A weighing system for measuring logistics goods

Abstract

The present application relates to the technical field of logistics cargo measurement, and discloses a weighing measurement system for logistics cargo. The system comprises a cargo information intelligent sensing module, a dynamic rule engine, a correlation graph construction module and a comprehensive risk assessment module. The intelligent sensing module collects multi-dimensional original parameters through multi-source sensors, filters noise and fuses data to generate accurate and rich enhanced cargo data. The dynamic rule engine loads matching weighing rules accordingly, calculates the applicability index to form a highly adaptive application result. The correlation graph module maps according to the cargo identifier and the transportation path, quantifies the node connection strength and clarifies the cargo correlation. The comprehensive risk assessment module integrates the result to analyze the risk propagation path and generates a comprehensive assessment report. The system solves the short board of traditional equipment, realizes accurate acquisition of cargo information and comprehensive risk control, and improves the intelligent level of logistics weighing.

Description

Technical Field

[0001] This invention relates to the field of logistics cargo measurement technology, specifically a weighing and measuring system for logistics cargo. Background Technology

[0002] In the rapid development of the modern logistics industry, the weighing and measurement data of goods is the fundamental information that runs through the entire process of warehousing management, transportation scheduling, and cost accounting, directly affecting the efficiency and standardization of logistics operations. With the continuous expansion of logistics scale and the increasing complexity of goods, from lightweight document packages to heavy industrial components, from standardized goods with regular shapes to irregularly shaped special items, higher requirements are placed on the accuracy and adaptability of weighing and measurement.

[0003] Most current cargo weighing and measuring devices on the market are based on a single sensor, capable of collecting only basic physical parameters such as weight. This limited data dimension makes it difficult to comprehensively reflect the actual attributes of the goods. Furthermore, these devices generally lack effective data processing mechanisms. The raw data collected by the sensors is susceptible to environmental vibrations, electromagnetic interference, and other factors, resulting in significant noise and potentially large data deviations when used directly. Regarding the application of weighing rules, traditional systems often employ fixed rule settings. Pre-entered weighing standards cannot be dynamically adjusted according to the specific attributes of the goods. For goods with special materials or irregular shapes, rule mismatches frequently occur, leading to reduced reliability of the weighing results.

[0004] Existing technologies often treat the weighing process of each item in isolation, ignoring the interrelationships between goods, such as the transportation routes of the same batch of goods or the hierarchical relationships between goods in the upstream and downstream supply chains. This isolated approach prevents logistics companies from grasping the overall flow of goods. When a weighing anomaly occurs in one item, it is difficult to quickly pinpoint the root cause of the problem and the potential impact on other goods. In the risk assessment stage, traditional systems often rely solely on single weighing data, lacking consideration of interrelated factors and failing to comprehensively identify potential risks during the weighing process. This results in biased risk assessments that are unreliable for providing reliable references for logistics decisions. These problems make existing weighing and measurement systems unable to meet the demands of modern logistics for efficient, accurate, and comprehensive acquisition of cargo information and risk management, thus hindering the intelligent development of the logistics industry. Summary of the Invention

[0005] The purpose of this invention is to provide a weighing and measuring system for logistics goods to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a weighing and measuring system for logistics goods, the system comprising:

[0007] The cargo information intelligent sensing module is used to collect the raw physical parameters of logistics cargo through multi-source sensors, perform noise filtering and data fusion processing on the raw physical parameters, and generate enhanced cargo data.

[0008] The dynamic rule engine module is used to dynamically load a set of weighing rules that match the attributes of the goods from the rule base based on enhanced cargo data, calculate the applicability index of the weighing rule set, and generate a set of rule application results.

[0009] The association graph construction module is used to construct an association graph between cargo entities based on cargo identifiers and transportation route information in the enhanced cargo data, and to calculate the connection strength between nodes in the association graph.

[0010] The comprehensive risk assessment module is used to integrate the connection strength of the rule application result set and the correlation graph, conduct risk propagation path analysis, and generate a weighing risk assessment report.

[0011] Preferably, the intelligent cargo information sensing module is implemented in the following ways:

[0012] The raw weight value, raw size data and real-time location information of logistics goods are collected by weight sensors, size scanners and positioning devices to form a set of raw physical parameters;

[0013] The original set of physical parameters is processed by Kalman filtering to eliminate measurement noise and obtain the filtered physical parameters.

[0014] The filtered physical parameters are fused from multiple sources, and a weighted average algorithm is used to calculate the confidence level of each physical parameter. The parameter values ​​are then adjusted based on the confidence level to generate enhanced cargo data.

[0015] Preferably, the dynamic rule engine module is implemented in the following ways:

[0016] Extract cargo type codes and weight thresholds from enhanced cargo data, perform fuzzy matching in the rule base, and retrieve all relevant weighing rules;

[0017] The retrieved weighing rules are parsed, the condition expressions and constraint values ​​of the rules are extracted, and the matching score between each rule and the cargo attributes is calculated.

[0018] The rules are sorted according to their matching scores, and the top-ranked rules are selected to form a set of weighing rules. The average matching score of the set of weighing rules is then calculated as an applicability index.

[0019] The applicability index is logically combined with the rule condition expression to generate a set of rule application results.

[0020] Preferably, the process of generating the rule application result set includes:

[0021] Each rule in the weighing rule set is compared with the weight and size values ​​in the enhanced cargo data item by item to check whether the weight value is within the threshold range of the rule constraint and whether the size value meets the tolerance standard defined by the rule.

