5kg", "route: inter-province", "delivery time: normal", "season: peak season") to ensure accurate matching with the corresponding rules in the knowledge base.

[0055] S403. Based on the query request, perform a matching query in the billing rule query knowledge base to determine the dynamic billing rule that matches the package to be billed.

[0056] In this embodiment, the generated structured query request is input into the billing rule query knowledge base to activate the intelligent search engine. During the search process, a preliminary set of rules that meet the criteria is first filtered out through keyword matching. Then, based on semantic association analysis and rule priority determination, the matching scope is further narrowed down. For example, for a 2.5kg package transported across provinces, rules related to weights of 1-3kg and cross-provincial routes are first filtered out. Then, combined with peak season parameters and capacity tension coefficients, the final applicable tiered fare rate and surcharge rules are determined. After matching is completed, a unique dynamic billing rule is output, and a rule matching detail is generated, explaining the matched keywords, priority determination basis, and real-time parameter impact, ensuring the traceability of the billing rule.

[0057] In this embodiment, a query knowledge base containing multi-dimensional pre-designed billing rules is first constructed, covering various scenario rules such as weight-based tiered rates, volumetric billing standards, route surcharges, and time-sensitive pricing, ensuring comprehensive rule coverage. Then, dynamic business parameters are acquired in real time, and targeted query requests are generated based on the comprehensive characteristics of the package. This ensures that the query conditions include both the inherent attributes of the package and fit the real-time business scenario. Finally, based on the query request, the knowledge base is quickly matched to select the dynamic billing rule most suitable for the current package and business status, breaking the rigid limitations of traditional fixed rules. The overall process achieves precise linkage between package attributes, real-time business, and billing rules, improving the efficiency of rule matching and ensuring the dynamic adaptability of billing rules. It can quickly respond to the complex changes in express delivery business, providing flexible rule support for accurate billing.

[0058] Please see Figure 5 The fifth embodiment of the intelligent weighing and billing method for express delivery in this invention includes: S501. Determine the rate formula based on dynamic billing rules.

[0059] In this embodiment, core rate parameters and calculation logic are extracted from the matched dynamic billing rules to construct a standardized rate formula. The formula type is dynamically adjusted according to package attributes and business scenarios: for heavy goods, a formula combining weight-based tiered rates and route surcharges is used; for lightweight goods, a formula combining volume-converted weight rates and regional surcharges is used; for packages with special time-sensitive requirements, a time-sensitive surcharge coefficient is added to the basic formula. The rate formula clearly defines the definition and calculation method of each parameter, such as the weight-based tiered rate calculated as "initial weight + additional weight × number of additional weights," and the route surcharge set as a fixed amount or percentage based on segmented transportation distances.

[0060] S502. Substitute the weight and volume values ​​of the comprehensive characteristics into the rate formula to calculate the basic shipping cost of the parcel to be billed.

[0061] In this embodiment, precise weight and volume values ​​are extracted from the comprehensive characteristics of the package. Data preprocessing is performed according to the requirements of the rate formula: if the formula requires weight to be converted by volume, the volume value is converted into converted weight using a preset conversion factor (e.g., 1m³=300kg); if it is a weight-based tiered pricing system, the tier range of the weight value is determined to determine the initial weight fee and the number of additional weight charges. The processed weight or volume data is substituted into the rate formula, and calculation is performed step by step according to the formula logic: first, the basic rate portion (e.g., initial weight fee + additional weight fee) is calculated, and then a fixed surcharge (e.g., basic route surcharge) is added to obtain the initial shipping cost of the package.

[0062] S503. Combining the route identification information of comprehensive features, the basic freight cost is adjusted for timeliness to obtain the predicted freight cost.

[0063] In this embodiment, the route identification information in the comprehensive features is analyzed to extract key transportation elements, including transportation distance (short / medium / long distance), number of transfer nodes, and destination area attributes (urban / suburban / remote areas). Based on these elements, a preset timeliness adjustment rule base is queried to determine the corresponding adjustment coefficient or surcharge: for example, a 5% distance surcharge is added to the initial freight for long-distance transportation, a fixed remoteness surcharge is added for remote areas, and a transfer surcharge is added if there are more than 3 transfers. The initial freight and various adjustment fees are added together to obtain the final predicted freight.

