IoT-based logistics data monitoring and management system and methods

Abstract

This invention belongs to the field of IoT logistics management technology. It provides an IoT-based logistics data monitoring and management system and method, including: an event triggering module, a status summary module, an event aggregation module, and a situational early warning module. The system collects raw status data within the logistics network, sets up edge intelligent nodes with an integrated lightweight rule engine to determine whether event judgment rules are triggered. If triggered, the event data packet is encapsulated; if not triggered, a status summary is generated from the raw status data within a set aggregation time window. The window length of the aggregation time window is dynamically adjusted based on resource quantification assessment. The system receives and stores the event data packets and status summaries on the cloud service platform. An event aggregation analysis engine is set up and used to analyze the event data packets, aggregating and generating composite events. A situational awareness engine is constructed and used to perform real-time analysis and prediction based on the composite event stream, generating a logistics situational prediction trajectory and providing tiered early warnings.

Description

Technical Field

[0001] This invention belongs to the field of Internet of Things (IoT) logistics management technology, specifically, it relates to an IoT-based logistics data monitoring and management system and method. Background Technology

[0002] With the deep integration of IoT technology into the logistics field, efficient data monitoring and management of goods, vehicles, and facilities scattered globally has become a core challenge for modern supply chains. Existing technical solutions typically rely on two mainstream architectures: one is a centralized cloud processing model, which uploads all sensor data to the cloud platform indiscriminately; the other is an edge computing model with simple filtering capabilities. However, in practical applications, we have found that these existing technologies have a series of technical problems that urgently need to be solved.

[0003] First, existing technologies face a contradiction between data deluge and resource bottlenecks. Traditional centralized processing requires the continuous uploading of massive amounts of raw data, which not only consumes valuable network bandwidth but also places enormous storage and computing pressure on the cloud, resulting in high system response latency and high operating costs. Existing edge computing solutions mostly adopt fixed data upload strategies. When edge intelligent nodes are under resource constraints, rigid upload mechanisms may lead to node overload, data loss, and even affect the stability of core monitoring functions. Second, at the event recognition and processing level, existing technologies lack the ability to identify complex event patterns and lack correlation analysis between discrete event alarms. It is difficult to detect global operational failures or security risks, and there is also a lack of quantitative assessment of the overall operational status of the logistics system and the ability to predict future trends. It is impossible to know potential efficiency bottlenecks or security risks in advance, and can only respond passively, losing the initiative in decision-making.

[0004] To address the above problems, this invention proposes a logistics data monitoring and management system and method based on the Internet of Things. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0006] The technical solution adopted by this invention to solve the technical problem is: a logistics data monitoring and management system based on the Internet of Things, comprising:

[0007] Event Triggering Module: Collects raw status data within the logistics network, sets up edge smart nodes with integrated lightweight rule engine to determine whether to trigger event judgment rules, and if triggered, encapsulates event data packets;

[0008] Status summary module: If not triggered, a status summary is generated for the original status data within the set aggregation time window, and the window length of the aggregation time window is dynamically adjusted based on the resource quantification assessment;

[0009] Event aggregation module: Receives event data packets and status summaries on the cloud service platform and stores them separately. It sets up and uses an event aggregation analysis engine to analyze the event data packets and aggregate them to generate composite events.

[0010] Situational warning module: Constructs and adopts a situational awareness engine, performs real-time analysis and prediction based on composite event streams, generates logistics situational prediction trajectories and provides hierarchical warnings;

[0011] Furthermore, the event data packet is obtained as follows:

[0012] Set up an edge intelligent node with an integrated lightweight rule engine. The edge intelligent node has a built-in local rule library with pre-defined event judgment rules based on business logic, including threshold judgment rules, pattern matching rules and sequence recognition rules. The lightweight rule engine performs logical operations on the raw state data that comes in in real time based on the pre-defined event judgment rules, identifies whether the event judgment rules are triggered, and generates a structured event data packet if triggered.

[0013] Furthermore, the state summary is generated as follows:

[0014] Set an aggregation time window and initialize the window length to the preset standard window length. If the event judgment rule is not triggered, at the end of the current aggregation time window, obtain the original state data within the current aggregation time window and integrate it into a state time sequence according to the time sequence. Extract features from the state time sequence to generate a state summary. The state summary encapsulates the peak value, valley value, arithmetic mean and standard deviation of various original state data within the state time sequence.

