A refrigerated truck cargo full-automatic loading and unloading monitoring and real-time cargo quantity tracking method

Abstract

The present application relates to the technical field of refrigerated truck cargo monitoring, and discloses a refrigerated truck cargo full-automatic loading and unloading monitoring and real-time cargo quantity tracking method. The method comprises the following steps: deploying multiple types of intelligent sensing devices, collecting cargo weight signals, position signals and loading and unloading action signals of the loading and unloading area of the refrigerated truck; extracting the correlation characteristics of the cargo weight and position signals, and the adjustment characteristics of the loading and unloading events on the two types of signals; generating state characteristics representing the cargo quantity state based on these characteristics, and obtaining a target cargo quantity probability distribution; using the probability distribution to suppress interference on the original sensing signals, and separating out the target weight waveform and target position signal of the target cargo; and finally, combining GPS / Beidou positioning data, realizing real-time cargo quantity tracking of the cargo through a cloud big data analysis platform. Through multi-source signal fusion, feature deep mining and interference suppression processing, the method realizes full-automatic monitoring of the refrigerated truck cargo loading and unloading process and dynamic tracking of the cargo quantity.

Description

Technical Field

[0001] This invention relates to the field of refrigerated truck cargo volume monitoring technology, specifically a method for fully automated loading and unloading monitoring and real-time cargo volume tracking of refrigerated trucks. Background Technology

[0002] In modern logistics systems, refrigerated trucks serve as the core transport vehicle for perishable goods such as fresh produce and pharmaceuticals. Their loading and unloading efficiency and cargo volume tracking accuracy directly impact cargo quality assurance and the smooth operation of the supply chain. With the continuous growth of market demand for cold chain logistics, traditional refrigerated truck cargo loading and unloading and cargo volume management models are gradually revealing numerous limitations. Currently, most refrigerated truck cargo loading and unloading monitoring still relies on manual recording or single-device detection, which is not only time-consuming and labor-intensive but also prone to data distortion due to human error.

[0003] In the cargo loading and unloading process, existing technologies mostly rely on manual counting or simple weighing equipment for cargo volume statistics, making it difficult to capture dynamic changes in real time. For example, manual recording is susceptible to factors such as fatigue and negligence, leading to inaccurate recording of key information such as cargo quantity and weight; while a single weighing device can only obtain the total weight of the cargo, failing to distinguish the distribution of cargo in different areas, and even more difficult to correlate loading and unloading actions with changes in cargo volume. This limitation is particularly pronounced when multiple batches and types of goods are mixed together, easily causing cargo volume statistics errors and creating problems for subsequent warehousing scheduling and distribution planning.

[0004] Refrigerated truck transportation operates in a complex environment, with loading and unloading areas often experiencing vibrations, temperature fluctuations, and electromagnetic interference. This makes sensor signals susceptible to interference, further reducing the reliability of cargo volume detection. Traditional methods lack effective signal processing mechanisms and cannot specifically suppress interference components in the original sensor signals, resulting in significant errors in the extracted cargo weight and location information. Furthermore, the lack of deep integration between cargo tracking and positioning systems makes it difficult to achieve dynamic monitoring of goods throughout the entire transportation process. When cargo is misaligned, lost, or experiences abnormal volume, it cannot be detected and alerted in a timely manner, increasing cargo loss and management risks.

[0005] With the development of intelligent logistics, the requirements for intelligent and automated loading and unloading of refrigerated trucks and cargo tracking are increasing. However, existing technologies are insufficient in areas such as multi-source signal fusion, dynamic feature extraction, interference suppression, and real-time data analysis, resulting in refrigerated truck cargo management remaining at a semi-automated stage, failing to meet the demands of efficient and precise modern cold chain logistics. Therefore, developing a method that enables fully automated loading and unloading monitoring and real-time cargo tracking has become an important direction for upgrading cold chain logistics technology. Summary of the Invention

[0006] The purpose of this invention is to provide a method for fully automated loading and unloading monitoring and real-time cargo tracking of refrigerated trucks, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides a method for fully automated loading and unloading monitoring and real-time cargo volume tracking of refrigerated trucks, the method comprising:

[0008] Deploy multiple types of intelligent sensing devices to collect raw sensing signals from the loading and unloading area of ​​refrigerated trucks, and extract cargo weight signals, cargo position signals, and action signals corresponding to loading and unloading events from the raw sensing signals;

[0009] The correlation characteristics between the cargo weight signal and the cargo location signal, as well as the modulation characteristics of the loading and unloading events on the cargo weight signal and / or the cargo location signal, are obtained.

[0010] Based on the correlation features and the adjustment features, a state feature representing the cargo volume status is generated; and based on the state feature, a target cargo volume probability distribution is obtained.

[0011] The original sensing signal is subjected to interference suppression processing based on the target cargo quantity probability distribution, so as to separate the target weight waveform and target position waveform of the target cargo from the original sensing signal.

[0012] By combining GPS / BeiDou positioning data and using a cloud-based big data analysis platform, real-time cargo volume tracking is achieved.

[0013] Preferably, acquiring the correlation characteristics between the cargo weight signal and the cargo location signal, and the modulation characteristics of the loading / unloading event on the cargo weight signal and / or the cargo location signal, includes the following steps:

[0014] Calculate the weight change rate and location-weight correlation coefficient during the loading and unloading cycle to obtain correlation characteristics;

[0015] Establish loading / unloading-position coordination criteria, analyze the stability index and weight recovery rate of the cargo position signal before and after the loading / unloading event, and obtain the adjustment characteristics;

[0016] The correlation feature is used to adjust the generation process of the target cargo volume probability distribution, and the adjustment feature is used to optimize the interference suppression process.

