Intelligent logistics data supervision freight scheduling system and method based on wireless communication

Abstract

The invention discloses an intelligent logistics data supervision freight scheduling system and method based on wireless communication, and belongs to the technical field of logistics data supervision. Through extracting an instantaneous connection request density abnormal value and an average connection retention time length deviation value in a data uploading process, a server port resource dynamic prediction model is constructed; a port exhaustion risk is judged in real time, when the risk exists, throttling control and an uploading time slice scheduling strategy of terminal equipment are automatically implemented, the connection peak load is effectively dispersed, the system further carries out feedback evaluation on the strategy execution effect and predicts the connection failure probability, and an uploading scheduling scheme is optimized accordingly. And an end-to-end closed-loop regulation and control mechanism is constructed, so that the connection stability and the system reliability in a high-concurrency environment are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of logistics data supervision, and in particular to an intelligent logistics data supervision and freight dispatching system and method based on wireless communication. Background Art

[0002] Intelligent logistics data monitoring and freight scheduling based on wireless communications refers to the real-time collection, transmission, and analysis of vehicle, cargo, route, and time data during logistics transportation processes through wireless communication technologies (such as 4G / 5G and the Internet of Things), thereby achieving intelligent monitoring and scheduling optimization of freight tasks. This approach can improve transportation efficiency, ensure cargo safety, reduce costs, and enhance visibility and controllability of the entire transportation process.

[0003] The existing technology has the following shortcomings:

[0004] In large-scale logistics scenarios, if numerous devices simultaneously initiate frequent new connections to cloud service APIs, particularly in systems deployed using NAT, this can lead to instantaneous exhaustion of server port resources, resulting in port exhaustion. Because the API server lacks connection reuse mechanisms and lacks upload throttling and distributed device scheduling strategies, the system struggles to handle high concurrency, ultimately leading to device connection failures, widespread data upload interruptions, and even system crashes. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent logistics data supervision freight scheduling system and method based on wireless communication to solve the shortcomings of the background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent logistics data supervision and freight scheduling based on wireless communication, comprising:

[0007] Upload the operating data collected by the terminal equipment in the logistics transportation vehicles to the cloud platform in real time;

[0008] During the data upload process, the deviation value ACT between the transient connection request density abnormal value TCR and the average connection duration used to represent the upload behavior is obtained;

[0009] A dynamic prediction model for server port resource usage is built based on TCR and ACT to determine whether there is a risk of instantaneous exhaustion of server port resources in the current scenario.

[0010] If there is a risk of exhaustion, implement throttling control on the terminal device, including limiting the connection rate; or schedule the terminal device upload time in slices to disperse the connection request peak;

[0011] Comprehensively analyze the risk of instantaneous exhaustion of server port resources after policy execution and the effectiveness of policy execution, predict the connection failure probability of terminal devices, and optimize the upload scheduling plan based on the prediction results.

[0012] Preferably, the method for obtaining the transient connection request density abnormal value TCR is:

[0013] In the sliding time window W, collect the timestamp sequence of all terminal devices uploading connections: D = {t1, t2, ..., t i ,...,t n}; where each t i represents the time of a connection request, n is the total number of connection requests recorded, and the connection requests are aggregated by seconds or milliseconds to obtain the number of connection requests per second: X = {x1, x2, ..., x i ,...,x m}; where x i Represents the number of connection requests in the i-th second, m is the total number of time slices divided into equal time periods; calculate the local density ρ of each point i , the expression is: i =∑ j χ(d ij -d c );where: d ij Indicates the density difference distance between point i and point j; d c is the cutoff distance; χ is the cutoff kernel function. For all points with a density greater than the current point, find the nearest density gap and calculate the density peak distance δ i , the expression is: If ρ i is the global maximum density point, and its Map all points to two-dimensional coordinates (ρ i , δ i ); determine the cluster center, and set the time point corresponding to the cluster center as t k , and its corresponding connection density is x k , define the transient connection request density abnormal value TCR as:

[0014] That is, the cluster center with the highest number of connection requests is the instantaneous connection outlier of the time window.