[0022] Record the comparison results for each rule, including compliance status, deviation value, and confidence level;

[0023] The comparison results of all rules are summarized, sorted according to rule priority, and a set of rule application results is generated.

[0024] Preferably, the implementation of the association map construction module includes:

[0025] Extract cargo identifiers, transport route nodes, and transport vehicle numbers from the enhanced cargo data, and uniquely encode each entity;

[0026] Calculate the co-occurrence frequency between cargo identifiers, the overlap between transport route nodes, and the number of times transport vehicle numbers are shared;

[0027] The correlation between entities is calculated based on co-occurrence frequency, overlap, and number of sharing, which serves as the connection strength.

[0028] Using cargo entities as nodes and connection strength as edge weights, construct an association graph.

[0029] Preferably, the process of constructing the association map includes:

[0030] Cluster analysis is performed on cargo identifiers to group similar identifiers into entity classes;

[0031] Calculate the center point for the identifier within each entity class, and use it as a graph node;

[0032] Calculate the path similarity between nodes and adjust the connection strength based on the path similarity;

[0033] Use a graph database to store nodes and edge weights to generate a relational graph.

[0034] Preferably, the comprehensive risk assessment module is implemented in the following ways:

[0035] Extract the compliance status and deviation value of each rule from the rule application result set, and calculate the basic risk score of a single cargo;

[0036] Extract cargo-related connection paths from the association graph and calculate the risk propagation coefficient based on the connection strength;

[0037] By combining the basic risk score and the risk propagation coefficient, a weighted fusion is performed to obtain the comprehensive risk value;

[0038] A weighing risk assessment report is generated based on the comprehensive risk value, including the risk level and the source of the risk.

[0039] Preferably, the process of generating the enhanced cargo data includes:

[0040] Time series analysis is performed on the filtered physical parameters to detect outliers.

[0041] Interpolation compensation is performed on outliers to ensure data continuity;

[0042] The compensated parameters are compared with historical data to calculate the trend of parameter changes;

[0043] By incorporating the changing trends into the parameter values, enhanced cargo data is generated.

[0044] Preferably, the dynamic loading process of the weighing rule set includes:

[0045] Monitor rule base update events and automatically reload rules when new rules are added;

[0046] The matching strategy is adjusted in real time based on changes in the cargo type code;

[0047] Use a caching mechanism to store frequently used rules to improve loading speed.

[0048] Preferably, the calculation process of the comprehensive risk value includes:

[0049] The basic risk score is normalized to fall within the range of zero to one.

[0050] Logarithmic transformation of the risk propagation coefficient reduces the impact of extreme values;

[0051] A linear weighting method is used to fuse the normalized basic risk score and the transformed risk propagation coefficient;

[0052] Based on the fusion results, risk levels are classified, and a weighing risk assessment report is generated.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] At the cargo data acquisition level, the intelligent cargo information sensing module adopts a multi-source sensor fusion acquisition method, breaking through the limitations of traditional single sensors. It can simultaneously acquire multiple dimensions of raw physical parameters of the cargo, such as weight, volume, and density, achieving comprehensive cargo information collection. Addressing the noise issue in the raw data, the module's built-in noise filtering algorithm effectively removes abnormal data caused by environmental interference. Furthermore, through data fusion technology, the multi-source data is complementary and integrated, generating enhanced cargo data that is not only rich in dimensions but also has improved accuracy and reliability, providing a high-quality data foundation for subsequent weighing rule matching and risk assessment.

[0055] The application of the dynamic rule engine module solves the problem of fixed weighing rules in traditional systems. This module can dynamically filter and load matching weighing rule sets from the rule base based on specific attributes such as material, size, and weight class contained in the enhanced cargo data, avoiding weighing errors caused by mismatches between rules and cargo attributes. By calculating the applicability index of the weighing rule set, the module prioritizes the weighing rules that best suit the current cargo, further ensuring the rationality of the weighing results. This allows the system to flexibly adapt to different types and characteristics of cargo, expanding its applicability and improving its practicality in complex logistics scenarios.

[0056] The association graph construction module links isolated cargo information, constructing an association graph between cargo entities based on cargo identifiers and transportation route information, clearly presenting the flow relationships of cargo within the logistics network. The calculation of node connection strength quantifies the tightness of association between cargo, helping logistics companies grasp the overall association status of cargo. When a cargo experiences a weighing anomaly, the association graph can quickly locate other potentially affected cargo, providing clear guidance for problem tracing and subsequent handling, and improving the coordination of logistics operations and the efficiency of problem handling.

[0057] The comprehensive risk assessment module abandons the traditional single-data assessment model. By integrating the result set of rule applications and the connection strength of the correlation graph, it achieves multi-dimensional analysis of weighing risks. The analysis of risk propagation paths clearly shows the diffusion path of risks among related goods. The generated weighing risk assessment report not only includes the weighing risk of individual goods but also covers the overall impact range of the risks. This enables logistics companies to comprehensively and accurately identify potential risks in the weighing process, take targeted prevention and control measures in advance, reduce logistics disputes and increased costs caused by weighing problems, and promote the development of intelligent and refined logistics weighing and measurement. Attached Figure Description

[0058] Figure 1 This is a schematic diagram illustrating the working principle of the weighing and measuring system for logistics goods according to the present invention.

[0059] Figure 2 A flowchart illustrating the operation of the intelligent cargo information sensing module;

[0060] Figure 3 A flowchart illustrating how the dynamic rules engine module works;

[0061] Figure 4 An optimized histogram of entity connection strength distribution;

[0062] Figure 5 This is a scatter plot showing the relationship between the basic risk score and the risk propagation coefficient. Detailed Implementation

[0063] 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.