[0015] Furthermore, the method for dynamically adjusting the window length of the aggregation time window is as follows:

[0016] An embedded resource status quantification evaluator and a time window dynamic adjuster are deployed inside the edge intelligent node. After the status summary is encapsulated, the resource status quantification evaluator calculates the resource stress index. The time window dynamic adjuster takes the generated resource stress index as input and matches it with the low, middle and high intervals of the preset resource stress index. When the resource stress index is in the low or high interval, the window length of the aggregation time window is set to the corresponding preset lower or upper limit of the window length. When it enters the middle interval, the adaptive adjustment of the window length is triggered. A negative correlation function based on the resource stress index is set to adjust the window length of the aggregation time window piecewise linearly between the preset lower and upper limit of the window length.

[0017] Furthermore, the specific calculation method for the resource scarcity index is as follows:

[0018] The resource status quantification evaluator continuously monitors the performance indicators of the core hardware resources of the edge intelligent node. The resource status quantification evaluator has a built-in weighted fusion algorithm based on the entropy weight method and the analytic hierarchy process. It performs dimensionless processing on each performance indicator, performs weighted fusion and normalizes it into a scalar value between 0 and 1, and marks it as the resource tension index.

[0019] Furthermore, the generation method for composite events is as follows:

[0020] Event data packets are received and verified on the cloud service platform side. After verification, they are stored in the event database. An event aggregation analysis engine is set up to analyze the event data packets that continuously flow into the event database. The event aggregation analysis engine is based on a spatiotemporal correlation calculation framework, which sets a sliding time window and a geographic grid. The geographic grid is divided into grid units. For event data packets within the current sliding time window, the co-occurrence probability assessment model and the causal strength assessment model are used to calculate the co-occurrence probability and causal effect value between event data packets, respectively. If event data packets are in the same grid unit and the co-occurrence probability exceeds the preset co-occurrence probability standard, or the causal effect value between event data packets is significantly non-zero, the event data packets are aggregated into a composite event.

[0021] Furthermore, the specific calculation methods for co-occurrence probability and causal effect value are as follows:

[0022] The co-occurrence probability assessment model uses a co-occurrence probability algorithm to calculate the frequency of different event types co-occurring in the same sliding time window and the same grid cell in the event library. It evaluates the co-occurrence probability between event types using a conditional probability formula. The causal strength assessment model is based on a causal discovery algorithm, which analyzes the temporal dependencies of event sequences in the event library, constructs a causal graph by combining domain knowledge, and calculates the causal effect value between event data packets.

[0023] Furthermore, the method for generating the logistics situation prediction trajectory is as follows:

[0024] A situational awareness engine is constructed, which has a built-in real-time analysis model and a prediction model. An evaluation period is set. At the end of the current evaluation period, the real-time analysis model analyzes the composite events and state summaries within the evaluation period and quantifies the multi-dimensional situational indicators of the current evaluation period. A historical time period is set, and the multi-dimensional situational indicators within the historical time period are integrated into an indicator time series according to the time sequence. The prediction model takes the indicator time series as input and uses a spatiotemporal graph convolutional network based on the attention mechanism to infer and output the predicted values ​​of the multi-dimensional situational indicators for a preset number of future evaluation periods. Combined with the multi-dimensional situational indicators within the current evaluation period, a logistics situation prediction trajectory is formed.

[0025] Furthermore, the specific calculation method for the multi-dimensional situation indicators is as follows:

[0026] The real-time analysis model includes a composite event classification sub-model and a multi-dimensional situation indicator sub-model. The composite event classification sub-model classifies composite events into efficiency-impacting events and security-impacting events. At the end of the current assessment period, the multi-dimensional situation indicator sub-model extracts composite events and status summaries within the assessment period, and integrates them according to time sequence to obtain composite event assessment sequences and status summary assessment sequences. The multi-dimensional situation indicator sub-model defines multi-dimensional situation indicators including efficiency indicators, security indicators, and stability indicators. The efficiency indicators and security indicators are calculated based on the proportion of the duration of efficiency-impacting events and security-impacting events in the composite event assessment sequence within the assessment period, respectively. The stability indicator is calculated by analyzing the fluctuation characteristics and trend changes in the status summary assessment sequence.

