Internet-based logistics transportation intelligent monitoring system

Abstract

The invention discloses a logistics transportation intelligent monitoring system based on the Internet. The system comprises a multi-source data acquisition module, an edge processing module, a communication management module, a cloud intelligent analysis platform and a visual monitoring terminal. The multi-source data acquisition module is used for acquiring original data in a logistics transportation process; the edge processing module is locally connected with the multi-source data acquisition module and is used for locally preprocessing the original data; the communication management module is used for remotely transmitting the locally preprocessed data to the cloud intelligent analysis platform based on the Internet; the cloud intelligent analysis platform is used for receiving the locally preprocessed data based on the Internet, and performing current logistics transportation behavior analysis on the locally preprocessed data based on a pre-trained transportation behavior recognition model; and the visual monitoring terminal is used for displaying the state information of the current logistics transportation based on the behavior analysis result of the cloud intelligent analysis platform.

Description

TECHNICAL FIELD

[0001] The present application relates to an internet-based intelligent monitoring system for logistics transportation. BACKGROUND

[0002] With the rapid development of the logistics transportation industry, especially in the cold chain, fresh food, e-commerce and other time-sensitive or require fine management of transportation scenarios, information collection and remote monitoring in the logistics transportation process have become the key to ensuring transportation efficiency and cargo safety. Traditional logistics monitoring systems are mostly based on GPS positioning terminals, vehicle sensors and platform dispatching systems, which report location, temperature and other data at regular intervals to achieve basic control of the transportation status. However, such systems generally have the problems of single information collection dimension, lagging response mechanism, low intelligence level, poor system compatibility, etc., and are difficult to meet the needs of data integrity, real-time and intelligent decision-making in complex transportation scenarios.

[0003] Chinese patent CN107527078A discloses a logistics information collection and monitoring system based on network technology, which includes barcode scanning, RFID identification, video collection, positioning and wireless communication modules. The microcontroller collects item information and location information and uploads the data to the background server for storage and display. However, this system still cannot meet the needs of data integrity, real-time and intelligent decision-making in complex transportation scenarios.

[0004] Therefore, there is an urgent need for a logistics transportation monitoring system that can improve the intelligence, reliability and response speed of the system to meet the management and control needs of modern logistics in high-complexity transportation environments. SUMMARY

[0005] (I) Technical problems solved

[0006] The present application provides an internet-based intelligent monitoring system for logistics transportation to solve the technical problem that the prior art cannot meet the needs of data integrity, real-time and intelligent decision-making in complex transportation scenarios.

[0007] (II) Technical solutions

[0008] To solve the above technical problems, the present application provides the following technical solutions:

[0009] The present application provides an internet-based intelligent monitoring system for logistics transportation, comprising:

[0010] a multi-source data collection module, an edge processing module, a communication management module, a cloud intelligent analysis platform and a visual monitoring terminal;

[0011] The multi-source data collection module is used to collect raw data in the logistics transportation process.

[0012] The edge processing module is locally connected with the multi-source data acquisition module, and is configured to perform local preprocessing on the raw data.

[0013] The communication management module is configured to remotely transmit the locally preprocessed data to the cloud intelligent analysis platform based on the Internet. The communication management module has access capabilities of at least two network protocols in a cellular network, a local wireless network, a low-power wide-area network, satellite communication, a wired network, and a relay enhanced network, and transmits the locally preprocessed data based on network transmission satisfying the following objective function according to real-time signal strength S, available bandwidth B, communication delay T, and packet loss rate L:

[0014]

[0015] Wherein, C is a communication quality cost function, and a, β, γ, and δ are weight coefficients.

[0016] The cloud intelligent analysis platform is configured to receive the locally preprocessed data based on the Internet, and perform behavior analysis on the locally preprocessed data based on a pre-trained transportation behavior recognition model.

[0017] The visual monitoring terminal is configured to display state information of the current logistics transportation based on the behavior analysis result of the cloud intelligent analysis platform.

[0018] In an optional embodiment, the raw data includes location information, environmental information, transportation equipment state information, and driving behavior information of the current logistics transportation.

[0019] And / or,

[0020] The local preprocessing includes data preprocessing, abnormality identification, feature extraction, and event judgment.

[0021] In an optional embodiment, the edge processing module is configured to perform at least one of the following local preprocessing on the raw data:

[0022] The local mean and the change rate of temperature data and humidity data included in the raw data are calculated by using a preset sliding window algorithm.

[0023] Vehicle acceleration data included in the raw data is input into a pre-trained support vector machine model, and a driving behavior classification result is output. The driving behavior classification result includes vehicle longitudinal dynamic behavior, vehicle lateral dynamic behavior, vehicle stability behavior, and / or non-standard driving behavior.

[0024] Image data included in the raw data is subjected to pixel processing, and Tiny-YOLO is used to perform object detection and driver facial expression analysis on the image data.

[0025] According to a preset combination judgment logic rule, it is judged whether a to-be-processed event is triggered, and in the case that the to-be-processed event is triggered, the corresponding data segment is marked as high-priority data.