[0015] Preferably, the method for obtaining the average connection holding time deviation value ACT is:

[0016] Collect the connection behavior data d of all terminal devices within a certain time window T z , represents the duration of the connection of device z, and divides the connection duration into h intervals B k, count the number of connections falling into each interval, and get the probability e of each interval z , the expression is:

[0017] Z is the total number of devices; d q Indicates the duration of the connection of device q; calculates the discrete degree H of the connection duration distribution T , the expression is:

[0018] Use the sliding window method to maintain the average discrete degree of the past M time periods and standard deviation σ H , calculate the deviation measure δ of the current window entropy value H , the expression is: The calculation expression of ACT is defined as:

[0019] Where, The average connection duration in the current time window.

[0020] Preferably, a dynamic prediction model for server port resource occupancy is constructed based on TCR and ACT to determine whether there is a risk of instantaneous exhaustion of server port resources in the current scenario, specifically including:

[0021] The instantaneous connection request density anomaly value TCR and the average connection duration deviation value ACT are converted into a comprehensive feature vector, and the comprehensive feature vector is used as the input of the machine learning model. The machine learning model uses each group of comprehensive feature vectors to predict the risk score value label of instantaneous exhaustion of server port resources as the prediction target, and takes minimizing the sum of the prediction errors of the risk score value labels of instantaneous exhaustion of server port resources as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The risk score value of instantaneous exhaustion of server port resources is determined according to the model output results. The machine learning model is a polynomial regression model.

[0022] Preferably, the obtained risk score value of instantaneous exhaustion of server port resources is compared with the preset risk threshold. If the risk score value of instantaneous exhaustion of server port resources is greater than or equal to the preset risk threshold, it is considered that the server has a high risk of instantaneous port exhaustion under the current connection behavior characteristics, and the throttling strategy or distributed scheduling means should be immediately enabled to intervene; if the risk score value of instantaneous exhaustion of server port resources is less than the preset risk threshold, it is considered that the server is currently in a stable state, the port resource consumption is within a controllable range, and the system continues to operate according to the current upload strategy without triggering additional protection mechanisms.

[0023] Preferably, a comprehensive analysis is conducted on the risk of instantaneous exhaustion of server port resources after policy execution and the policy execution effect, the probability of connection failure of the terminal device is predicted, and the upload scheduling plan is optimized based on the prediction results, specifically including:

[0024] In each time window Q, the risk score value before the strategy execution is collected Risk score after strategy execution Current terminal device connection failure rate

[0025] FailRate t ; Number of active disconnections on the server port ConnDrop t ; Average number of terminal reconnections Retry t ; Average connection establishment time Latency t ; Calculate the strategy execution effect index R t , indicating whether the policy is reducing the risk of server port exhaustion, the expression is:

[0026] If R t >0, indicating that the risk is reduced and effective; if R t ≤0, indicating that the strategy is invalid.

[0027] Preferably, a connection failure probability model is constructed to calculate the connection failure probability P of the terminal device. fail :

[0028]

[0029] Of which: ConnTotal t is the total number of connections in the current period; ConnTotal is the total number of active disconnections of the server port, α1, α2, α3, α4 are empirical coefficients; if the predicted P fail >0.1, indicating that the system control strategy still has bottlenecks and needs to be optimized.

[0030] Preferably, a strategy execution anomaly index AEI is generated to comprehensively evaluate the volatility and stability of the strategy execution effect, and the expression is:

[0031] in:

[0032] Indicates the fluctuation value of the risk score value in the most recent g windows;

[0033] γ1, γ2, and γ3 are weighting coefficients.