[0064] Please see Figure 1 This invention provides a weighing and measurement system for logistics goods. The system includes: a cargo information intelligent sensing module responsible for acquiring the original physical parameters of the cargo using multi-source sensors, performing noise filtering and data fusion to produce enhanced cargo data; a dynamic rule engine module dynamically matching weighing rules in a rule base based on the enhanced cargo data, calculating applicability indices, and generating a set of rule application results; an association graph construction module parsing cargo identifiers and transportation route information, constructing an entity association graph, and quantifying connection strength; and a comprehensive risk assessment module integrating rule application results and association graph data, performing risk propagation analysis, and finally generating a weighing risk assessment report.

[0065] Example 1: See Figure 2 In practice, a high-precision strain gauge sensor is installed on the conveyor belt support structure to acquire the raw weight value of the goods at a sampling frequency of 100 times per second. A dimension scanner uses 3D LiDAR scanning technology to capture the length, width, and height point cloud data of the goods. A positioning device integrates a dual-frequency GPS receiver module to record the real-time latitude and longitude coordinates and timestamp information of the goods. These raw physical parameters are aggregated to form a raw physical parameter set. This raw physical parameter set is transmitted to the data processing unit for Kalman filtering. The Kalman filtering algorithm is based on the state equation and observation equation of a linear dynamic system. Through prediction and update steps, it recursively estimates the true values ​​of the parameters, effectively eliminating random noise introduced during the measurement process, such as fluctuation noise in weight values ​​and discrete errors in dimensional data. The filtered physical parameters exhibit higher stability and accuracy.

[0066] In practical implementation, the filtered physical parameters enter the multi-source data fusion stage. A weighted average algorithm is used to calculate the confidence level of each physical parameter. The weight values ​​are dynamically allocated based on the accuracy level of the sensor devices and historical calibration data. For example, the confidence level of a weight sensor is determined based on the error range in its calibration certificate, the confidence level of a dimensional scanner depends on the density and completeness of the point cloud data, and the confidence level of a positioning device considers the influence of GPS signal strength and multipath effects. The parameter values ​​are then weighted and fused according to the confidence levels to generate enhanced cargo data. This enhanced cargo data not only includes the corrected parameter values ​​but also includes confidence level labels, providing reliable input for subsequent modules. The generation process of enhanced cargo data further includes time series analysis, performing trend detection on the filtered physical parameters over continuous time points. A sliding window mechanism is used to calculate the moving average and standard deviation of the parameters, identifying outliers that deviate from the normal range, such as sudden jumps in weight values ​​or abnormal shifts in dimensional data.

[0067] In some embodiments, time series analysis uses a statistical process control-based method to set dynamic thresholds. Outliers are marked and then compensated by interpolation. Linear interpolation or cubic spline interpolation is used to ensure the continuity and smoothness of data points. The compensated parameters are compared with similar cargo data in a historical database. Historical data is stored in a distributed database, and parameter records for similar cargo are retrieved through a query interface. The comparison process calculates parameter change trends, for example, using linear regression to analyze the long-term drift trend of weight values ​​or exponential smoothing to predict seasonal changes in size values. The change trend is quantified as a trend coefficient and incorporated into the current parameter values ​​to generate the final enhanced cargo data. The enhanced cargo data is stored in a structured format, containing timestamps, parameter values, confidence levels, and trend information, supporting efficient querying and updating. The hardware deployment of the intelligent cargo information sensing module needs to consider industrial environment adaptability. Weight sensors are mounted on shockproof bases to resist mechanical vibration interference, and size scanners are equipped with protective covers to prevent dust and light interference. Positioning devices communicate with the central server via wired or wireless networks to ensure real-time and reliable data transmission. In practical implementation, the sensor data acquisition unit adopts an embedded system design, running a real-time operating system to process high-frequency data streams. The data preprocessing module uses digital signal processing technology to optimize the computational efficiency of filtering algorithms. The multi-source data fusion module is deployed on edge computing nodes to reduce network transmission latency. The software implementation of the cargo information intelligent sensing module adopts a modular architecture. The data acquisition submodule is responsible for sensor driving and data caching, the filtering submodule implements the Kalman filtering algorithm, the data fusion submodule manages weight calculation and fusion logic, and the time series analysis submodule performs anomaly detection and trend calculation. All submodules communicate asynchronously through message queues to ensure the system's high concurrency processing capability.

[0068] Optionally, the generation process of enhanced cargo data can integrate machine learning methods to assist in anomaly detection, such as using the isolated forest algorithm to identify abnormal patterns in multidimensional parameters. However, the core implementation still relies primarily on statistical methods to ensure the transparency and interpretability of the process. The output interface of the intelligent cargo information sensing module provides a standardized data format, facilitating parsing and use by the dynamic rule engine module. In practical implementation, the module's performance monitoring includes sensor health status checks and data quality assessments, regularly generating diagnostic reports. Maintenance personnel use these reports to perform equipment calibration and software updates, ensuring long-term stable system operation. The overall workflow of the intelligent cargo information sensing module follows Industrial Internet of Things (IIoT) standards, with data acquisition, filtering, fusion, and enhancement steps processed in a streamlined manner. It supports both batch processing and real-time stream processing modes to adapt to the needs of different logistics scenarios. The implementation of the intelligent cargo information sensing module relies on precise sensor calibration and algorithm parameter optimization. The calibration process is conducted in a laboratory environment, using standard weight blocks and dimensional reference objects to calibrate the sensor output. Algorithm parameters, such as the process noise covariance and observation noise covariance of the Kalman filter, are determined through experimental data training. In practical implementation, the weight adjustment strategy for multi-source data fusion can be dynamically optimized according to the actual application scenario. For example, in high-speed sorting scenarios, the weight of weight parameters can be increased, while in large-item cargo scenarios, the emphasis can be placed on size parameters. The generation cycle of enhanced cargo data can be synchronized with the logistics operation rhythm to ensure the timeliness of the evaluation results. As the data entry point of the system, the reliability and accuracy of the cargo information intelligent sensing module directly affect the decision-making quality of subsequent modules. Therefore, both hardware selection and software implementation must undergo rigorous testing and verification.