[0026] In an optional embodiment, the edge processing module and the multi-source data acquisition module communicate in a local connection mode, and the local connection mode includes at least two of the following physical communication interfaces: a controller area network (CAN) bus interface, a recommended standard (RS)-485 interface, a universal serial bus (USB) interface, an analog voltage input interface, and a wireless short-distance communication interface.

[0027] The edge processing module includes:

[0028] An interface identification unit is configured to identify a communication physical layer standard of the multi-source data acquisition module according to a connected port type.

[0029] A data transmission unit is configured to perform data transmission with the multi-source data acquisition module based on a corresponding communication protocol analysis logic selected according to an identification result.

[0030] In an optional embodiment, α, β, and γ in the target function satisfy:

[0031] α(W) = α0 + k1·W, β(W) = β0 + k2·W, γ(W) = γ0 - k3·W

[0032] wherein α0, β0, γ0, k1, k2, and k3 are preset constants, and W is an importance coefficient of current to-be-uploaded data.

[0033] The communication management module includes:

[0034] A network state detection unit is configured to collect real-time signal strength S, available bandwidth B, communication delay T, and packet loss rate L of a currently supported communication link.

[0035] A link switching control unit is configured to switch the current communication link corresponding to C0 to the communication link corresponding to C i in the case that minimize(C0) > minimize(C i ) and the current to-be-uploaded data includes data marked as high priority by the edge processing module.

[0036] In an optional embodiment, the locally preprocessed data includes longitudinal acceleration A x , lateral acceleration A y , vehicle speed V, steering wheel angle change rate Δθ / Δt, cargo compartment image feature I, current environment temperature T, and current environment humidity H.

[0037] The cloud intelligent analysis platform is configured to construct the local preprocessed data as a feature vector F = [A x ,A y ,V, Δθ / Δt, I, T, H], and input the feature vector into the transportation behavior recognition model to recognize vehicle longitudinal dynamic behavior, vehicle lateral dynamic behavior, vehicle stability behavior, non-standard driving behavior, and / or cargo state.

[0038] In an optional embodiment, the visualization monitoring terminal comprises a map display interface, an information display interface, and a behavior display interface.

[0039] The map display interface is configured to display the transportation path and the current position of the current logistics transportation on a digital map.

[0040] The information display interface is configured to display the vehicle speed V, the current environment temperature T, and the current environment humidity H corresponding to the current logistics transportation.

[0041] The behavior display interface is configured to display the vehicle longitudinal dynamic behavior, the vehicle lateral dynamic behavior, the vehicle stability behavior, the non-standard driving behavior, and / or the cargo state recognized by the cloud intelligent analysis platform.

[0042] In an optional embodiment, when displaying, the behavior display interface displays in a corresponding form according to the confidence of the recognition result of the cloud intelligent analysis platform; wherein the recognition result with high confidence is displayed in a display form with high priority.

[0043] In an optional embodiment, the transportation behavior recognition model comprises a support vector machine (SVM) and a long short-term memory (LSTM).

[0044] The SVM is configured to perform instantaneous behavior classification based on single-frame features in the local preprocessed data.

[0045] The LSTM is configured to perform behavior trend recognition on a sequence of continuous feature vectors in the local preprocessed data.

[0046] The cloud intelligent analysis platform calculates a joint confidence C b based on the output results of the SVM and the LSTM according to the following formula:

[0047] C b = ω s · C SVM + ω l · C LSTM

[0048] wherein C SVM , C LSTMrespectively, are behavior recognition confidence values of the SVM and the LSTM s ,ω l is a weight parameter.

[0049] In an alternative embodiment, the cloud intelligent analysis platform comprises:

[0050] A platform model updating unit is configured to construct an incremental training sample set based on the historical n-day labeled preprocessed data, and to perform incremental training on the transportation behavior recognition model.

[0051] An edge processing updating unit is configured to perform remote updating on the edge processing module based on the incrementally trained transportation behavior recognition model.

[0052] (III) Beneficial Effects

[0053] The technical solution provided by the present application has at least the following beneficial effects:

[0054] 1. The present application integrates a multi-source data acquisition module to realize full-dimensional information acquisition of the logistics transportation process, and performs local preprocessing on the original data based on the edge processing module to improve the response speed and reduce the frequency of uploading invalid data.

[0055] 2. The communication management module of the present application supports multiple network protocols and dynamically selects the link through a communication quality cost function to ensure data reliability and real-time performance.

[0056] 3. The present application combines SVM and LSTM models through a cloud intelligent analysis platform to realize joint recognition of instantaneous behavior and behavior trend, and dynamically updates the model based on historical data to continuously improve the analysis accuracy.

[0057] 4. The visual monitoring terminal of the present application displays through map trajectory, behavior label and confidence level classification, etc., to improve the intuitiveness, interpretability and operation efficiency of the monitoring system.

[0058] 5. The interface structure of the present application supports multi-protocol analysis, and the model and index can be remotely configured and updated, which is suitable for intelligent monitoring requirements of various logistics scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0060] Figure 1A system module architecture schematic diagram of an Internet-based logistics transportation intelligent monitoring system is provided in an embodiment of the present application.