[0034] The present invention also provides an intelligent logistics data supervision and freight dispatching system based on wireless communication, which includes a data acquisition module, an upload behavior feature extraction module, a risk prediction and assessment module, an upload dispatching control module, and a strategy feedback evaluation and optimization module;

[0035] Data collection module: uploads the operating data collected by terminal devices in logistics transportation vehicles to the cloud platform in real time;

[0036] Upload behavior feature extraction module: During the data upload process, it obtains the instantaneous connection request density anomaly value TCR and the average connection duration deviation value ACT used to represent the upload behavior;

[0037] Risk Prediction and Assessment Module: This module builds a dynamic prediction model for server port resource usage based on TCR and ACT to determine whether there is a risk of instantaneous exhaustion of server port resources in the current scenario.

[0038] Upload scheduling control module: If there is a risk of exhaustion, throttling control of the terminal device is implemented, including limiting the connection rate; or slicing the terminal device upload time to disperse the connection request peak;

[0039] Policy feedback evaluation and optimization module: This module comprehensively analyzes the risk of instantaneous exhaustion of server port resources after policy execution and the effectiveness of policy execution, predicts the probability of connection failure of terminal devices, and optimizes the upload scheduling plan based on the prediction results.

[0040] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0041] 1. This invention uses the outliers of instantaneous connection request density and the deviation from the average connection duration as connection behavior features to construct a dynamic prediction model for server port resources, accurately assessing the risk of instantaneous exhaustion of server port resources in high-concurrency scenarios. Based on this assessment, the system can dynamically implement throttling control and upload fragmentation scheduling strategies to effectively disperse connection peaks, reduce the instantaneous occupancy of port resources, and avoid connection failures and system crashes. This method uses a machine learning model to predict risk scores and intelligently regulate upload behavior accordingly, achieving predictable and controllable management of large-scale terminal connections.

[0042] 2. This invention further performs effect evaluation and risk feedback after policy execution. By calculating the risk reduction ratio, connection failure probability, and policy execution anomaly index, it quantitatively evaluates the effectiveness of the control strategy. It then adaptively optimizes the scheduling strategy based on the evaluation results, forming a closed-loop feedback control mechanism. The overall system enhances stability, scalability, and intelligent resource scheduling in high-concurrency wireless communication environments, significantly improving the responsiveness and operational safety of traditional logistics data upload systems facing port exhaustion. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments described in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0044] Figure 1 This is a mind map of the method of the present invention.

[0045] Figure 2 This is a mind map of the system modules of the present invention. DETAILED DESCRIPTION

[0046] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0047] Example 1, please refer to Figure 1 As shown, the method for intelligent logistics data supervision and freight scheduling based on wireless communication described in this embodiment includes:

[0048] Upload the operating data collected by the terminal equipment in the logistics transportation vehicles to the cloud platform in real time;

[0049] During the data upload process, the deviation value ACT between the transient connection request density abnormal value TCR and the average connection duration used to represent the upload behavior is obtained;

[0050] A dynamic prediction model for server port resource usage is built based on TCR and ACT to determine whether there is a risk of instantaneous exhaustion of server port resources in the current scenario.

[0051] If there is a risk of exhaustion, implement throttling control on the terminal device, including limiting the connection rate; or schedule the terminal device upload time in slices to disperse the connection request peak;

[0052] Comprehensively analyze the risk of instantaneous exhaustion of server port resources after policy execution and the effectiveness of policy execution, predict the connection failure probability of terminal devices, and optimize the upload scheduling plan based on the prediction results.

[0053] In a wireless communication-based intelligent logistics system, uploading operational data collected by terminal devices in logistics transport vehicles to a cloud platform in real time is a key step in achieving transport process visualization, risk warning, and intelligent scheduling. This process typically involves a complete set of processes from data collection, processing, packaging, uploading to the cloud for reception and analysis, which can be broken down into the following:

[0054] First, intelligent terminal devices installed on transport vehicles continuously collect various operational data. These devices typically include a GPS module, environmental sensors (such as temperature, humidity, and vibration sensors), a vehicle operating status interface (such as an OBD-II interface), and a built-in microcontroller (MCU) or edge processing module. The GPS module captures the vehicle's geographic location, speed, and direction in real time; environmental sensors collect parameters such as the temperature and humidity of the cargo's environment; and the vehicle interface provides operational information such as fuel consumption, engine status, and brake status. This data is centrally managed by the MCU and undergoes preliminary processing, such as denoising, standardization, and outlier filtering.