[0069] Example 2: See Figure 3 In its implementation, the dynamic rule engine module extracts cargo type codes and weight thresholds from the enhanced cargo data. The cargo type codes employ standardized classification systems such as UNSPSC or custom hierarchical coding, while the weight thresholds are preset upper and lower weight limits parsed from the cargo information. The module then performs fuzzy matching in a rule base to retrieve all relevant weighing rules. This rule base is stored using a relational database management system, and its table structure includes rule identifiers, conditional expressions, constraint values, priority fields, and effective timestamps. The fuzzy matching algorithm is based on string similarity calculation and numerical range lookup. The cargo type codes are compared with predefined cargo type patterns in the rule base, using either edit distance or cosine similarity algorithms. The weight thresholds are intersected with the weight constraint ranges stored in the rule base, and the retrieved relevant weighing rules form the initial rule set.

[0070] In practice, the retrieved weighing rules are parsed to extract their conditional expressions and constraint values. The conditional expression is a set of predicates connected by logical operators, such as a combination of weight greater than a certain value and size less than a certain value. The constraint value is a specific numerical threshold involved in the conditional expression. The parsing process uses a parser to convert the conditional expressions into an abstract syntax tree, facilitating subsequent calculation of the matching score. Each rule is then calculated to match the cargo attributes. The matching score is a numerical indicator quantifying the rule's applicability. This calculation involves substituting actual parameter values ​​from the enhanced cargo data into the rule's conditional expression for partial evaluation and assigning a score based on the degree to which the conditions are met. The rules are then sorted according to their matching scores, using descending order to place rules with higher matching scores at the top of the list. The top-ranked rules form the weighing rule set, the size of which can be determined by a preset threshold or dynamically adjusted rule count. The average matching score of the weighing rule set is calculated as an applicability indicator, reflecting the overall matching quality of the rule set. Finally, the applicability indicator is logically combined with the rule conditional expressions to generate a rule application result set. This logical combination process integrates the applicability indicator as a weighting factor with the conditional judgment results of each rule.

[0071] In some embodiments, the process of generating the rule application result set includes comparing each rule in the weighing rule set with the weight and size values ​​in the enhanced cargo data item by item. The comparison operation checks whether the weight value is within the threshold range of the rule constraint and whether the size value meets the tolerance standard defined by the rule. The tolerance standard is the allowable deviation range of the size specified in the rule, such as a tolerance of ± a certain percentage for length. The comparison result for each rule is recorded, containing three elements: compliance status, deviation value, and confidence level. The compliance status is a Boolean value indicating whether the rule condition is met; the deviation value is the absolute or relative difference between the actual measured value and the rule threshold; and the confidence level is a probabilistic estimate based on the reliability of the data source. The comparison results of all rules are summarized. The summarization process collects the comparison results of all rules and sorts them according to rule priority. Rule priority is a preset importance indicator for each rule in the rule base. After sorting by rule priority, a rule application result set is generated. The rule application result set is a structured data list, where each entry contains a rule identifier, compliance status, deviation value, confidence level, and priority information.

[0072] It's understandable that performance optimization of the dynamic rule engine module involves multiple aspects. The indexing strategy of the rule base directly impacts the efficiency of fuzzy matching, requiring a full-text index on the cargo type code field and a range index on the weight threshold field. Matching score calculation can employ a caching mechanism, pre-calculating matching scores for frequently occurring cargo type and weight combinations. The generation of the rule application result set can be parallelized, distributing comparison tasks for different rules to multiple computing nodes for simultaneous execution. The dynamic rule engine module needs to be real-time, capable of quickly updating the weighing rule set when new rules are added to the rule base. This is achieved through database triggers or message queues listening for rule base change events.

[0073] Optionally, the generation of the rule application result set can incorporate uncertain reasoning methods to handle fuzzy or conflicting rule conditions, such as using DS evidence theory to merge the support of multiple rules. However, the core implementation remains primarily based on deterministic logic judgments to ensure the transparency and traceability of the rule evaluation process. The output interface of the dynamic rule engine module provides a standardized data serialization format, and the rule application result set is converted into JSON or XML format for use by subsequent modules. In specific implementations, the monitoring and logging functions of the dynamic rule engine module record the detailed process of rule matching, including the number of rules retrieved, the distribution of matching scores, and rule application result statistics. These logs are used for continuous optimization of the rule base and system performance analysis.

[0074] It is understandable that the implementation of the dynamic rule engine module is highly dependent on the quality and completeness of the rule base, which needs to be maintained and updated regularly to reflect changes in business rules. The design of rule condition expressions needs to balance expressiveness and computational complexity, avoiding overly complex logical conditions that could impact real-time performance. The matching score calculation model needs to be calibrated according to actual application scenarios to ensure that high-scoring rules truly represent high applicability. The interface between the dynamic rule engine module and the intelligent cargo information perception module needs to define a clear data contract, ensuring that the fields of the enhanced cargo data are consistent with the variables in the rule condition expressions. As the decision-making core of the system, the accuracy and efficiency of the dynamic rule engine module are crucial to overall performance; therefore, it requires thorough testing and verification.