[0061] Figure 2 A structure schematic diagram of an edge processing module is provided in an embodiment of the present application. DETAILED DESCRIPTION

[0062] The present application will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are implemented on the premise of the technical solution of the present application, and detailed implementation manners and specific operation processes are given, but the protection scope of the present application is not limited to the following embodiments.

[0063] In the present application, the words such as "in a possible embodiment", "exemplary" or "for example" are used to represent an example, illustration or description. Any embodiment or design scheme described as "in a possible embodiment", "exemplary" or "for example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. On the contrary, the words such as "in a possible embodiment", "exemplary" or "for example" are intended to present the relevant concept in a specific way.

[0064] Figure 1 A structure schematic diagram of an Internet-based logistics transportation intelligent monitoring system is provided in an embodiment of the present application. As shown in Figure 1 An Internet-based logistics transportation intelligent monitoring system 10 includes:

[0065] a multi-source data acquisition module 101, an edge processing module 102, a communication management module 103, a cloud intelligent analysis platform 104 and a visual monitoring terminal 105;

[0066] The multi-source data acquisition module is used for acquiring original data in a logistics transportation process.

[0067] The edge processing module is locally connected with the multi-source data acquisition module, and is used for locally pre-processing the original data.

[0068] The communication management module is used for remotely transmitting the locally pre-processed data to the cloud intelligent analysis platform based on the Internet. The communication management module has at least two network protocol access capabilities in a cellular network, a local wireless network, a low-power wide-area network, satellite communication, a wired network and a relay enhancement network, and transmits the locally pre-processed data based on network transmission meeting the following objective function based on real-time signal strength S, available bandwidth B, communication delay T and packet loss rate L:

[0069]

[0070] Wherein, C is a communication quality cost function, and a, b, g, d are weight coefficients.

[0071] The cloud intelligent analysis platform is configured to receive the local preprocessed data based on the Internet, and perform behavior analysis on the local preprocessed data based on a pre-trained transportation behavior recognition model.

[0072] The visual monitoring terminal is configured to display the state information of the current logistics transportation based on the behavior analysis result of the cloud intelligent analysis platform.

[0073] In an optional embodiment, the original data may, for example, include location information, environmental information, transportation equipment state information, and driving behavior information of the current logistics transportation.

[0074] The location information may, for example, include latitude and longitude, speed, and a sequence of trajectory points of the current vehicle or goods; the environmental information may, for example, include temperature, humidity, and air pressure values in the vehicle cabin or goods compartment, and is particularly suitable for cold chain or sensitive goods transportation scenarios; the transportation equipment state information may, for example, include engine speed, vehicle speed, oil level, voltage, brake state, gear state, and tire state data; and the driving behavior information may, for example, include longitudinal acceleration, lateral acceleration, steering wheel angle change rate, and driver face image data.

[0075] Correspondingly, the multi-source data acquisition module may integrate multiple types of sensors.

[0076] For example, the sensors for acquiring location information may, for example, include a global positioning system (GPS) unit, a global navigation satellite system (GNSS) unit, and the like, and may be used to acquire real-time positioning, driving trajectory, speed, and the like of the transportation vehicle.

[0077] For another example, the sensors for acquiring environmental information may, for example, include temperature sensors, humidity sensors, air pressure sensors, illumination sensors, gas sensors, and the like, and may be particularly used for environmental monitoring in cold chain and hazardous chemical transportation.

[0078] For another example, the sensors for acquiring dynamic or motion information may, for example, include three-axis acceleration sensors, vibration sensors, tilt sensors, and the like, and may be used for detection of jolting, collision, severe acceleration / braking, and dumping.

[0079] For another example, the sensors for acquiring image or video information may, for example, include visible light or infrared cameras, dash cams, thermal imaging cameras, and the like, and may be used for driver state recognition (e.g., fatigue, distraction) and abnormal goods recognition.

[0080] For example, sensors for collecting vehicle status information, such as fuel consumption sensors, tire pressure monitoring systems, etc., can be used to monitor engine status, fuel consumption, vehicle speed, braking, steering, alarm light status, etc.

[0081] For example, sensors for collecting identity and cargo tracking information, such as radio frequency identification (RFID) readers, near field communication (NFC) units, barcode or two-dimensional code scanners, etc., can be used for cargo identity recognition, loading and unloading state detection, and item matching.

[0082] For example, sensors for collecting personnel behavior information, such as voice interaction modules, etc., can be used to monitor voice commands and rule violations.

[0083] It should be noted that the above is only an exemplary description, and in actual application, it can be set according to different needs, and the present embodiment does not limit this.

[0084] In the present embodiment, the local pre-processed data is remotely transmitted to the cloud intelligent analysis platform based on the Internet by the communication management module, and the edge processing module and the multi-source data acquisition module are connected locally, i.e., part of the data is processed and analyzed near the data acquisition terminal (such as a vehicle terminal or a sensor node), rather than being uploaded to the cloud.