[0027] The IoT-based logistics data monitoring and management method includes the following steps:

[0028] Collect raw status data within the logistics network, set up edge smart nodes with integrated lightweight rule engines to determine whether event judgment rules are triggered, and if triggered, encapsulate event data packets;

[0029] If not triggered, a status summary is generated for the original status data within the set aggregation time window, and the window length of the aggregation time window is dynamically adjusted based on the resource quantification assessment.

[0030] The cloud service platform receives and stores event data packets and status summaries respectively, sets up and uses an event aggregation analysis engine to analyze the event data packets, and aggregates them to generate composite events.

[0031] A situational awareness engine is built and adopted to perform real-time analysis and prediction based on composite event streams, generate logistics situation prediction trajectories, and provide hierarchical early warnings.

[0032] The beneficial effects of this invention are as follows:

[0033] 1. This invention achieves efficient preprocessing and intelligent decision-making of data at the source by deploying edge intelligent nodes with integrated lightweight rule engines. By directly executing rules such as thresholding, pattern matching, and sequence recognition at the edge, it can instantly identify key events and immediately encapsulate and report them, greatly shortening the delay from data generation to alarm response and meeting the stringent real-time requirements of logistics monitoring. Secondly, for normal data that has not triggered events, the system dynamically adjusts the aggregation time window through resource quantification assessment. When resources are abundant, it generates high-precision summaries to retain details, and when resources are scarce, it extends the window length for macroscopic statistics, significantly reducing network transmission load and cloud storage pressure. This achieves intelligent and adaptive saving of edge computing resources and communication bandwidth, ensuring the long-term stable operation of the system under complex working conditions.

[0034] 2. This invention utilizes a cloud service platform with an event aggregation analysis engine and a situational awareness engine to achieve a leap in cognition from discrete events to system-level situational awareness. It aggregates isolated alarm events into composite events with business significance, thereby revealing deep-seated interconnected risks. This solves the problems of isolated alarms and superficial insights in traditional systems. Based on composite event streams and historical state data, it constructs multi-dimensional situational indicators and makes accurate predictions, generating logistics situational prediction trajectories. This achieves a fundamental shift from passive alarms to proactive early warnings, providing managers with forward-looking decision support and comprehensively improving the operational resilience and intelligent management level of the logistics system. Attached Figure Description

[0035] The invention will now be further described with reference to the accompanying drawings.

[0036] Figure 1 This is a module architecture diagram of the IoT-based logistics data monitoring and management system described in an embodiment of the present invention;

[0037] Figure 2 This is a flowchart illustrating the steps of the IoT-based logistics data monitoring and management method described in this embodiment of the invention. Detailed Implementation

[0038] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0039] Example 1

[0040] Please see Figure 1 As shown in the embodiments of the present invention, the IoT-based logistics data monitoring and management system aims to solve the problems of edge resource pressure and transmission bottlenecks caused by data deluge in existing logistics monitoring, as well as insufficient management insight caused by isolated events and lagging situational awareness. By deploying edge intelligent nodes integrating a lightweight rule engine, the system performs real-time stream processing on raw state data. When an event judgment rule is triggered, the system immediately encapsulates the event data packet; otherwise, it generates a state summary within an aggregation time window and adaptively adjusts the window length based on resource quantification assessment to achieve resource-sensitive data simplification. On the cloud platform side, the event aggregation analysis engine performs correlation analysis on the event data packets, aggregating and generating composite events. Then, the situational awareness engine calculates multi-dimensional situational indicators based on the composite event stream and state summary, and uses an attention-based spatiotemporal graph convolutional network to predict and generate logistics situational prediction trajectories, ultimately achieving hierarchical early warning and completing closed-loop intelligent management from data to decision. Specifically, it includes the following modules:

[0041] Event Triggering Module: Collects raw status data within the logistics network, sets up edge smart nodes with integrated lightweight rule engine to determine whether to trigger event judgment rules, and if triggered, encapsulates event data packets;

[0042] Specifically, lightweight, low-power IoT sensor nodes are planned and deployed in key physical objects and operational links of the logistics network to form an IoT sensor node network and collect raw status data in real time.

[0043] Key physical objects and operational links include cargo carriers, storage locations, handling equipment, and access gates. The IoT sensor nodes integrate multiple types of micro-sensing units, including temperature and humidity sensors, triaxial accelerometers, optical recognition modules, and RFID readers. These are directly installed and embedded in the corresponding planned monitoring locations to continuously collect raw state data of the physical world according to a predefined unified data model and acquisition frequency. The raw state data includes ambient temperature data, ambient humidity data, vibration data, attitude data, visual image data, and electronic tag information.