[0075] Example 3: In specific implementation, the association graph construction module extracts cargo identifiers, transportation route nodes, and transportation vehicle numbers from the enhanced cargo data. The cargo identifier is a unique code identifying a single item or batch of cargo; the transportation route node is the code of the physical location points the cargo passes through in the logistics network; and the transportation vehicle number is the identifier of the transportation unit carrying the cargo. Each entity is uniquely coded, using international standards or custom naming rules. The cargo identifier, transportation route node, and transportation vehicle number are mapped to a unified resource identifier format. The co-occurrence frequency between cargo identifiers is calculated, based on historical transportation order data and real-time logistics event streams. The overlap between transportation route nodes is calculated using a set similarity measurement method. The number of times transportation vehicle numbers are shared is calculated, extracted from logistics transportation records. The association degree between entities is calculated based on co-occurrence frequency, overlap, and sharing frequency, serving as the connection strength. The association degree calculation is a multi-factor fusion process, where connection strength... The calculation can be expressed using the following relation:

[0076]

[0077] in: This represents the connection strength between entity i and entity j. This represents the co-occurrence frequency of entity i and entity j. This represents the maximum co-occurrence frequency of all entity pairs in the system. This represents the overlap of transportation path nodes between entity i and entity j, with a value between 0 and 1. It also represents the number of times the transportation vehicle IDs of entity i and entity j are shared. This represents the maximum number of times all entity pairs in the system share data. , and These are the weighting coefficients for co-occurrence frequency, overlap, and number of sharing events, respectively, and satisfy the following conditions: .

[0078] The construction process of the association graph includes clustering analysis of cargo identifiers using a density-based clustering algorithm to group cargo identifiers with similar characteristics into entity classes. Centroids are calculated for identifiers within each entity class, serving as graph nodes; the centroid calculation uses the mean method. Path similarity between nodes is calculated based on a sequence comparison of cargo transportation routes. Connection strength is adjusted based on path similarity, with stronger connections between nodes exhibiting high path similarity. A graph database is used to store nodes and edge weights, generating the association graph.

[0079] In some embodiments, the association graph is stored using an attribute graph model, where nodes have labels and attributes, and edges have relation type and weight attributes. The construction of the association graph is an incremental process, dynamically updating nodes and edges as new cargo data continuously flows in. The query interface of the association graph supports complex network analysis. The computational complexity of the association graph construction module is high, requiring the support of a distributed computing framework. The selection of the graph database considers scalability and query performance. Data consistency of the association graph is guaranteed through a transaction mechanism.

[0080] It is understandable that a time decay factor can be introduced into the calculation of connection strength, making the contribution of recently occurring co-occurring, overlapping, and shared events to connection strength greater. The calculation of path similarity can consider the network topology of transportation paths. The visualization module of the association graph can provide a graphical interface. The association graph construction module can be integrated with external knowledge graphs to enrich entity attributes and relationship types. The construction cycle of the association graph can be set according to business needs. Performance monitoring of the association graph construction module includes indicators such as graph size statistics, construction time, and memory usage. The parameter selection of the node clustering algorithm needs to be adjusted according to the distribution of cargo data. The dynamic update mechanism of edge weights needs to balance real-time performance and computational overhead. The interface design between the association graph construction module and the front-end and back-end modules needs to consider data serialization efficiency and network transmission load. The security and access control of the association graph need to be implemented through a permission management mechanism. Long-term maintenance of the association graph includes data archiving, version management, and backup and recovery strategies.

[0081] See Figure 4 This chart focuses on the association graph construction module of the logistics cargo weighing and measurement system, and is a key visualization result for quantifying the tightness of the connections between cargo entities. The horizontal axis, entity connection strength, is generated by integrating three dimensions: the co-occurrence frequency of cargo identifiers, the overlap of transportation route nodes, and the number of times transportation vehicle numbers are shared; the vertical axis, frequency, counts the number of cargo entity pairs in different connection strength ranges. This chart provides core indicators for the subsequent comprehensive risk assessment module to analyze risk propagation paths: the higher the connection strength, the stronger the potential for risk diffusion between entities. It reflects the technological breakthrough of the system from isolated analysis of single cargo to global insight into network connections, and is a key data support for achieving comprehensive and refined assessment of logistics weighing risks. It also verifies the effectiveness of the association graph construction module in uncovering implicit cargo connections and quantifying connection strength.

[0082] Example 4: In specific implementation, the comprehensive risk assessment module extracts the compliance status and deviation value of each rule from the rule application result set. The compliance status is a Boolean logic value, indicating whether the cargo parameters meet the conditions of a specific weighing rule. The deviation value is a numerical variable, quantifying the difference between the actual measured value and the rule threshold. A basic risk score for each cargo is calculated using a weighted scoring model. The compliance status of each rule is converted into a risk contribution value. A true compliance status results in a zero risk contribution value, while a false compliance status results in a positive risk contribution value. The magnitude of the risk contribution value is proportional to the priority weight of the rule. The deviation value is mapped to a risk multiplier through a linear or nonlinear function. The basic risk score is the weighted sum of the risk contribution values ​​of each rule and their corresponding risk multipliers.