[0064] In this embodiment, the corresponding rate formula is first accurately extracted based on dynamically matched billing rules to ensure that the basic billing logic conforms to business specifications. Then, the weight and volume values, the most critical features in the comprehensive characteristics, are substituted into the formula to calculate the initial freight cost, ensuring the core accuracy of the billing. Finally, the initial freight cost is adjusted in a targeted manner by combining the route identification information in the comprehensive characteristics and considering time-sensitive factors such as transportation distance, number of transit nodes, and regional attributes. This ensures that the final freight cost not only conforms to the basic billing standards but also adapts to the differences in specific transportation scenarios. The overall process avoids the traditional one-size-fits-all billing method, balancing billing standardization with scenario-based adaptation. This effectively reduces errors caused by the basic billing not considering scenario differences, improves the accuracy and rationality of freight cost calculation, and the calculation process is simple and efficient, adapting to the rapid operational needs of express delivery outlets.

[0065] Please see Figure 6 The sixth embodiment of a smart weighing and billing method for express delivery in this invention includes: S601. Collect multiple parcels to be billed where the difference between the actual freight and the predicted freight exceeds a preset difference threshold as sample parcels, record the associated data corresponding to each sample parcel, and use the associated data corresponding to each sample parcel to construct a difference sample set. The associated data includes initial features, comprehensive features, dynamic billing rules, and actual freight.

[0066] In this embodiment, data for all billed packages are periodically (e.g., daily) analyzed to calculate the absolute difference between the predicted and actual shipping costs. A preset difference threshold (e.g., 5 yuan or 10% of the total shipping cost) is set, and packages with differences exceeding the threshold are selected as sample packages. For each sample package, comprehensive related data is collected, including initial features (weight, volume, route identifier), comprehensive features, matching dynamic billing rules, actual shipping cost, and preliminary annotations of the reasons for the difference (e.g., rule matching deviation, inaccurate feature generation). This data is then organized in a standardized format to construct a structured difference sample set. The sample set includes complete link information such as input data, billing process, output results, actual results, and reasons for the difference.

[0067] S602. Using the initial features in the differential sample set as input and the billing attributes included in the corresponding actual freight as the optimization objective, the pre-trained billing attribute feature generation model is retrained in a supervised manner to obtain the optimized billing attribute feature generation model.

[0068] In this embodiment, the initial features from the differential sample set are used as the model input data, and the actual billing attributes implied by the actual freight charges corresponding to the samples (such as the actual applicable rate level and surcharge type) are used as supervision labels to construct a model retraining dataset. The model retraining dataset is divided into a training set and a validation set in an 8:2 ratio. The parameters of the original pre-trained model are loaded as initial weights, and the model is iteratively trained using supervised learning. During training, the matching degree between the comprehensive features generated by the model and the actual billing attributes is used as the optimization objective. The internal weight parameters and transformation logic of the model are adjusted using the gradient descent method. The model performance is evaluated on the validation set after each round of training. If the feature matching degree on the validation set reaches a preset standard (such as above 95%), training is stopped, the optimized model parameters are saved, and the optimized billing attribute feature generation model is obtained, ensuring that the model can specifically correct previous deviations.

[0069] S603. Analyze the key factors in the differential sample set that cause deviations in rule matching or rate calculation, and based on the analysis results, perform at least one of the following operations on the preset billing rule query knowledge base: modify the calculation parameters of existing rules, supplement new constraint rules, or adjust the priority weights between rules, to generate an optimized billing rule query knowledge base.

[0070] In this embodiment, a deep analysis of the discrepancy sample set is conducted to identify key factors causing billing deviations: if the deviation is due to inaccurate rule parameters (e.g., a lower rate for a certain weight range), the calculation parameters of the corresponding rule are corrected; if the deviation is due to a lack of specific scenario rules (e.g., no corresponding rule for a newly added short-distance same-city express delivery scenario), new constraint rules are added to clarify the rate formula and surcharge standard for that scenario; if the deviation is due to unreasonable rule priorities (e.g., a conflict between volume and weight rules leading to matching errors), the priority weights between rules are adjusted to clarify the order of rule application in conflict scenarios. After analysis and adjustment, the optimized rules are updated to the billing rule query knowledge base, and a rule adjustment log is recorded (including adjustment content, the sample cases used, and the adjustment time). The updated knowledge base is synchronously applied to subsequent billing processes, working in conjunction with the optimized model to achieve a comprehensive improvement in billing accuracy.

[0071] In this embodiment, packages with significant discrepancies between predicted and actual shipping costs are first precisely selected as difference samples. A comprehensive collection of associated data from these samples provides a rich and accurate data source for optimization. Then, based on these difference samples, the billing attribute feature generation model undergoes supervised retraining to specifically correct model biases and improve the accuracy of feature generation. Simultaneously, the causes of these discrepancies are analyzed in depth, and the billing rule query knowledge base is optimized in multiple dimensions (parameter correction, rule supplementation, and priority adjustment) to improve the rule system.