[0044] It should be noted that the unified data model specifies the data type, unit, encoding format and timestamp, providing a consistent interface standard for subsequent processing;

[0045] Set up edge intelligent nodes with integrated lightweight rule engines. Edge intelligent nodes are embedded intelligent units deployed at logistics operation sites, which have the ability to collect data, perform real-time calculations and make local decisions. They are networked with IoT sensor nodes through near-field communication links to perform real-time streaming processing on the raw status data uploaded by IoT sensor nodes.

[0046] Specifically, the edge intelligent node has a built-in local rule library, which contains pre-defined event judgment rules based on business logic, including threshold judgment rules, pattern matching rules, and sequence recognition rules. The lightweight rule engine performs logical operations on the real-time incoming raw state data based on the pre-defined event judgment rules to identify whether the event judgment rules are triggered.

[0047] For example, the threshold judgment rule includes a temperature over-limit rule. When the ambient temperature data in the original state data exceeds the temperature safety threshold set by the temperature over-limit rule, the temperature over-limit rule is triggered.

[0048] For example, the pattern matching rules include bearing fault feature identification rules, which calculate the matching degree between vibration data and pre-stored fault feature templates. When it is found that the matching degree with the bearing wear template exceeds the limit, the bearing fault feature identification rules are triggered.

[0049] For example, the sequence recognition rules include outbound process compliance verification rules, which continuously monitor the event sequence of electronic tag information reading, visual recognition, and access control scanning. When the event sequence is reversed or a key link fails to respond within a timeout period, the outbound process compliance verification rules are triggered.

[0050] When real-time stream processing identifies the event judgment rules triggered by the raw state data, it immediately generates a structured event data packet. The event data packet encapsulates the event type, trigger source identifier, occurrence location, timestamp, and associated data snapshot, and reports it in real time through a priority communication queue.

[0051] Among them, the event type is obtained based on the event judgment rule triggered, the trigger source identifier is the unique code of the corresponding IoT sensor node, the occurrence location is obtained based on the location of the IoT sensor node corresponding to the trigger source identifier, the timestamp records the time of triggering the event judgment rule, and the associated data snapshot stores the time sequence of the original state data within a specific time window before and after triggering the event judgment rule.

[0052] It should be noted that this step involves deploying edge intelligent nodes with integrated lightweight rule engines at the edge to perform real-time streaming processing of raw state data, enabling instantaneous identification and response to events. This significantly reduces the latency from data collection to decision-making. At the same time, the structured event data packets significantly reduce the transmission of invalid data, saving network bandwidth and cloud resources. Embedding event judgment rules into edge intelligent nodes at the logistics site enables intelligent filtering and decision-making of data at the source, changing the traditional model that relies on centralized cloud processing and improving the system's real-time performance and autonomy.

[0053] Status summary module: If the event judgment rule is not triggered, a status summary is generated for the original status data within the set aggregation time window, and the window length of the aggregation time window is dynamically adjusted based on the resource quantitative assessment.

[0054] Specifically, set the aggregation time window and initialize the window length to the preset standard window length. The end of the current aggregation time window is also the start of the next aggregation time window.

[0055] At the end of the current aggregation time window, the original state data within the current aggregation time window is obtained and integrated into a state time series according to the time sequence. Feature extraction is performed on the state time series to generate a state summary. The state summary encapsulates the peak value, valley value, arithmetic mean and standard deviation of various original state data in the state time series.

[0056] After the state summary is encapsulated, the window length of the aggregation time window is adaptively adjusted based on the resource quantification assessment. Specifically, a resource state quantification assessor and a time window dynamic adjuster are embedded and deployed inside the edge intelligent node.

[0057] Among them, the resource status quantification evaluator continuously monitors the performance indicators of the core hardware resources of the edge intelligent node. The performance indicators include CPU utilization, memory usage, instantaneous availability of network bandwidth and the depth of the data queue to be processed. The resource status quantification evaluator has a built-in weighted fusion algorithm based on entropy weighting method and analytic hierarchy process to perform dimensionless weighted fusion of various performance indicators and normalize them into scalar values ​​between 0 and 1, which are marked as resource tension index.

[0058] It should be noted that the higher the resource scarcity index, the more strained the computing, storage, and communication resources of the edge intelligent nodes are, and the heavier the burden of data transmission and processing.