[0083] In practical implementation, the comprehensive risk assessment module extracts connection paths related to the currently assessed goods from the association graph. This extraction employs a graph traversal algorithm, starting with the node corresponding to the current goods and exploring connected nodes in the association graph using a depth-first or breadth-first strategy. The traversal depth is set to two- or three-degree association based on actual needs. A risk propagation coefficient is calculated based on connection strength, representing the diffusion capacity of risk factors in the association network. This calculation involves traversing all connection paths and aggregating the connection strength of each edge along the path. Aggregation methods include taking the maximum, minimum, or arithmetic mean of the path values. A weighted fusion is then performed by combining the basic risk score and the risk propagation coefficient to obtain a comprehensive risk value. This weighted fusion uses a linear combination model, where the basic risk score and the risk propagation coefficient are multiplied by configured weights and then summed. These configured weights are dynamically adjusted based on the risk preferences of the business scenario.

[0084] Referring to Table 1, the calculation process of the comprehensive risk value includes: normalizing the basic risk score to fall within the range of zero to one; using a min-max scaling method to determine the range of the basic risk score based on historical data or preset theoretical extreme values; performing a logarithmic transformation on the risk propagation coefficient to reduce the impact of extreme values, using the natural logarithm or commonly used logarithm to compress the long-tailed distribution of the risk propagation coefficient into a smaller numerical range; and using a linear weighting method to fuse the normalized basic risk score and the transformed risk propagation coefficient, with the weighting coefficients determined through expert experience or machine learning training. Based on the fusion results, risk levels are classified, and a weighing risk assessment report is generated. Risk level classification uses a threshold method, setting multiple risk intervals corresponding to different level labels.

[0085] Table 1: Risk Level Classification Standards

[0086] Comprehensive risk range Risk level label Recommended treatment code [0.0,0.3) Low risk NOR001 [0.3,0.7) Medium risk CHE002 [0.7,1.0] High risk INV003

[0087] It is understandable that the algorithm implementation of the comprehensive risk assessment module needs to consider computational efficiency. For large-scale cargo assessment scenarios, the calculation of basic risk scores can be parallelized, and the calculation tasks of risk contribution values ​​under different rules can be distributed to multiple computing units. The calculation of the risk propagation coefficient can introduce approximation algorithms; when the correlation graph is large, the propagation coefficient can be estimated through random walks or sampling methods. The update mechanism of the comprehensive risk value supports real-time streaming processing, incrementally updating the risk assessment results as new data arrives. The output interface of the comprehensive risk assessment module provides a structured data format, and the weighing risk assessment report includes header information, risk assessment details, and decision support information.

[0088] In some embodiments, the risk level classification criteria can be configured as a multi-dimensional matrix, considering not only the comprehensive risk value but also the type distribution and spatiotemporal characteristics of risk factors. The generation of the weighing risk assessment report supports template customization, outputting content with varying levels of detail based on different user roles. Integration of the comprehensive risk assessment module with external systems is achieved through a standard application programming interface (API), allowing risk assessment results to be pushed to inventory management systems or transportation scheduling systems. Performance metrics of the comprehensive risk assessment module include assessment accuracy, response time, and system throughput, which are collected and analyzed in real time through monitoring components.

[0089] Optionally, the calculation of the basic risk score can incorporate fuzzy logic to handle conflicts between rules. When multiple rules lead to contradictory conclusions, the comprehensive risk contribution is calculated using a membership function. The calculation of the risk propagation coefficient can consider a path length attenuation factor; nodes farther from the current cargo node have lower risk impact weights. The fusion model for the comprehensive risk value can be extended to a nonlinear function, such as a neural network model, but sufficient training data is required. The configuration management interface of the comprehensive risk assessment module allows authorized users to adjust weight parameters and risk level thresholds to adapt to changes in business strategies. The digital signature mechanism of the weighing risk assessment report ensures the integrity and non-repudiation of the report; a hash algorithm is used to generate a digital fingerprint of the report content. The audit log of the comprehensive risk assessment module records the calculation parameters and intermediate results of each assessment task, supporting traceability and review of the risk assessment process. The fault-tolerant processing mechanism of the comprehensive risk assessment module includes data verification, anomaly capture, and fault recovery. When input data is abnormal or the calculation process fails, the system can record error information and execute predefined emergency procedures.

[0090] Understandably, the implementation of the comprehensive risk assessment module requires establishing a reliable data exchange protocol with the front-end module. Rule application result sets and correlation graph data are transmitted via message queues or remote procedure calls. The deployment architecture of the comprehensive risk assessment module supports horizontal scaling, distributing assessment tasks to multiple risk assessment instances through load balancing. Testing and verification of the comprehensive risk assessment module requires building test case sets covering various risk scenarios, including normal, boundary, and abnormal conditions, to ensure the correctness and robustness of the assessment logic. Version management of the comprehensive risk assessment module follows software engineering standards; the codebase uses a branch strategy to manage development and defect fixing, and changes to the core algorithm of the risk assessment engine require rigorous regression testing.