[0085] For example, the edge processing module can use an edge artificial intelligence (AI) chip and can have the function of buffering and delayed uploading under a network outage condition.

[0086] In an optional embodiment, the local preprocessing includes data preprocessing, anomaly identification, feature extraction, and event judgment.

[0087] For example, data preprocessing such as data cleaning, through filtering out obvious error values, packet loss filling, sensor calibration (such as temperature and humidity compensation) to ensure data quality; for example, time synchronization, through time synchronization of various sensor data to fuse the basis; for example, data downsampling and compression, through segmental mean, sliding window, etc. to reduce data upload amount; for example, data standardization and normalization, to unify the data structure to be processed.

[0088] For example, anomaly identification such as rule threshold detection, through setting temperature range, vibration threshold to quickly filter out abnormal events; for example, model prediction deviation detection, through comparison with short-term prediction value, alarm above threshold to improve accuracy; for example, behavior recognition, through driver fatigue detection, cargo dumping recognition to prevent accident risks.

[0089] Feature extraction, such as time series feature extraction, e.g. moving average, volatility, slope, amplitude, etc., is used to support subsequent judgment; image / video feature extraction, such as extracting key objects and actions in the frame based on edge AI model, can be used for image analysis; multi-modal fusion features, combined with environment, behavior, vehicle state, etc., form a joint feature vector, which can be used to support complex judgment.

[0090] Event judgment, such as conditional combination judgment, e.g. if the temperature is out of limits and the GPS is stationary, it is determined that the cold chain fails, which can be used for joint judgment; model reasoning, such as using edge AI model to identify "abnormal driving" and "illegal parking" events, which can be used for intelligent judgment; event caching and reporting, such as temporarily storing data packets and marking emergency upload after local judgment of abnormal events, which can be used for priority upload of key data.

[0091] In an optional embodiment, the edge processing module is configured to perform at least one of the following local preprocessing on the raw data:

[0092] The local mean and rate of change of the temperature data and humidity data included in the raw data are calculated using a preset sliding window algorithm;

[0093] The vehicle acceleration data included in the raw data is input into a pre-trained support vector machine model, and a driving behavior classification result is output, which includes vehicle longitudinal dynamic behavior, vehicle lateral dynamic behavior, vehicle stability behavior, and / or non-standard driving behavior;

[0094] The image data included in the raw data is subjected to pixel processing, and Tiny-YOLO is used for object detection and driver facial expression analysis on the image data;

[0095] According to a preset combination judgment logic rule, it is determined whether a to-be-processed event is triggered, and in the case that the to-be-processed event is triggered, the corresponding data segment is marked as high-priority data.

[0096] For example, in the preprocessing of calculating the local mean and rate of change of the temperature data and humidity data included in the raw data using the preset sliding window algorithm, the local mean calculation, i.e. the sliding window average, such as calculating the average temperature in a certain time period, can be used to smooth the data, reduce random fluctuations, and improve robustness, and can be used for denoising or as a stable value reference; the rate of change evaluation is to measure the change speed of a variable in a unit of time, which can be used to identify mutations, trend judgment, and trigger alarms, and can be used to identify whether the "change" itself is abnormal. The rate of change can be calculated by the formula ΔT=(Ti-T(i-n)) / n, where ΔT is the temperature change rate in a unit of time, Ti represents the current temperature, and n represents the sliding step.

[0097] Exemplarily, in the preprocessing of the driving behavior classification result output by inputting the vehicle acceleration data included in the original data into the pre-trained support vector machine model, the vehicle longitudinal dynamic behavior such as rapid acceleration is embodied in the driving behavior such as rapid accelerator pedal depression, vehicle forward rush, and is embodied in the sensor data such as rapid rise of instantaneous longitudinal acceleration; and for example, rapid uphill / downhill is embodied in the driving behavior such as uphill heavy accelerator, downhill without braking, and is embodied in the sensor data such as longitudinal acceleration change combined with slope data (inclination). The vehicle lateral dynamic behavior such as sharp turning is embodied in the driving behavior such as sharp turning, sudden lane change, and is embodied in the sensor data such as large lateral acceleration and rapid steering wheel angle change; and for example, no speed reduction on a curve is embodied in the driving behavior such as maintaining a high speed in a curve, and is embodied in the sensor data such as GPS route curvature combined with non-decrease of longitudinal acceleration. The vehicle stability behavior such as snake driving is embodied in the driving behavior such as continuous left and right turning, lane deviation, and is embodied in the sensor data such as frequent lateral acceleration fluctuation and lane deviation identified by image recognition; and for example, lane departure is embodied in the driving behavior such as non-lane change with light and emergency lane driving, and is embodied in the sensor data such as camera combined with trajectory model. The non-standard driving behavior such as driver fatigue is embodied in the driving behavior such as frequent lane change and unstable accelerator control, and is embodied in the sensor data such as expression recognition, eye movement tracking and behavior characteristics; and for example, distracted driving is embodied in the driving behavior such as frequent steering wheel operation but unreasonable path, and is embodied in the sensor data such as camera-identified inconsistency between visual line and behavior.