[0059] The input of the time window dynamic adjuster is connected to the output of the resource status quantification evaluator to obtain the generated resource tension index and match it with the low, middle and high ranges of the preset resource tension index.

[0060] When the resource scarcity index is in a low range, resources are considered sufficient. The window length of the aggregation time window is set to the preset lower limit value to generate a high-frequency, high-precision state summary, which preserves the details and timeliness of the original state data to the greatest extent.

[0061] When the resource stress index enters the median range, it is determined that resources are becoming tight, triggering an adaptive adjustment of the window length. A negative correlation function based on the resource stress index is set to adjust the window length of the aggregation time window piecewise linearly. The window length dynamically takes a value between the preset lower limit and the upper limit of the window length.

[0062] When the resource stress index reaches a high level, it is determined that resources are in short supply. The window length of the aggregation time window is set to the upper limit of the preset window length, and a low-frequency state summary with more macroscopic statistical characteristics is generated. This significantly reduces the amount of data generated and the reporting frequency, so as to ensure the stability of the core functions of the node.

[0063] It should be noted that the purpose of this step is to generate a state summary containing statistical characteristics within the aggregation time window for normal data that has not triggered events, thereby achieving effective data compression and archiving. Through the resource status quantification evaluator and the time window dynamic adjuster, the window length of the aggregation time window is dynamically adjusted according to the resource tension index. When resources are tight, the data reporting frequency is reduced to ensure node stability, and when resources are sufficient, data details are preserved. This achieves intelligent balance between edge computing resources and communication load, effectively solving the resource bottleneck problem of edge intelligent nodes under variable loads.

[0064] Event aggregation module: Receives event data packets and status summaries on the cloud service platform and stores them separately. It sets up and uses an event aggregation analysis engine to analyze the event data packets and aggregate them to generate composite events.

[0065] Specifically, on the cloud service platform side, a high-throughput data access gateway is set up as a unified data aggregation entry point. The data access gateway adopts a distributed cluster architecture, receives and verifies event data packets and status summaries reported from edge intelligent nodes across the entire domain. The data access gateway performs format compliance checks, source identity authentication, and data integrity verification on the received event data packets and status summaries. After verification, the event data packets and status summaries are stored in different topic databases of the distributed storage system, including an event database and a status history database.

[0066] Configure an event aggregation and analysis engine to analyze the event data packets that continuously flow into the event database;

[0067] Specifically, the event aggregation analysis engine is based on a spatiotemporal correlation calculation framework and adopts a co-occurrence probability assessment model and a causal strength assessment model. It performs joint analysis of concurrent events by setting a sliding time window and a geographic grid. The sliding time window is defined as a configurable time interval that traces back from the current time and is used to capture the time sequence of event data packets. The geographic grid divides the logistics monitoring area into uniform grid units, and each grid unit is assigned a unique geographic code for spatial clustering of the event location.

[0068] It should be noted that the duration of the sliding time window is dynamically set according to business needs, and the granularity of the geographic grid is adjusted based on the monitoring accuracy.

[0069] Among them, the co-occurrence probability assessment model uses the co-occurrence probability algorithm to calculate the frequency of different event types co-occurring in the same sliding time window and the same grid cell in the event library. It evaluates the co-occurrence probability between event types through the conditional probability formula. The causal strength assessment model is based on the causal discovery algorithm, which analyzes the temporal dependency of event sequences in the event library, constructs a causal graph by combining domain knowledge, and calculates the causal effect value between event data packets.

[0070] For event data packets within the current sliding time window, if the event data packets are in the same grid cell and the co-occurrence probability exceeds the preset co-occurrence probability standard, or the causal effect value between event data packets is significantly non-zero, the event aggregation analysis engine will aggregate the discrete event data packets into composite events.

[0071] It should be noted that the purpose of this step is to use the event aggregation analysis engine on the cloud service platform side to aggregate discrete event data packets into more business-meaning composite events through the co-occurrence probability assessment model and the causal strength assessment model. This reveals the inherent relationship and causal chain between isolated events, and enhances the system's deep insight into complex logistics risks. The event aggregation analysis engine integrates spatiotemporal dimensions and causal analysis, going beyond simple threshold alarms. It can discover potential correlation patterns and root causes from concurrent event streams, achieving a cognitive upgrade from point to surface.