[0055] After completing local processing, the terminal device formats and encapsulates the collected data, often using data structured formats such as JSON, XML, or Protobuf. The package usually contains information such as the device's unique identifier, timestamp, various sensor values, and operating status codes. The encapsulated data is then sent to the cloud platform by the communication module. The communication module accesses the wireless network through the embedded SIM card, including 4G / 5G cellular communication networks or low-power wide-area Internet of Things (such as NB-IoT or Cat-M1). During the network access process, the communication module will complete the APN setting, PDP context activation, and establish a connection with the target server through NAT (if deployed in a private network environment) or a public IP address.

[0056] The data upload process typically utilizes communication protocols such as MQTT, HTTP(S), TCP / IP, or CoAP. MQTT is suitable for scenarios with limited device resources and high upload frequency, supporting persistent sessions and reconnection. HTTP or HTTPS are commonly used for interacting with RESTful APIs and are suitable for structured uploads and command responses. For scenarios with extremely low latency requirements, WebSocket can also be used for two-way persistent communication. During the upload process, the system implements a heartbeat mechanism to maintain a persistent connection and deploys a retransmission strategy to address occasional communication interruptions.

[0057] To ensure data upload continuity and system stability, terminal devices have a built-in local caching mechanism. In the event of a temporary network outage, data is cached in local storage (such as Flash) and automatically re-transmitted in batches upon network restoration. Each piece of data carries a unique identification number and timestamp, enabling the cloud to deduplicate and sequence data, preventing data confusion or duplicate recording.

[0058] After data arrives at the cloud platform, it is initially received by an API gateway or message-based middleware. Subsequently, the data is forwarded to a parsing module for formatting, verification, classification, and cleansing. The processed data is then written to a database or data lake for subsequent use in system modules such as visualization, intelligent analysis, and scheduling optimization.

[0059] The entire upload process emphasizes high real-time performance, high reliability, and high scalability. The system design must ensure that server resources are not exhausted even with high concurrent connections. Connection reuse, throttling control, asynchronous uploads, and distributed scheduling are used to address potential communication bottlenecks or traffic bursts. This ensures that vehicle terminal devices can stably and securely transmit critical operational data to the cloud platform in real time, regardless of geographic location and network conditions.

[0060] During the data upload process, the deviation value ACT between the transient connection request density abnormal value TCR and the average connection duration, which are used to represent the upload behavior, is obtained.

[0061] The method for obtaining the transient connection request density abnormal value TCR is as follows:

[0062] In the sliding time window W (e.g., 5 minutes), collect the timestamp sequence of all terminal devices uploading connections: D = {t1, t2, ..., t i ,...,t n}; where each t i represents the time of a connection request, n is the total number of connection requests recorded, and the connection requests are aggregated by seconds or milliseconds to obtain the number of connection requests per second (time density) sequence: X = {x1, x2, ..., x i ,...,x m}; where x i Represents the number of connection requests in the i-th second, m is the total number of time slices divided into equal time periods; calculate the local density ρ of each point i , the expression is: i =∑ j χ(d ij -d c );where: d ij Indicates the density difference distance between point i and point j; d c is the cutoff distance (density similarity threshold); χ is the cutoff kernel function, χ(s) = 1 if s < 0, otherwise it is 0 (indicating that the density is only counted within the neighborhood).

[0063] For all points with a density greater than the current point, find the nearest density gap, measure the relative position of the current point to the density peak, and calculate the density peak distance δ i , the expression is:

[0064] If ρ i is the global maximum density point, and its Map all points to two-dimensional coordinates (ρ i , δ i ); the cluster center satisfies: high density (ρ i large) and away from other high-density areas (δ i These points are selected as “abnormal peaks”, that is, candidate time periods for TCR abnormal points.