[0098] Exemplarily, in the preprocessing of the image data included in the original data by pixel processing and adopting Tiny-YOLO for object detection and driver facial expression analysis, for example, whether the goods are dumped, slipped, misaligned can be detected by the in-cabin camera image, whether the non-driver personnel is close can be identified by the rear door camera image, whether the vehicle is parked off-road or by the side can be detected by the front camera image combined with GPS to judge whether the vehicle is parked in violation of the rules or in danger, and whether the front obstacle or accident risk can be identified by the front image. In addition, the image input resolution can be set to be not higher than 416x416, and the model inference delay can be controlled to be within 200 milliseconds, so as to improve the processing efficiency. In addition, other ways such as equivalent lightweight convolutional neural network model can also be adopted, and in actual application, different requirements can be set, which is not limited in the embodiment.

[0099] For example, in the case of triggering the pending event according to the preset combination judgment logic rule, in the pre-processing of marking the corresponding data segment as high-priority data, for example, the combination judgment logic rule can include that the environmental temperature exceeds the threshold value and the vehicle is stationary for a long time, and the edge processing module packages and marks the event corresponding data segment as high-priority data for uploading.

[0100] In an optional embodiment, the edge processing module and the multi-source data acquisition module communicate in a local connection mode, and the local connection mode includes at least two physical communication interfaces: a controller area network (CAN) bus interface, a recommended standard (RS)-485 interface, a universal serial bus (USB) interface, an analog voltage input interface, and a wireless short-distance communication interface.

[0101] Figure 2 The structure diagram of the edge processing module provided in the embodiment of the application is shown in FIG. 1. Figure 2 As shown in FIG. 1, the edge processing module 102 includes:

[0102] An interface identification unit 1021 is configured to identify the communication physical layer standard of the multi-source data acquisition module according to the type of the connected port.

[0103] A data transmission unit 1022 is configured to perform data transmission with the multi-source data acquisition module based on a corresponding communication protocol analysis logic selected according to the identification result.

[0104] In addition, the edge processing module and the multi-source data acquisition module can also be connected and communicated in other manners, and in actual applications, different requirements can be set, and the embodiment is not limited in this regard.

[0105] For example, in a cold chain transport vehicle:

[0106] The temperature and humidity sensor module can be connected to the edge processing module through the RS-485 interface and periodically transmit temperature and humidity data using the Modbus RTU protocol.

[0107] The vehicle state acquisition module can be connected to the edge processing module through the CAN bus to obtain vehicle speed, engine speed, fuel consumption, and other information.

[0108] The vehicle-mounted camera module can be connected through the USB interface to transmit image data to the edge AI module for real-time driver state recognition.

[0109] The vibration sensor module can use analog voltage output and perform data acquisition through the ADC channel built in the edge processing module.

[0110] The RFID reader module can be short-range connected with the edge processing module through wireless Bluetooth communication, for identifying the goods label information.

[0111] In addition, the edge processing module can include an interface identification unit configured with multi-channel detection circuit and protocol identification logic, for judging the type of physical communication interface adopted by the access module based on electrical characteristics (such as pin level, voltage, handshake signal, etc.), and identifying the communication physical layer standard used by the module accordingly. After identification is completed, the data transmission unit in the edge processing module calls the corresponding communication protocol analysis logic (such as Modbus parser, CAN frame decoder, USB class driver or custom frame structure analysis module), to realize data transmission and interaction with the multi-source data acquisition module.

[0112] In addition, the edge processing module can support multi-channel parallel access and unified management of heterogeneous devices, and through configuration of interface mapping table and data identification tag, realize unified structured processing and encoding transmission of different types of data sources, to provide data basis for subsequent local preprocessing and event judgment.

[0113] In the embodiment, the communication management module can have access capability of at least two network protocols among cellular network, local wireless network, low-power wide-area network, satellite communication, wired network and relay enhanced network.

[0114] Specifically, the cellular network refers to public network wide-area wireless communication, such as Global System for Mobile Communications GSM, General Packet Radio Service GPRS, Enhanced Data Rates for Global System for Mobile Communications Evolution EDGE, Wideband Code Division Multiple Access WCDMA, High Speed Packet Access HSPA, Fourth Generation Long Term Evolution 4G-LTE, Fifth Generation New Radio Access Technology 5G-NR, Narrow Band Internet of Things NB-IoT, Enhanced Machine Type Communication eMTC (LTE Cat-M1), which is characterized by wide coverage, high speed and low delay (especially 5G).

[0115] The local wireless network refers to local / short-distance wireless communication, such as Wireless Fidelity Wi-Fi (IEEE 802.11 series of wireless local area network protocols 802.11a / b / g / n / ac / ax), ZigBee based on IEEE 802.15.4 standard, Bluetooth technology Bluetooth (BT / BLE), which is characterized by low power consumption, fast transmission, but small coverage and easy to be interfered.

[0116] The low-power wide-area network refers to private / public wide-area low-speed communication, such as Long Range Wireless Communication Technology LoRa, Sigfox Bee Network Communication Technology, Weightless Wireless Communication Technology, which is characterized by extremely low power consumption, strong penetration, low speed and suitable for uploading state data.