[0072] Situational warning module: Constructs and adopts a situational awareness engine, performs real-time analysis and prediction based on composite event streams, generates logistics situational prediction trajectories and provides hierarchical warnings;

[0073] Specifically, a situational awareness engine is built on the cloud service platform side, and the situational awareness engine has built-in real-time analysis model and prediction model;

[0074] The real-time analysis model includes a composite event classification sub-model and a multi-dimensional situation indicator sub-model. The composite event classification sub-model classifies the received composite events into efficiency-impact events and security-impact events, and sets an evaluation period. At the end of each evaluation period, the multi-dimensional situation indicator sub-model extracts the composite events and status summaries within the corresponding evaluation period from the composite event stream and status history database, and integrates them according to the time sequence to obtain the composite event evaluation sequence and status summary evaluation sequence within the evaluation period.

[0075] The multidimensional situation indicator sub-model defines a set of multidimensional situation indicators that reflect the overall operating status of the logistics network. The multidimensional situation indicators include efficiency indicators, safety indicators and stability indicators. The efficiency indicators and safety indicators are calculated based on the proportion of the duration of efficiency-influencing events and safety-influencing events in the composite event assessment sequence within the assessment period, respectively. The stability indicator is calculated by analyzing the fluctuation characteristics and trend changes in the status summary assessment sequence.

[0076] At the end of the current assessment period, the real-time analysis model performs analysis within the current assessment period and quantitatively calculates the multi-dimensional situation indicators for the current assessment period.

[0077] Set a historical time period ending at the current time. The historical time period contains a preset number of evaluation cycles. The multi-dimensional situation indicators within the historical time period are integrated into an indicator time series according to the time sequence.

[0078] The prediction model takes the time series of indicators as input and uses a spatiotemporal graph convolutional network based on the attention mechanism to capture the long-term dependence and periodic patterns of multidimensional situation indicators in the time dimension. Through the prediction model, it outputs the predicted values ​​of multidimensional situation indicators for a preset number of assessment periods in the future.

[0079] The multidimensional situation indicators within the current assessment period are integrated with the output multidimensional situation indicator prediction values ​​to form a logistics situation prediction trajectory.

[0080] A pre-set early warning threshold system for various multi-dimensional situation indicators is provided, which includes multiple warning levels corresponding to various multi-dimensional situation indicator ranges. The warning levels include normal, attention, warning and severe. The multi-dimensional situation indicators within the logistics situation prediction trajectory are matched with the early warning threshold system. The highest warning level among the various multi-dimensional situation indicators is used as the current matching warning level. The early warning information, including the logistics situation prediction trajectory and the warning level, is encapsulated and sent to the logistics management terminal through the notification service of the cloud service platform to realize the monitoring and management of logistics data.

[0081] It should be noted that the purpose of this step is to build a situational awareness engine, which transforms composite events and state summaries into quantifiable multidimensional situational indicators, and uses a spatiotemporal graph convolutional network based on an attention mechanism for prediction to generate logistics situational prediction trajectories. Finally, a hierarchical early warning system is implemented through an early warning threshold system, which transforms management from post-event response to pre-event prediction, provides forward-looking decision support, and forms a closed-loop intelligent management system.

[0082] The technical solution of this invention is as follows: Raw state data is collected within the logistics network; an edge intelligent node integrating a lightweight rule engine is set to determine whether an event judgment rule is triggered; if triggered, an event data packet is encapsulated; if not triggered, a state summary is generated from the raw state data within a set aggregation time window; and the window length of the aggregation time window is dynamically adjusted based on resource quantification assessment. Event data packets and state summaries are received and stored separately on the cloud service platform; an event aggregation analysis engine is set up and used to analyze the event data packets, aggregating and generating composite events; a situational awareness engine is constructed and used to perform real-time analysis and prediction based on the composite event stream, generating a logistics situation prediction trajectory and providing tiered early warnings.

[0083] Example 2

[0084] like Figure 2 As shown in the embodiment of the present invention, the logistics data monitoring and management method based on the Internet of Things includes the following steps:

[0085] Collect raw status data within the logistics network, set up edge smart nodes with integrated lightweight rule engines to determine whether event judgment rules are triggered, and if triggered, encapsulate event data packets;

[0086] If the event judgment rule is not triggered, a status summary is generated for the original status data within the set aggregation time window, and the window length of the aggregation time window is dynamically adjusted based on the resource quantification assessment.