[0072] The above describes the intelligent weighing and billing method for express delivery in the embodiments of the present invention. The following describes the device in the embodiments of the present invention. Please refer to [link / reference]. Figure 7 The implementation methods of the intelligent weighing and billing device for express delivery in this embodiment of the invention include: Extraction module 701 is used to acquire multidimensional data of the package to be billed and extract the initial features of the multidimensional data, the initial features including weight value, volume value and path identification information; The generation module 702 is used to generate comprehensive features for characterizing package billing attributes based on the initial features using a pre-trained billing attribute feature generation model. The matching module 703 is used to obtain real-time business parameters, match the comprehensive features and the real-time business parameters with the pre-designed billing rules, and determine the dynamic billing rules that match the package to be billed. The calculation module 704 is used to calculate the predicted shipping cost of the package to be billed based on the dynamic billing rules and the comprehensive features. The optimization module 705 is used to collect packages to be billed where the difference between the actual freight cost and the predicted freight cost exceeds a preset difference threshold as difference samples, and to use the difference samples to optimize the pre-trained billing attribute feature generation model to obtain an optimized billing attribute feature generation model.

[0073] In this embodiment, the extraction module 701 includes: an acquisition unit 7011, used to acquire multidimensional data of the package to be billed, the multidimensional data including real-time weight, three-dimensional dimensions and current transportation stage information; a preprocessing unit 7012, used to perform calibration, noise reduction and format standardization processing on the multidimensional data to obtain preprocessed data; and an extraction unit 7013, used to extract initial features from the preprocessed data, the initial features including weight value, volume value and path identification information.

[0074] In this embodiment, the generation module 702 includes: an encoding unit 7021, used to vectorize the weight value, the volume value, and the path identification information to obtain a weight feature vector, a volume feature vector, and a path feature vector; an enhancement unit 7022, used to perform nonlinear transformation and feature enhancement on the weight feature vector, the volume feature vector, and the path feature vector respectively through a pre-trained billing attribute feature generation model to obtain an enhanced weight feature vector, an enhanced volume feature vector, and an enhanced path feature vector; and a first generation unit 7023, used to aggregate the enhanced weight feature vector, the enhanced volume feature vector, and the enhanced path feature vector to generate a comprehensive feature for characterizing the package billing attributes.

[0075] In this embodiment, the matching module 703 includes: a construction unit 7031, used to construct a billing rule query knowledge base, the billing rule query knowledge base including multi-dimensional pre-designed billing rules; a second generation unit 7032, used to obtain real-time business parameters and generate a query request based on the comprehensive features and the real-time business parameters; and a matching unit 7033, used to perform a matching query in the billing rule query knowledge base based on the query request to determine the dynamic billing rule that matches the package to be billed.

[0076] In this embodiment, the calculation module 704 includes: a determination unit 7041, used to determine a rate formula based on dynamic billing rules; a calculation unit 7042, used to substitute the weight and volume values ​​of the comprehensive features into the rate formula to calculate the basic shipping cost of the package to be billed; and an adjustment unit 7043, used to adjust the basic shipping cost based on the timeliness of the route identification information of the comprehensive features to obtain a predicted shipping cost.

[0077] In this embodiment, the optimization module 705 includes: a collection unit 7051, used to collect multiple packages to be billed where the difference between actual and predicted freight costs exceeds a preset difference threshold as sample packages, record the associated data corresponding to each sample package, and construct a difference sample set using the associated data corresponding to each sample package, wherein the associated data includes initial features, comprehensive features, dynamic billing rules, and actual freight costs; a first optimization unit 7052, used to take the initial features in the difference sample set as input, take the billing attributes included in the corresponding actual freight costs as optimization targets, and perform supervised retraining on a pre-trained billing attribute feature generation model to obtain an optimized billing attribute feature generation model; and a second optimization unit 7053, used to analyze the key factors in the difference sample set that cause deviations in rule matching or rate calculation, and perform at least one operation on a preset billing rule query knowledge base based on the analysis results, such as correcting the calculation parameters of existing rules, supplementing new constraint rules, or adjusting the priority weights between rules, to generate an optimized billing rule query knowledge base.