[0117] Satellite communication can cross geographical wide-area communication, such as Beidou satellite navigation system short message communication, which is characterized by high reliability, suitable for public network area, and high cost.

[0118] Wired network can communicate at high speed and stability, such as Ethernet technology Ethernet, controller area network CAN, recommended standard 485 bus RS-485, Modbus communication protocol, optical fiber, etc., which is characterized by high bandwidth and low delay, suitable for local use (such as warehouse, park).

[0119] Relay enhanced network, such as multi-network automatic switching, dual-SIM dual-network, mesh networking, can ensure no interruption and no drop, and improve stability.

[0120] In an optional embodiment, a, b and g in the target function satisfy:

[0121] a(W)=a0+k1·W, b(W)=b0+k2·W, g(W)=g0-k3·W

[0122] Wherein, a0, b0, g0, k1, k2, k3 are preset constants, and W is the importance coefficient of the current data to be uploaded.

[0123] The communication management module comprises:

[0124] A network state detection unit is configured to collect real-time signal strength S, available bandwidth B, communication delay T and packet loss rate L of the current supported communication link.

[0125] A link switching control unit is configured to switch the current communication link corresponding to C0 to the communication link corresponding to C in the case that minimize(C0) > minimize(C i ) and the current data to be uploaded contains data marked as high priority by the edge processing module. i

[0126] That is, the backup link can be switched only when the uploaded data is a high-priority abnormal event and the current network cost is too high. In addition, a communication link stability record table can be maintained to record the average connection success rate and average score value of each type of network in the historical transmission process, and the link with better historical performance can be preferentially selected in the link switching process.

[0127] In this embodiment, on the one hand, the edge processing module performs intelligent local preprocessing on the original data, and on the other hand, the data is transmitted to the cloud intelligent analysis platform through the Internet, and the data is further processed and analyzed by the cloud intelligent analysis platform.

[0128] ​In an alternative embodiment, the local pre-processed data comprises longitudinal acceleration A x , lateral acceleration A y , vehicle speed V, steering wheel angle change rate Δθ / Δt, cargo compartment image features I, current ambient temperature T and current ambient humidity H.

[0129] The cloud intelligent analysis platform is configured to construct the local pre-processed data into a feature vector F = [A x , A y , V, Δθ / Δt, I, T, H] and input the feature vector into the transportation behavior recognition model to recognize vehicle longitudinal dynamic behavior, vehicle lateral dynamic behavior, vehicle stability behavior, non-standard driving behavior and / or cargo state.

[0130] In the present embodiment, the visual monitoring terminal can be an interactive monitoring interface system running in a browser front end, a desktop client or a mobile terminal application program, for visually displaying the analysis results and state information of the logistics transportation behavior provided by the cloud intelligent analysis platform.

[0131] In an alternative embodiment, the visual monitoring terminal comprises a map display interface, an information display interface and a behavior display interface.

[0132] The map display interface is configured to display the transportation path and current position of the current logistics transportation on a digital map.

[0133] The information display interface is configured to display the vehicle speed V, current ambient temperature T and current ambient humidity H corresponding to the current logistics transportation.

[0134] The behavior display interface is configured to display the vehicle longitudinal dynamic behavior, vehicle lateral dynamic behavior, vehicle stability behavior, non-standard driving behavior and / or cargo state recognized by the cloud intelligent analysis platform.

[0135] For example, the map display interface can be constructed based on an open source map component, and the current position and historical path trajectory of the logistics transportation vehicle can be rendered on a digital map in real time by calling the current GPS coordinate information of the vehicle. The path can be connected with multiple positioning points to form a trajectory line in the form of a polyline, the color can be dynamically changed according to the transportation state (normal, abnormal), and the current position icon supports clicking to pop up a current state information window.

[0136] The information display interface, for example, can present the basic transportation indicators in real time in the form of a dashboard or a panel, such as the vehicle speed V, current ambient temperature T and current ambient humidity H corresponding to the current logistics transportation, etc.

[0137] The behavior display interface can be arranged in an event stream or timeline manner, supports filtering and viewing according to time, risk level, and automatically labels the behavior event location point on the map interface at the corresponding time point.

[0138] In an optional embodiment, the behavior display interface displays the recognition result according to the confidence level of the cloud intelligent analysis platform when displaying.

[0139] For example, different colors can be used for labeling, such as orange for the confidence interval [0.7, 0.85) and red for [0.85, 1.0].

[0140] In an optional embodiment, the transportation behavior recognition model includes a support vector machine (SVM) and a long short-term memory (LSTM).

[0141] The SVM is used for instantaneous behavior classification based on single-frame features in the locally preprocessed data.

[0142] The LSTM is used for behavior trend recognition of continuous feature vector sequences in the locally preprocessed data.