[0087] The cloud service platform receives and stores event data packets and status summaries respectively, sets up and uses an event aggregation analysis engine to analyze the event data packets, and aggregates them to generate composite events.

[0088] A situational awareness engine is built and adopted to perform real-time analysis and prediction based on composite event streams, generate logistics situation prediction trajectories, and provide hierarchical early warnings.

[0089] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. The logistics data monitoring and management system based on the Internet of Things, characterized in that: Comprise: Event triggering module: collect original state data in the logistics network, set up edge intelligent nodes integrated with lightweight rule engine to judge whether to trigger event judgment rules, if triggered, encapsulate event data packets; State summary module: if not triggered, generate state summary for original state data within the set aggregation time window, and dynamically adjust the window length of the aggregation time window based on resource quantification evaluation; The generation method of state summary is: Set the aggregation time window and initialize the window length to the preset standard window length, if the event judgment rule is not triggered, at the end of the current aggregation time window, obtain the original state data within the current aggregation time window and integrate it into a state time sequence according to the time sequence, extract features from the state time sequence to generate a state summary, and the state summary encapsulates the peak value, valley value, arithmetic mean and standard deviation of each type of original state data in the state time sequence; The dynamic adjustment method of the window length of the aggregation time window is: Embed the set resource state quantification evaluator and time window dynamic adjuster in the edge intelligent node, after the state summary is encapsulated, the resource state quantification evaluator calculates the resource shortage index, the time window dynamic adjuster inputs the generated resource shortage index, and matches with the low interval, median interval and high interval of the pre-set resource shortage index, when the resource shortage index is in the low interval or high interval, set the window length of the aggregation time window to the corresponding preset lower limit value or upper limit value of the window length, when entering the median interval, trigger the adaptive adjustment of the window length, set the negative correlation function based on the resource shortage index, and adjust the window length of the aggregation time window in the preset window length lower limit value and window length upper limit value in a segmented linear manner; The specific calculation method of resource shortage index is: The resource state quantification evaluator continuously monitors the performance indicators of the core hardware resources of the edge intelligent node, the resource state quantification evaluator internally has a weighted fusion algorithm based on entropy weight method and analytic hierarchy process, which weights and fuses the performance indicators after de-dimensioning and normalizes them into scalar values between 0 and 1, which are marked as resource shortage indexes; Event aggregation module: receive event data packets and state summaries on the cloud service platform side and store them respectively, set and use event aggregation analysis engine to analyze event data packets, and aggregate to generate composite events; Situation early warning module: build and use situation awareness engine, perform real-time analysis and prediction based on composite event stream, generate logistics situation prediction trajectory and perform hierarchical early warning. 2.The logistics data monitoring and management system based on Internet of Things according to claim 1, characterized in that: The acquisition method of event data packet is: Set the edge intelligent node integrated with lightweight rule engine, the edge intelligent node has a local rule library, the local rule library has pre-set event judgment rules based on business logic definition, including threshold judgment rules, pattern matching rules and sequence recognition rules, the lightweight rule engine performs logical operation on real-time incoming original state data based on the pre-set event judgment rules, identifies whether to trigger the event judgment rule, if triggered, generates a structured event data packet. 3.The logistics data monitoring and management system based on Internet of Things according to claim 1, characterized in that: The generation method of composite event is: The event data packet is received and verified at the cloud service platform side, and is stored in an event library after verification. An event aggregation analysis engine is set up to analyze the continuously flowing event data packet in the event library. The event aggregation analysis engine is based on a time-space correlation calculation framework, and a sliding time window and a geographic grid are set up. The geographic grid divides the grid unit. For the event data packet in the current sliding time window, a co-occurrence probability evaluation model and a causal strength evaluation model are used to calculate the co-occurrence probability and the causal effect value between the event data packets. If the event data packets are in the same grid unit and the co-occurrence probability exceeds the preset co-occurrence probability standard, or the causal effect value between the event data packets is significantly non-zero, the event data packets are aggregated into a composite event. 4.The logistics data monitoring and management system based on Internet of Things according to claim 3, characterized in that: The specific calculation method of the co-occurrence probability and the causal effect value is as follows: The co-occurrence probability evaluation model uses a co-occurrence probability algorithm to calculate the frequency of different event types in the event library co-occurring in the same sliding time window and the same grid unit. The co-occurrence probability between event types is evaluated by a conditional probability formula. The causal strength evaluation model is based on a causal discovery algorithm to analyze the time sequence dependence relationship of the event sequence in the event library, and constructs a causal graph combined with domain knowledge to calculate the causal effect value between the event data packets. 5.The logistics data monitoring and management system based on Internet of Things according to claim 1, characterized in that: The generation method of the logistics situation prediction trajectory is as follows: A situation awareness engine is constructed, which is built-in with a real-time analysis model and a prediction model. An evaluation period is set. The real-time analysis model analyzes the composite event and the state summary in the evaluation period at the end of the current evaluation period, and quantitatively calculates the multi-dimensional situation index of the current evaluation period. A historical period is set. The multi-dimensional situation index in the historical period is integrated into an index time sequence according to the time sequence. The prediction model takes the index time sequence as input, uses a spatio-temporal graph convolution network based on an attention mechanism, and deduces the multi-dimensional situation index prediction value of a preset number of evaluation periods in the future. Combined with the multi-dimensional situation index in the current evaluation period, the logistics situation prediction trajectory is formed. 6.The logistics data monitoring and management system based on Internet of Things according to claim 5, characterized in that: The specific calculation method of the multi-dimensional situation index is as follows: The real-time analysis model includes a composite event classification sub-model and a multi-dimensional situation index sub-model. The composite event classification sub-model classifies the composite event into efficiency impact events and safety impact events. At the end of the current evaluation period, the multi-dimensional situation index sub-model intercepts the composite event and the state summary in the evaluation period, and integrates the composite event and the state summary into a composite event evaluation sequence and a state summary evaluation sequence according to the time sequence, respectively. The multi-dimensional situation index sub-model defines the multi-dimensional situation index, including efficiency index, safety index and stability index. The efficiency index and the safety index are calculated based on the proportion of the time length of the efficiency impact events and the safety impact events in the composite event evaluation sequence in the evaluation period, respectively. The stability index is calculated by analyzing the fluctuation characteristics and trend changes in the state summary evaluation sequence.