[0065] Let the time point corresponding to the cluster center be t k , and its corresponding connection density is x k , define the transient connection request density abnormal value TCR as: That is, the cluster center with the highest number of connection requests is the instantaneous connection outlier of the time window.

[0066] Outputs the set of TCR values ​​identified within the time window (which can be single or multiple values), which can be used to dynamically adjust the connection throttling strategy, start the connection multiplexing mechanism, or predict the risk of system port exhaustion.

[0067] The method for obtaining the average connection hold time deviation value ACT is as follows:

[0068] Collect the connection behavior data d of all terminal devices within a certain time window T (such as 5 minutes) z , represents the duration of the connection of device z, in seconds or milliseconds. The connection duration is divided into h intervals B k For example, the connection duration is divided into: 0-30 seconds, 30-60 seconds, 60-120 seconds, 120-300 seconds, and >300 seconds; the number of connections falling into each interval is counted to obtain the probability e of each interval. z , the expression is:

[0069] Z is the total number of devices; d q Indicates the duration of the connection of device q; calculates the discrete degree H of the connection duration distribution T , the expression is:

[0070] Use the sliding window method to maintain the average discrete degree of the past M time periods and standard deviation σ H , calculate the deviation measure δ of the current window entropy value H , the expression is: If δ H >θ (if the threshold is set to 2 or 3), it is considered that there is an ACT behavior deviation in the current time period. The calculation expression of ACT is defined as: Where, The average connection duration in the current time window.

[0071] A dynamic prediction model for server port resource usage is built based on TCR and ACT to determine whether there is a risk of instantaneous exhaustion of server port resources in the current scenario. Specifically, the model includes:

[0072] The instantaneous connection request density anomaly value TCR and the average connection duration deviation value ACT are converted into a comprehensive feature vector, and the comprehensive feature vector is used as the input of the machine learning model. The machine learning model uses each group of comprehensive feature vectors to predict the risk score value label of instantaneous exhaustion of server port resources as the prediction target, and takes minimizing the sum of the prediction errors of the risk score value labels of instantaneous exhaustion of server port resources as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The risk score value of instantaneous exhaustion of server port resources is determined according to the model output results. The machine learning model is a polynomial regression model.

[0073] The obtained risk score value of instantaneous exhaustion of server port resources is compared with the preset risk threshold. If the risk score value of instantaneous exhaustion of server port resources is greater than or equal to the preset risk threshold, it is considered that the server has a high risk of instantaneous port exhaustion under the current connection behavior characteristics, and the connection reuse mechanism, throttling strategy or distributed scheduling means should be immediately enabled for intervention; if the risk score value of instantaneous exhaustion of server port resources is less than the preset risk threshold, it is considered that the server is currently in a stable state and the port resource consumption is within a controllable range. The system can continue to operate according to the current upload strategy without triggering additional protection mechanisms.

[0074] The system uses a polynomial regression model based on TCR and ACT to output a server port resource exhaustion risk score for the current time window. When the risk score is greater than or equal to the preset risk threshold, the system determines the current state is high risk and triggers throttling control and upload shard scheduling mechanisms.

[0075] Identify high-frequency connection terminal devices: The system counts the connection frequency of each terminal device in the recent time period and identifies devices with a connection rate higher than the threshold Rmax.

[0076] Dynamic rate limiting instructions: The cloud platform sends rate limiting parameters to frequently connected devices. For example, these parameters include setting a minimum connection interval (e.g., no less than 10 seconds between connections), a maximum number of uploads per minute (e.g., no more than 6), and a reconnection backoff strategy (exponential backoff or random delay). Devices implement local rate limiting mechanisms: Devices that receive control instructions cache uploaded data locally and delay sending according to the rate limit parameters, reducing the number of instantaneous connections.

[0077] A complete upload cycle T (eg, 60 seconds) is divided into a number of equal-time segments Δt, for example, one segment of 5 seconds, for a total of 12 segments.

[0078] Terminal devices are assigned to different time slices based on the device ID hash value, geographic location, or task priority, such as: hash mod 12 → corresponding time slice number; high-priority devices are assigned to the earlier slices, and low-frequency devices are assigned to the later slices.