[0143] The cloud intelligent analysis platform calculates the joint confidence C based on the output results of the SVM and the LSTM according to the following formula: b

[0144] C b = ω s · C SVM + ω l · C LSTM

[0145] where C SVM , C LSTM are the behavior recognition confidence values of the SVM and the LSTM, respectively, and ω s , ω l are weight parameters.

[0146] In an optional embodiment, the cloud intelligent analysis platform includes:

[0147] A platform model updating unit is configured to construct an incremental training sample set based on n days of labeled locally preprocessed data and perform incremental training on the transportation behavior recognition model.

[0148] An edge processing updating unit is configured to remotely update the edge processing module based on the incrementally trained transportation behavior recognition model.

[0149] ​Exemplarily, the cloud intelligent analysis platform can further include a data processing unit, a model training unit, a model inference unit, a model updating unit, and a rule engine unit.

[0150] The data processing unit can receive data uploaded from the edge processing module, and classify, label, and arrange the data in time series.

[0151] The model training unit can establish various transportation behavior recognition models based on historical transportation data, including but not limited to: a cargo transportation path prediction model, a vehicle abnormal driving behavior detection model, a transportation time deviation analysis model, and a transportation environment risk assessment model.

[0152] The model inference unit can make real-time judgments on the current transportation task, and output indicators including risk level, recommended path, and estimated delivery time.

[0153] The model updating unit can periodically perform incremental learning using new transportation data to dynamically adjust the model weights and parameters.

[0154] The rule engine unit can trigger alarm events based on model output and preset rules, and the alarm rules support user-defined configuration, including alarm threshold, trigger condition combination logic, and notification strategy.

[0155] Based on the above structure, the present application realizes full-dimensional information acquisition of the logistics transportation process by integrating multi-source data acquisition modules, and performs local preprocessing on the original data based on the edge processing module to improve response speed and reduce the frequency of uploading invalid data. The communication management module supports multiple network protocols and dynamically selects links through a communication quality cost function to ensure data reliability and real-time performance. The cloud intelligent analysis platform uses a combination of SVM and LSTM models to realize joint recognition of instantaneous behavior and behavior trends, and dynamically updates the model based on historical data to continuously improve analysis accuracy. The visual monitoring terminal displays through map trajectory, behavior label, and confidence level classification to improve the intuitiveness, interpretability, and operation efficiency of the monitoring system. The interface structure supports multi-protocol analysis, and the model and indicators can be remotely configured and updated, suitable for intelligent monitoring needs in various logistics scenarios.

[0156] In addition, it should be noted that the present application can be provided as a method, device, or computer program product. Therefore, the embodiments of the present application can adopt a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present application can adopt the form of a computer program product implemented on one or more computer usable storage media containing computer usable program code.

[0157] The embodiments of the present application are described with reference to the flowchart illustrations and / or block diagrams of the methods, terminal devices (systems) and computer program products according to the embodiments of the present application. It is understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing terminal devices to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal devices, create means for implementing the functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams.

[0158] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal devices to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means that implement the function specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams.

[0159] It should also be noted that, in the present document, the terms such as first and second, and the like are used merely to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between such entities or operations. The terms "comprising", "containing" or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, method, article or terminal device that comprises a list of elements does not only include those elements, but can also include other elements not expressly listed or other elements inherent to such process, method, article or terminal device. Without more limitations, an element defined by the phrase "comprising a" does not exclude the existence of additional identical elements in the process, method, article or terminal device that includes the element.

[0160] Finally, it should be noted that the above description is of preferred embodiments of the application, and that although preferred embodiments of the application have been described, numerous changes and modifications can be made to the preferred embodiments without departing from the principles of the application, and that such changes and modifications are contemplated as falling within the scope of the application. Accordingly, the appended claims are intended to embrace all such changes and modifications.

Claims

1. An intelligent monitoring system for logistics transportation based on the Internet, characterized in that, include: Multi-source data acquisition module, edge processing module, communication management module, cloud-based intelligent analysis platform, and visual monitoring terminal; The multi-source data acquisition module is used to collect raw data during the logistics and transportation process; The edge processing module is locally connected to the multi-source data acquisition module and is used to perform local preprocessing on the raw data; The communication management module is used to remotely transmit the locally preprocessed data to the cloud-based intelligent analysis platform via the Internet. The communication management module has access capabilities for at least two network protocols selected from cellular networks, local area networks, low-power wide area networks, satellite communication, wired networks, and relay-enhanced networks. It transmits the locally preprocessed data based on network transmission that satisfies the following objective function: real-time signal strength S, available bandwidth B, communication delay T, and packet loss rate L. Where C is the communication quality cost function, and α, β, γ, and δ are weighting coefficients; The cloud-based intelligent analysis platform is used to receive the locally preprocessed data via the Internet and to perform current logistics transportation behavior analysis on the locally preprocessed data based on a pre-trained transportation behavior recognition model. The visualization monitoring terminal is used to display the current logistics transportation status information based on the behavior analysis results of the cloud-based intelligent analysis platform.

2. The Internet-based intelligent monitoring system for logistics transportation as described in claim 1, characterized in that, The raw data includes the current logistics transportation location information, environmental information, transportation equipment status information, and driving behavior information; And / or, The local preprocessing includes: data preprocessing, anomaly detection, feature extraction, and event judgment.