[0091] See Figure 5 This chart focuses on the comprehensive risk assessment module of the logistics cargo weighing and measurement system, serving as a core visualization result for quantifying the correlation between individual basic risks and network propagation risks in cargo weighing. The horizontal axis, the basic risk score, originates from the rule application result set and is a quantitative assessment of information such as the rule compliance status and deviation values ​​of individual cargo. The vertical axis, the risk propagation coefficient, comes from the connection strength analysis of the correlation graph, reflecting the ability of risk to spread within the cargo entity network. The color bar on the right, representing the comprehensive risk value, is the final risk result obtained by weighted fusion of the basic risk score and the risk propagation coefficient; the color intensity intuitively reflects the level of risk. The data distribution shows that when both the basic risk score and the risk propagation coefficient are high, the comprehensive risk value increases significantly; conversely, when both are low, the comprehensive risk value is also low. This correlation intuitively presents the technical logic of the comprehensive risk assessment module integrating individual rule matching results and network correlation strength to assess weighing risks from multiple dimensions. It provides data support for logistics companies to accurately identify high-risk cargo and predict risk propagation paths, serving as a key tool for achieving accurate identification and overall control of weighing risks. This demonstrates the system's technological breakthrough from single-data assessment to multi-dimensional risk fusion.

[0092] Example 5: In specific implementation, rule base update event monitoring is achieved through a combination of database triggers and message queue mechanisms. For example, in a relational database management system, triggers are created for rule tables. When an insert, update, or delete operation occurs in the rule table, the trigger is activated and sends an event notification to the message middleware. The event notification includes the rule identifier and operation type. The event notification in the message queue is consumed by the background listening service of the dynamic rule engine module. The listening service parses the event content and calls the rule loading interface. The rule loading interface retrieves the complete rule definition from the rule base based on the rule identifier, including metadata such as the rule's condition expression, constraint values, and priority. The new rule is added to the rule set in memory, and existing rules are updated or removed according to the change type. When a new rule is added, the automatic reloading process involves version management and consistency control. The new rule is assigned a unique version number, and the rule loading interface determines whether the rule instance in memory needs to be updated by comparing the version number. Automatic rule reloading uses atomic operations to ensure the consistency of the rule set during the update process, avoiding inconsistent rule states. The matching strategy is adjusted in real time based on changes in cargo type codes. Changes in cargo type codes are detected in real time by monitoring the enhanced cargo data stream. The adjustment of the matching strategy includes rebuilding the rule index and updating the matching algorithm parameters. For example, when the cargo type code changes from "general cargo" to "dangerous goods", the matching strategy will increase the weight of rules related to dangerous goods and prioritize matching rules that contain dangerous goods constraints.

[0093] A caching mechanism is used to store frequently used rules, improving loading speed. This mechanism employs a Least Recently Used (LRU) algorithm to manage in-memory rule copies. Frequently used rules are determined based on historical matching frequency and recent access time. The cached rule copies are synchronized with the master records in the rule base, and cached content is updated through periodic expiration or event-driven methods. The dynamic loading process is integrated into the rule engine's runtime environment, ensuring seamless execution within the data processing pipeline. Upon receiving enhanced cargo data, the rule engine's workflow first checks the state of the rule set, triggering the rule loading process if necessary, before executing rule matching and application logic.

[0094] In some embodiments, rule base update events can be subdivided into different priorities. High-priority rule changes trigger immediate reloading, while low-priority rule changes can be delayed until the system is idle for batch processing. The automatic rule reloading process can be configured with a rollback mechanism; if a new rule fails to load or verification fails, the system automatically reverts to the previous stable version of the rule set. Real-time adjustments to the matching strategy can be based on machine learning models to predict trends in cargo type changes and preload potentially needed rule sets. The caching mechanism can employ a multi-level caching architecture, with the first-level cache stored in the rule engine process's memory and the second-level cache stored in a distributed caching system.

[0095] It's understandable that the dynamic loading process of the weighing rule set in the dynamic rule engine module needs to handle concurrent access issues. When multiple cargo data streams trigger rule loading simultaneously, locking mechanisms or transaction control are needed to ensure the integrity of the rule set. The monitoring of rule base update events needs high reliability to avoid event loss due to network interruptions or service failures. The performance impact of automatic rule reloading needs to be evaluated; overly frequent reloading may affect system throughput, requiring a balance between real-time performance and overall performance. The effectiveness of the caching mechanism depends on the cache hit rate, necessitating continuous optimization of the caching strategy through monitoring metrics.

[0096] Optionally, rule base update event monitoring can be integrated into the configuration management platform, managing rule changes and release processes through a graphical interface. The automatic rule reloading process can support a canary release strategy, where new rules are first tested in a small-scale cargo data stream, and then fully released after verification. Real-time adjustment of matching strategies can incorporate an A / B testing framework to compare the application effects of rules under different matching strategies. The caching mechanism can support rule hotspot prediction, preloading frequently used rules based on business cycles and seasonal patterns. The dynamic loading process of the weighing rule set in the dynamic rule engine module can record detailed operation logs, including rule loading time, number of rules loaded, loading time, and loading results; log data is used for performance analysis and troubleshooting. The version history of the rule set can be persistently stored, supporting quick rollback to any historical version. The memory usage of the caching mechanism can be monitored in real time, automatically triggering cache cleanup operations when memory usage reaches a threshold.

[0097] Optionally, the dynamic loading process of the weighing rule set in the dynamic rule engine module can be extended to a distributed deployment mode. Multiple rule engine instances share a unified rule base and caching service, and rule loading tasks are allocated through a consistent hashing algorithm. Rule base update events can be broadcast among multiple rule engine instances to ensure that the rule sets of all instances remain synchronized. Matching strategy adjustment parameters can be centrally configured and dynamically distributed to each rule engine instance through a configuration center. The dynamic loading process of the weighing rule set in the dynamic rule engine module is a crucial component of system maintainability. The rule base structure needs to support flexible query and update operations, and the rule definition language needs to be sufficiently expressive while maintaining parsability. The automatic rule reloading verification process needs to include syntax checking, logical verification, and performance testing to ensure that new rules do not cause system anomalies. The matching strategy adjustment algorithm needs to avoid overfitting historical data and maintain adaptability to new types of goods. The cache expiration strategy needs to consider the importance and change frequency of rules; the cache expiration time for critical rules should be set shorter, while the cache expiration time for ordinary rules can be appropriately extended. The dynamic loading process of the weighing rule set of the dynamic rule engine module and its collaboration with other modules require clear interface contract definitions, standardized input and output data formats, and standardized exception handling procedures.