[0079] The system sends the upload time slice information to the terminal through instructions or configuration parameters. The device only establishes a connection and uploads data within the specified time slice.

[0080] During the scheduling process, the system continuously monitors the actual connection density of each time slice and can dynamically adjust the grouping strategy based on the skew situation to achieve a more balanced connection distribution.

[0081] If the score drops below the threshold for multiple consecutive time windows, the system determines that the risk has been mitigated. The speed limit parameters and shard scheduling control are withdrawn in stages, and normal upload behavior of the device is restored.

[0082] The system records all control actions and device response status, providing feedback for subsequent scheduling strategy optimization and system learning model retraining. Connection rate limiting effectively suppresses short-term upload bursts; upload sharding scheduling balances the time dimension of terminal upload loads; and control strategies dynamically respond to predicted risks, ensuring system resilience and stability.

[0083] Comprehensively analyze the risk of instantaneous exhaustion of server port resources after policy execution and the effectiveness of policy execution, predict the probability of connection failure of terminal devices, and optimize the upload scheduling plan based on the prediction results, including:

[0084] In each time window Q, the system collects the following indicators:

[0085] Risk score before strategy execution Risk score after strategy execution

[0086] Current terminal device connection failure rate (such as connection timeout rate, rejection rate)

[0087] FailRate t ; Number of active disconnections on the server port ConnDrop t ; Average number of terminal reconnections Retry t ; Average connection establishment time (milliseconds) Latency t .

[0088] Calculate the strategy execution effect index R t, indicating whether the policy is reducing the risk of server port exhaustion, the expression is: If R t >0, indicating that the risk is reduced and effective; if R t ≤0, indicating that the strategy is ineffective or even worsens the situation.

[0089] Construct a connection failure probability model (you can use logistic regression, probabilistic regression, or simple linear regression) to calculate the connection failure probability P of the terminal device. fail :

[0090]

[0091] Of which: ConnTotal t The total number of connections in the current period; ConnTotal is the total number of active disconnections on the server port.

[0092] α1, α2, α3, and α4 are empirical coefficients (which can be obtained by fitting historical data). This formula comprehensively considers connection failure rate, retry behavior, disconnection, and latency, and outputs the connection failure risk as a probability value.

[0093] If the predicted P fail >0.1, indicating that the system control strategy still has bottlenecks and needs to be optimized. The system can perform one of the following actions:

[0094] Reduce the shard width, increase the number of shards, and further dilute concurrent uploads;

[0095] Improve the speed limit of the equipment, such as increasing the minimum connection interval;

[0096] Adjust the shard allocation strategy to evenly distribute high-frequency devices to different time slices;

[0097] Increase the local cache limit to reduce the need for instant upload.

[0098] Generate the strategy execution anomaly index AEI, which is used to comprehensively evaluate the volatility and stability of the strategy execution effect. The expression is:

[0099] in:

[0100] Indicates the fluctuation value of the risk score value in the most recent g windows;

[0101] γ1, γ2, and γ3 are weighted coefficients that reflect sensitivity to different factors. If the AEI approaches 0, the control strategy is stable and effective. If it approaches 1 or continues to increase, it indicates that the strategy is unbalanced or has side effects, and the strategy structure should be reassessed.

[0102] Example 2, please refer to Figure 2As shown, the intelligent logistics data supervision and freight scheduling system based on wireless communication described in this embodiment includes a data acquisition module, an upload behavior feature extraction module, a risk prediction and assessment module, an upload scheduling control module, and a strategy feedback evaluation and optimization module;

[0103] Data collection module: uploads the operating data collected by terminal devices in logistics transportation vehicles to the cloud platform in real time;

[0104] Upload behavior feature extraction module: During the data upload process, it obtains the instantaneous connection request density anomaly value TCR and the average connection duration deviation value ACT used to represent the upload behavior;

[0105] Risk Prediction and Assessment Module: This module builds a dynamic prediction model for server port resource usage based on TCR and ACT to determine whether there is a risk of instantaneous exhaustion of server port resources in the current scenario.