[0078] In this embodiment, by collecting multi-dimensional core data such as package weight, volume, and route, initial features are extracted to provide a comprehensive foundation for billing, avoiding billing deviations caused by a single data dimension. Moreover, based on a pre-trained billing attribute feature generation model, the scattered initial features are transformed into comprehensive features that can accurately represent billing attributes, breaking the limitation of traditional billing relying solely on a single physical attribute. At the same time, dynamic matching is performed by combining real-time business parameters with a multi-dimensional billing rule knowledge base, allowing billing rules to adapt to real-time business scenarios and solving the rigidity problem of fixed rules. Furthermore, by calculating and predicting freight costs through comprehensive features and dynamic rules, both package attributes and actual business conditions are taken into account. In addition, by collecting difference samples to continuously optimize the model, iterative improvements in billing accuracy are achieved. This not only improves the accuracy and efficiency of freight calculation but also enables rapid response to business changes, thereby effectively reducing disputes and operating costs caused by billing errors.

[0079] Figure 7 The structure of the intelligent weighing and billing device for express delivery shown does not constitute a limitation on the intelligent weighing and billing device for express delivery, and can realize the steps of the intelligent weighing and billing methods for express delivery provided in the above-described method embodiments.

[0080] above Figure 7 The intelligent weighing and billing device for express delivery in this embodiment of the invention is described in detail from the perspective of modular functional entities. The intelligent weighing and billing device for express delivery in this embodiment of the invention is described in detail below from the perspective of hardware processing.

[0081] Figure 8 This is a schematic diagram of the structure of an intelligent express delivery weighing and billing device 800 provided in an embodiment of the present invention. The device 800 can vary significantly due to different configurations or performance, and may include one or more central processing units (CPUs) 810 (e.g., one or more processors) and a memory 820, and one or more storage media 830 (e.g., one or more mass storage devices) for storing application programs 833 or data 832. The memory 820 and storage media 830 can be temporary or persistent storage. The program stored in the storage media 830 may include one or more modules (not shown), each module may include a series of instruction operations on the device 800. Furthermore, the processor 810 may be configured to communicate with the storage media 830 and execute the series of instruction operations in the storage media on the device 800.

[0082] Device 800 may also include one or more power supplies 840, one or more wired or wireless network interfaces 850, one or more input / output interfaces 860, and / or one or more operating systems 831, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc.

[0083] This invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the express delivery intelligent weighing and billing method.

[0084] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0085] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0086] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart weighing and billing method for express delivery, characterized in that, The intelligent weighting and billing method for express delivery includes: Obtain multidimensional data of the package to be billed, and extract the initial features of the multidimensional data, including weight value, volume value and path identification information; Based on the initial features, a pre-trained billing attribute feature generation model is used to generate comprehensive features to characterize the package billing attributes. Obtain real-time business parameters, match the comprehensive features and the real-time business parameters with the pre-designed billing rules, and determine the dynamic billing rules that match the package to be billed; Based on the dynamic billing rules and the comprehensive features, the predicted shipping cost of the package to be billed is calculated.

2. The intelligent weighing and billing method for express delivery according to claim 1, characterized in that, The process involves acquiring multidimensional data of the package to be billed and extracting initial features from the multidimensional data. These initial features include weight, volume, and path identification information, including: Acquire multidimensional data of the package to be billed, including real-time weight, three-dimensional dimensions, and current transportation stage information; The multidimensional data is calibrated, denoised, and format-normalized to obtain preprocessed data; Initial features are extracted from the preprocessed data, including weight value, volume value and path identification information.

3. The intelligent weighing and billing method for express delivery according to claim 1, characterized in that, Based on the initial features, the pre-trained billing attribute feature generation model is used to generate comprehensive features to characterize package billing attributes, including: The weight value, the volume value, and the path identification information are vectorized and encoded to obtain a weight feature vector, a volume feature vector, and a path feature vector; The pre-trained billing attribute feature generation model performs nonlinear transformation and feature enhancement on the weight feature vector, the volume feature vector, and the path feature vector to obtain enhanced weight feature vector, enhanced volume feature vector, and enhanced path feature vector. The enhanced weight feature vector, the enhanced volume feature vector, and the enhanced path feature vector are aggregated to generate a comprehensive feature for characterizing the package billing attributes.