[0098] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0099] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A weighing and measuring system for logistics goods, characterized in that, The system includes: The cargo information intelligent sensing module is used to collect the raw physical parameters of logistics cargo through multi-source sensors, perform noise filtering and data fusion processing on the raw physical parameters, and generate enhanced cargo data. The dynamic rule engine module is used to dynamically load a set of weighing rules that match the attributes of the goods from the rule base based on enhanced cargo data, calculate the applicability index of the weighing rule set, and generate a set of rule application results. The association graph construction module is used to construct an association graph between cargo entities based on cargo identifiers and transportation route information in the enhanced cargo data, and to calculate the connection strength between nodes in the association graph. The comprehensive risk assessment module is used to integrate the connection strength of the rule application result set and the correlation graph, conduct risk propagation path analysis, and generate a weighing risk assessment report. The implementation methods of the association graph construction module include: Extract cargo identifiers, transport route nodes, and transport vehicle numbers from the enhanced cargo data, and uniquely encode each entity; Calculate the co-occurrence frequency between cargo identifiers, the overlap between transport route nodes, and the number of times transport vehicle numbers are shared; The correlation between entities is calculated based on co-occurrence frequency, overlap, and number of sharing, which serves as the connection strength. Using cargo entities as nodes and connection strength as edge weights, construct an association graph; The implementation methods of the comprehensive risk assessment module include: Extract the compliance status and deviation value of each rule from the rule application result set, and calculate the basic risk score of a single cargo; Extract cargo-related connection paths from the association graph and calculate the risk propagation coefficient based on the connection strength; By combining the basic risk score and the risk propagation coefficient, a weighted fusion is performed to obtain the comprehensive risk value; A weighing risk assessment report is generated based on the comprehensive risk value, including the risk level and the source of the risk.

2. The weighing and measuring system for logistics goods according to claim 1, characterized in that, The intelligent cargo information sensing module is implemented in the following ways: The raw weight value, raw size data and real-time location information of logistics goods are collected by weight sensors, size scanners and positioning devices to form a set of raw physical parameters; The original set of physical parameters is processed by Kalman filtering to eliminate measurement noise and obtain the filtered physical parameters. The filtered physical parameters are fused from multiple sources, and a weighted average algorithm is used to calculate the confidence level of each physical parameter. The parameter values ​​are then adjusted based on the confidence level to generate enhanced cargo data.

3. The weighing and measuring system for logistics goods according to claim 1, characterized in that, The implementation methods of the dynamic rule engine module include: Extract cargo type codes and weight thresholds from enhanced cargo data, perform fuzzy matching in the rule base, and retrieve all relevant weighing rules; The retrieved weighing rules are parsed, the condition expressions and constraint values ​​of the rules are extracted, and the matching score between each rule and the cargo attributes is calculated. The rules are sorted according to their matching scores, and the top-ranked rules are selected to form a set of weighing rules. The average matching score of the set of weighing rules is then calculated as an applicability index. The applicability index is logically combined with the rule condition expression to generate a set of rule application results.

4. The weighing and measuring system for logistics goods according to claim 3, characterized in that, The process of generating the rule application result set includes: Each rule in the weighing rule set is compared with the weight and size values ​​in the enhanced cargo data item by item to check whether the weight value is within the threshold range of the rule constraint and whether the size value meets the tolerance standard defined by the rule. Record the comparison results for each rule, including compliance status, deviation value, and confidence level; The comparison results of all rules are summarized, sorted according to rule priority, and a set of rule application results is generated.

5. The weighing and measuring system for logistics goods according to claim 1, characterized in that, The process of constructing the association map includes: Cluster analysis is performed on cargo identifiers to group similar identifiers into entity classes; Calculate the center point for the identifier within each entity class, and use it as a graph node; Calculate the path similarity between nodes and adjust the connection strength based on the path similarity; Use a graph database to store nodes and edge weights to generate a relational graph.

6. The weighing and measuring system for logistics goods according to claim 2, characterized in that, The process of generating the enhanced cargo data includes: Time series analysis is performed on the filtered physical parameters to detect outliers. Interpolation compensation is performed on outliers to ensure data continuity; The compensated parameters are compared with historical data to calculate the trend of parameter changes; By incorporating the changing trends into the parameter values, enhanced cargo data is generated.

7. The weighing and measuring system for logistics goods according to claim 3, characterized in that, The dynamic loading process of the weighing rule set includes: Monitor rule base update events and automatically reload rules when new rules are added; The matching strategy is adjusted in real time based on changes in the cargo type code; Use a caching mechanism to store frequently used rules to improve loading speed.

8. The weighing and measuring system for logistics goods according to claim 1, characterized in that, The calculation process for the comprehensive risk value includes: The basic risk score is normalized to fall within the range of zero to one. Logarithmic transformation of the risk propagation coefficient reduces the impact of extreme values; A linear weighting method is used to fuse the normalized basic risk score and the transformed risk propagation coefficient; Based on the fusion results, risk levels are classified, and a weighing risk assessment report is generated.

Images