3. The Internet-based intelligent monitoring system for logistics transportation as described in claim 1, characterized in that, The edge processing module is used to perform at least one of the following local preprocessing steps on the raw data: The local mean and rate of change of temperature and humidity data included in the original data are calculated using a preset sliding window algorithm. The vehicle acceleration data included in the original data is input into a pre-trained support vector machine model, and the driving behavior classification results are output. The driving behavior classification results include vehicle longitudinal dynamic behavior, vehicle lateral dynamic behavior, vehicle stability behavior and / or non-standard driving behavior. The image data included in the original data is processed pixel by pixel, and Tiny-YOLO is used to perform object detection and driver facial expression analysis on the image data; The system determines whether a pending event is triggered based on preset combination judgment logic rules. If a pending event is triggered, the corresponding data segment is marked as high-priority data.

4. The Internet-based intelligent monitoring system for logistics transportation as described in claim 1, characterized in that, The edge processing module communicates with the multi-source data acquisition module via a local connection method, which includes at least two of the following physical communication interfaces: Controller Area Network (CAN) bus interface, recommended standard RS-485 interface, Universal Serial Bus (USB) interface, analog voltage input interface, and wireless short-range communication interface. The edge processing module includes: An interface identification unit is used to identify the communication physical layer standard of the multi-source data acquisition module according to the type of the connected port. The data transmission unit is used to transmit data with the multi-source data acquisition module based on the parsing logic of the corresponding communication protocol selected according to the identification result.

5. The Internet-based intelligent monitoring system for logistics transportation as described in claim 1, characterized in that, The objective function α, β, and γ satisfy the following: α(W)=α0+k1·W, β(W)=β0+k2·W, γ(W)=γ0-k3·W Where α0, β0, γ0, k1, k2, and k3 are preset constants for y, and W is the importance coefficient of the data to be uploaded. The communication management module includes: The network status detection unit is used to collect the real-time signal strength S, available bandwidth B, communication delay T, and packet loss rate L of the currently supported communication links; The link switching control unit is used to switch the link from minimum(C0) to minimum(C0). i If the current data to be uploaded includes data marked as high priority by the edge processing module, then the current communication link corresponding to C0 will be switched to C. i The corresponding communication link.

6. The Internet-based intelligent monitoring system for logistics transportation as described in claim 1, characterized in that, The locally preprocessed data includes longitudinal acceleration A. x lateral acceleration A y Vehicle speed V, steering wheel angle change rate Δθ / Δt, cargo compartment image features I, current ambient temperature T, and current ambient humidity H; The cloud-based intelligent analysis platform is used to construct a feature vector F = [A] from the locally preprocessed data. x A y The feature vectors [V,Δθ / Δt,I,T,H] are input into the transportation behavior recognition model to identify vehicle longitudinal dynamic behavior, vehicle lateral dynamic behavior, vehicle stability behavior, non-standard driving behavior, and / or cargo status.

7. The Internet-based intelligent monitoring system for logistics transportation as described in claim 1, characterized in that, The visual monitoring terminal includes a map display interface, an information display interface, and a behavior display interface; The map display interface is used to display the current logistics transportation route and current location on a digital map; The information display interface is used to display the vehicle speed V, current ambient temperature T, and current ambient humidity H corresponding to the current logistics transportation. The behavior display interface is used to display the vehicle's longitudinal dynamic behavior, lateral dynamic behavior, vehicle stability behavior, non-standard driving behavior, and / or cargo status as identified by the cloud-based intelligent analysis platform.

8. The Internet-based intelligent monitoring system for logistics transportation as described in claim 1, characterized in that, When the behavior display interface is displayed, it adopts a corresponding form based on the confidence level of the recognition results of the cloud-based intelligent analysis platform; among them, recognition results with higher confidence levels are displayed in a higher priority form.

9. The Internet-based intelligent monitoring system for logistics transportation as described in claim 1, characterized in that, The transportation behavior recognition model includes a support vector machine (SVM) and a long short-term memory network (LSTM). The SVM is used to perform instantaneous behavior classification based on single-frame features in the locally preprocessed data. The LSTM is used to identify behavioral trends in the continuous feature vector sequence in the locally preprocessed data. The cloud-based intelligent analysis platform calculates the joint confidence level C based on the outputs of the SVM and the LSTM using the following formula. b : C b =ω s ·C SVM +oh l ·C LSTM Among them, C SVM C LSTM ω represents the behavior recognition confidence value of the SVM and the LSTM, respectively. s ,ω l These are the weight parameters.

10. The Internet-based intelligent monitoring system for logistics transportation as described in claim 1, characterized in that, The cloud-based intelligent analysis platform includes: The platform model update unit is used to construct an incremental training sample set based on the locally preprocessed data labeled over n days of history, and to incrementally train the transportation behavior recognition model. The edge processing update unit is used to remotely update the edge processing module based on the incrementally trained transportation behavior recognition model.

Images