[0106] Upload scheduling control module: If there is a risk of exhaustion, throttling control of the terminal device is implemented, including limiting the connection rate; or slicing the terminal device upload time to disperse the connection request peak;

[0107] Policy feedback evaluation and optimization module: This module comprehensively analyzes the risk of instantaneous exhaustion of server port resources after policy execution and the effectiveness of policy execution, predicts the probability of connection failure of terminal devices, and optimizes the upload scheduling plan based on the prediction results.

[0108] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0109] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0110] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0111] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A method for intelligent logistics data monitoring and freight dispatching based on wireless communication, characterized by: include: Upload the operating data collected by the terminal equipment in the logistics transportation vehicles to the cloud platform in real time; During the data upload process, the deviation value ACT between the transient connection request density abnormal value TCR and the average connection duration used to represent the upload behavior is obtained; A dynamic prediction model for server port resource usage is built based on TCR and ACT to determine whether there is a risk of instantaneous exhaustion of server port resources in the current scenario. If there is a risk of exhaustion, implement throttling control on the terminal device, including limiting the connection rate; or schedule the terminal device upload time in slices to disperse the connection request peak; Comprehensively analyze the risk of instantaneous exhaustion of server port resources after policy execution and the effectiveness of policy execution, predict the connection failure probability of terminal devices, and optimize the upload scheduling plan based on the prediction results.

2. The method for intelligent logistics data monitoring and freight dispatching based on wireless communication according to claim 1 is characterized in that: The method for obtaining the transient connection request density abnormal value TCR is as follows: In the sliding time window W, collect the timestamp sequence of all terminal devices uploading connections: D = {t1, t2, ..., t i ,...,t n }; where each t i represents the time of a connection request, n is the total number of connection requests recorded, and the connection requests are aggregated by seconds or milliseconds to obtain the number of connection requests per second: X = {x1, x2, ..., x i ,...,x m }; where x i Represents the number of connection requests in the i-th second, m is the total number of time slices divided into equal time periods; calculate the local density ρ of each point i , the expression is: i =∑ j χ(d ij -d c );where: d ij Indicates the density difference distance between point i and point j; d c is the cutoff distance; χ is the cutoff kernel function. For all points with a density greater than the current point, find the nearest density gap and calculate the density peak distance δ i , the expression is: If ρ i is the global maximum density point, and its Map all points to two-dimensional coordinates (ρ i , δ i ); determine the cluster center, and set the time point corresponding to the cluster center as t k , and its corresponding connection density is x k , define the transient connection request density abnormal value TCR as: That is, the cluster center with the highest number of connection requests is the instantaneous connection outlier of the time window.

3. The method for intelligent logistics data monitoring and freight dispatching based on wireless communication according to claim 2 is characterized in that: The method for obtaining the average connection hold time deviation value ACT is as follows: Collect the connection behavior data d of all terminal devices within a certain time window T z , represents the duration of the connection of device z, and divides the connection duration into h intervals B k , count the number of connections falling into each interval, and get the probability e of each interval z , the expression is: forz=1,2,...,h,Z is the total number of devices; d q Indicates the duration of the connection of device q; Calculate the discrete degree H of the connection duration distribution T , the expression is: Use the sliding window method to maintain the average discrete degree of the past M time periods and standard deviation σ H , calculate the deviation measure δ of the current window entropy value H , the expression is: The calculation expression of ACT is defined as: Where, The average connection duration in the current time window.

4. The method for intelligent logistics data monitoring and freight dispatching based on wireless communication according to claim 3 is characterized by: A dynamic prediction model for server port resource usage is built based on TCR and ACT to determine whether there is a risk of instantaneous exhaustion of server port resources in the current scenario. Specifically, the model includes: The instantaneous connection request density anomaly value TCR and the average connection duration deviation value ACT are converted into a comprehensive feature vector, and the comprehensive feature vector is used as the input of the machine learning model. The machine learning model uses each group of comprehensive feature vectors to predict the risk score value label of instantaneous exhaustion of server port resources as the prediction target, and takes minimizing the sum of the prediction errors of the risk score value labels of instantaneous exhaustion of server port resources as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The risk score value of instantaneous exhaustion of server port resources is determined according to the model output results. The machine learning model is a polynomial regression model.