4. The intelligent weighing and billing method for express delivery according to claim 3, characterized in that, The pre-trained billing attribute feature generation model performs nonlinear transformations and feature enhancements on the weight feature vector, volume feature vector, and path feature vector to obtain enhanced weight feature vector, enhanced volume feature vector, and enhanced path feature vector, including: A billing attribute feature generation model is constructed and trained based on a neural network. The billing attribute feature generation model includes an input layer, a parallel fully connected network layer, a feature enhancement module layer based on an attention mechanism, and an output layer. The input layer maps the weight feature vector, volume feature vector, and path feature vector into high-dimensional weight features, high-dimensional volume features, and high-dimensional path features, respectively. Then, a parallel fully connected network layer performs nonlinear transformations on the high-dimensional weight features, high-dimensional volume features, and high-dimensional path features to obtain various features after nonlinear transformation. The feature enhancement module layer enhances various features after nonlinear transformation to obtain enhanced weight feature vector, enhanced volume feature vector, and enhanced path feature vector. The enhanced weight feature vector, enhanced volume feature vector, and enhanced path feature vector are then output through the output layer.

5. The intelligent weighing and billing method for express delivery according to claim 1, characterized in that, The step of obtaining real-time service parameters, matching the comprehensive features and the real-time service parameters with pre-designed billing rules, and determining the dynamic billing rules that match the package to be billed includes: Construct a billing rule query knowledge base, which includes multi-dimensional pre-designed billing rules; Obtain real-time business parameters, and generate a query request based on the comprehensive features and the real-time business parameters; Based on the query request, a matching query is performed in the billing rule query knowledge base to determine the dynamic billing rule that matches the package to be billed.

6. The intelligent weighing and billing method for express delivery according to claim 1, characterized in that, The step of calculating the predicted shipping cost of the package to be billed based on the dynamic billing rules and the comprehensive features includes: The rate formula is determined based on dynamic billing rules; Substitute the weight and volume values ​​of the comprehensive characteristics into the rate formula to calculate the basic shipping cost of the package to be billed; By combining the route identification information of the comprehensive features, the basic freight cost is adjusted for timeliness to obtain the predicted freight cost.

7. The intelligent weighing and billing method for express delivery according to claim 1, characterized in that, After calculating the predicted shipping cost of the package to be billed, the method further includes collecting packages whose actual shipping cost and predicted shipping cost differ from the predicted shipping cost by more than a preset difference threshold as difference samples, and using the difference samples to optimize the pre-trained billing attribute feature generation model to obtain an optimized billing attribute feature generation model. The process of collecting packages whose actual shipping cost and predicted shipping cost differ from the predicted shipping cost by more than a preset difference threshold as difference samples, and using the difference samples to optimize the pre-trained billing attribute feature generation model to obtain the optimized billing attribute feature generation model includes: Collect multiple packages to be billed where the difference between the actual and predicted shipping costs exceeds a preset difference threshold as sample packages, record the associated data corresponding to each sample package, and use the associated data corresponding to each sample package to construct a difference sample set. The associated data includes initial features, comprehensive features, dynamic billing rules, and actual shipping costs. Using the initial features in the differential sample set as input and the billing attributes included in the corresponding actual freight as the optimization objective, the pre-trained billing attribute feature generation model is retrained in a supervised manner to obtain the optimized billing attribute feature generation model. The key factors causing deviations in rule matching or rate calculation in the differential sample set are analyzed, and at least one operation is performed on the preset billing rule query knowledge base based on the analysis results, such as correcting the calculation parameters of existing rules, supplementing new constraint rules, or adjusting the priority weights between rules, to generate an optimized billing rule query knowledge base.

8. A smart weighing and billing device for express delivery, characterized in that, include: The extraction module is used to acquire multidimensional data of the package to be billed and extract the initial features of the multidimensional data, including weight value, volume value and path identification information; The generation module is used to generate comprehensive features for characterizing package billing attributes based on the initial features using a pre-trained billing attribute feature generation model. The matching module is used to obtain real-time business parameters, match the comprehensive features and the real-time business parameters with the pre-designed billing rules, and determine the dynamic billing rules that match the package to be billed. The calculation module is used to calculate the predicted shipping cost of the package to be billed based on the dynamic billing rules and the comprehensive features.

9. A smart weighing and billing device for express delivery, characterized in that, It includes a memory and at least one processor, wherein the memory stores computer-readable instructions; The at least one processor invokes the computer-readable instructions in the memory to perform the steps of the express delivery intelligent weighing and billing method as described in any one of claims 1-7.

10. A computer-readable storage medium storing computer-readable instructions thereon, characterized in that, When the computer-readable instructions are executed by the processor, they implement the various steps of the intelligent weighing and billing method for express delivery as described in any one of claims 1-7.

Images