7. The logistics data monitoring and management method based on the Internet of Things, characterized in that: The following steps are included: Original state data is collected in the logistics network. An edge intelligent node integrated with a lightweight rule engine is set up to determine whether an event judgment rule is triggered. If it is triggered, an event data packet is encapsulated. If it is not triggered, a state summary is generated for the original state data within the set aggregation time window, and the window length of the aggregation time window is dynamically adjusted based on resource quantification evaluation. The state summary is generated in the following manner: An aggregation time window is set and the window length is initialized as a preset standard window length. If no event judgment rule is triggered, at the end of the current aggregation time window, the original state data in the current aggregation time window is obtained and integrated into a state time sequence according to the time sequence. The state time sequence is subjected to feature extraction to generate a state summary. The state summary encapsulates the peak value, valley value, arithmetic mean value and standard deviation of each type of original state data in the state time sequence. The dynamic adjustment of the window length of the aggregation time window is performed in the following manner: A resource state quantification evaluator and a time window dynamic adjuster are embedded in the edge intelligent node. After the state summary is encapsulated, the resource state quantification evaluator calculates a resource stress index. The time window dynamic adjuster inputs the generated resource stress index and matches it with a preset low range, middle range and high range of the resource stress index. When the resource stress index is in the low range or the high range, the window length of the aggregation time window is set to a corresponding preset lower limit value or upper limit value of the window length. When the resource stress index is in the middle range, adaptive adjustment of the window length is triggered. A negative correlation function based on the resource stress index is set. The window length of the aggregation time window is adjusted in a piecewise linear manner between the preset lower limit value and the upper limit value of the window length. The resource stress index is calculated in the following manner: The resource state quantification evaluator continuously monitors the performance indicators of the core hardware resources of the edge intelligent node. The resource state quantification evaluator internally has a weighted fusion algorithm based on the entropy weight method and the analytic hierarchy process. The weighted fusion of the performance indicators is de-dimensioned and normalized to a scalar value between 0 and 1, which is marked as the resource stress index. The event data packets and the state summaries are received at the cloud service platform side and stored respectively. An event aggregation analysis engine is set and used to analyze the event data packets to generate composite events by aggregation. A situational awareness engine is constructed and used to perform real-time analysis and prediction based on the composite event stream to generate a logistics situation prediction trajectory and perform hierarchical early warning.

Images