5. The method for intelligent logistics data monitoring and freight dispatching based on wireless communication according to claim 4 is characterized in that: Compare the obtained risk score of instantaneous port resource exhaustion to the preset risk threshold. If the risk score of instantaneous port resource exhaustion is greater than or equal to the preset risk threshold, the server is considered to have a high risk of instantaneous port exhaustion under the current connection behavior characteristics, and throttling strategies or distributed scheduling methods should be immediately used for intervention. If the risk score value of instantaneous exhaustion of server port resources is less than the preset risk threshold, the server is considered to be in a stable state and the port resource consumption is within a controllable range. The system continues to operate according to the current upload policy without triggering additional protection mechanisms.

6. The method for intelligent logistics data monitoring and freight dispatching based on wireless communication according to claim 5 is characterized in that: Comprehensively analyze the risk of instantaneous exhaustion of server port resources after policy execution and the effectiveness of policy execution, predict the probability of connection failure of terminal devices, and optimize the upload scheduling plan based on the prediction results, including: In each time window Q, the risk score value before the strategy execution is collected Risk score after strategy execution Current terminal device connection failure rate FailRate t ; Number of active disconnections on the server port ConnDrop t ; Average number of terminal reconnections Retry t ; Average connection establishment time Latency t ; Calculate the strategy execution effect index R t , indicating whether the policy is reducing the risk of server port exhaustion, the expression is: If R t >0, indicating that the risk is reduced and effective; if R t ≤0, indicating that the strategy is invalid.

7. The method for intelligent logistics data monitoring and freight dispatching based on wireless communication according to claim 6 is characterized in that: Construct a connection failure probability model to calculate the connection failure probability P of the terminal device fail : Of which: ConnTotal t is the total number of connections in the current period; ConnTotal is the total number of active disconnections of the server port, α1, α2, α3, α4 are empirical coefficients; if the predicted P fail >0.1, indicating that the system control strategy still has bottlenecks and needs to be optimized.

8. The method for intelligent logistics data monitoring and freight dispatching based on wireless communication according to claim 7 is characterized in that: Generate the strategy execution anomaly index AEI, which is used to comprehensively evaluate the volatility and stability of the strategy execution effect. The expression is: in: It represents the fluctuation value of the risk score value in the most recent g windows; γ1, γ2, and γ3 are weighting coefficients.

9. An intelligent logistics data monitoring freight dispatching system based on wireless communication, used to implement the intelligent logistics data monitoring freight dispatching method based on wireless communication according to any one of claims 1 to 8, characterized in that: It includes data collection module, upload behavior feature extraction module, risk prediction and assessment module, upload scheduling control module and strategy feedback evaluation and optimization module; Data collection module: uploads the operating data collected by terminal devices in logistics transportation vehicles to the cloud platform in real time; Upload behavior feature extraction module: During the data upload process, it obtains the instantaneous connection request density anomaly value TCR and the average connection duration deviation value ACT used to represent the upload behavior; Risk Prediction and Assessment Module: This module builds a dynamic prediction model for server port resource usage based on TCR and ACT to determine whether there is a risk of instantaneous exhaustion of server port resources in the current scenario. Upload scheduling control module: If there is a risk of exhaustion, throttling control of the terminal device is implemented, including limiting the connection rate; or slicing the terminal device upload time to disperse the connection request peak; Policy feedback evaluation and optimization module: This module comprehensively analyzes the risk of instantaneous exhaustion of server port resources after policy execution and the effectiveness of policy execution, predicts the probability of connection failure of terminal devices, and optimizes the upload scheduling plan based on the prediction results.

Images