[0017] Preferably, extracting the action signal corresponding to the loading / unloading event from the original sensing signal includes the following steps:

[0018] Energy gradient analysis is performed on adjacent signal frames of the original sensing signal to locate loading / unloading candidate intervals and generate loading / unloading event markers.

[0019] Non-stationary features are decoupled from the marker signal of the loading and unloading event to separate the frequency band feature components that characterize the loading and unloading action;

[0020] Based on the motion trajectory discrimination criterion, local loading and unloading actions are distinguished from global interference, and the motion intensity waveform is generated by fusing the frequency band feature components.

[0021] The motion intensity waveform is used to assist in generating the adjustment feature, and the loading / unloading event flag is used to trigger the calculation of the correlation feature.

[0022] Preferably, the step of performing energy gradient analysis on adjacent signal frames of the original sensing signal to locate the loading / unloading candidate interval and generate loading / unloading event markers includes the following steps:

[0023] The Teager energy operator difference sequence of adjacent signal frames of the original sensing signal is calculated by using a sliding time window;

[0024] Based on the real-time signal characteristics, a dynamic judgment threshold is obtained;

[0025] Based on the dynamic determination threshold, the start time and duration range of the loading and unloading event are obtained;

[0026] The loading and unloading candidate interval is located based on the start time and duration range of the loading and unloading event, and loading and unloading event markers are generated;

[0027] The output of the loading / unloading event marker is used to control the separation process of the frequency band characteristic components.

[0028] Preferably, the step of distinguishing local loading / unloading actions from global interference based on motion trajectory discrimination criteria and fusing the frequency band feature components to generate an action intensity waveform includes the following steps:

[0029] A three-dimensional motion trajectory map is constructed based on the Doppler phase change, and the Mahalanobis distance between each trajectory point and the historical motion reference trajectory is calculated.

[0030] When the cumulative Mahalanobis distance of continuous trajectory points exceeds the dynamic interference threshold, it is determined as a local loading and unloading action event and the spatial positioning coordinates are output.

[0031] The instantaneous energy of the frequency band characteristic components is spatiotemporally aligned with the spatial positioning coordinates to generate a temporally continuous motion intensity waveform;

[0032] The spatial positioning coordinates are used to optimize the establishment of the loading / unloading-position coordination criterion, and the data stream of the action intensity waveform is used to update the analysis of the weight recovery rate.

[0033] Preferably, after acquiring the correlation characteristics between the cargo weight signal and the cargo location signal, and the adjustment characteristics of the loading / unloading event on the cargo weight signal and / or the cargo location signal, before generating the state characteristics representing the cargo quantity status based on the correlation characteristics and the adjustment characteristics, the method further includes the following steps:

[0034] Determine the duration range of the loading / unloading event and the value of a preset threshold, and determine whether there is overlap of multiple target reflection cross-sections;

[0035] The decision on whether to perform secondary processing is based on the judgment result. The determination of the overlapping of the multi-target reflection cross sections is based on distinguishing between local loading and unloading actions and global interference, and the output of the action intensity waveform generated by fusing frequency band feature components.

[0036] Preferably, after determining the duration range of the loading / unloading event and the value of a preset threshold, and determining whether there is overlap of multiple target reflection cross-sections, the method further includes the following steps:

[0037] If the duration of the loading and unloading event exceeds a preset threshold, or if there is an overlap of multiple target reflection sections, a multi-source data fusion algorithm is used to perform secondary decoupling processing on the cargo weight signal and the cargo position signal.

[0038] Based on the three-dimensional motion trajectory map, the overlapping areas of the reflection sections corresponding to global interference are excluded, and the processed weight signal and processed position signal are re-extracted.

[0039] Based on the processed weight signal, the processed correlation characteristics of the target individual are obtained; based on the processed position signal, the processed adjustment characteristics are obtained; the processed correlation characteristics and the processed adjustment characteristics are used to optimize the generation of the state characteristics;

[0040] The processed weight signal and processed position signal are used as data inputs to update the calculation of the target cargo quantity probability distribution.

[0041] Preferably, after performing secondary decoupling processing on the cargo weight signal and the cargo position signal, the method further includes the following steps:

[0042] Verify the stability of the signal energy distribution after the secondary decoupling process;

[0043] If the verification is successful, the state features are corrected based on the processed correlation features and the processed adjustment features to obtain the processed features, and the processed features are used as input to generate the target cargo volume probability distribution;

[0044] The verification results of the signal energy distribution stability are used to control the parameter adjustment of the interference suppression process.

[0045] Preferably, the real-time cargo volume tracking via a cloud-based big data analysis platform includes the following steps:

[0046] When the sensor data traffic exceeds the threshold and the cloud resource utilization continues to rise, a global load value is generated after calling multi-source sensor data and constructing a time-series dataset.

[0047] When the global load value reaches the system's preset peak threshold, the priority and resource consumption of the evaluation model are assessed based on the scheduling coefficient, and the high-priority model is assigned to the active state.

[0048] Resource reclamation or preloading is performed based on the value accumulation function and model switching overhead, and cloud resource quotas are reserved based on recent value accumulation function prediction results;

[0049] The output of the global load value is used to dynamically adjust the calculation of the scheduling coefficient, and the data of the reserved cloud resource quota is used to support the continuity of the real-time cargo volume tracking.

[0050] Preferably, before generating the global load value after calling multi-source sensor data and constructing a time-series dataset, the method further includes the following steps:

[0051] Based on the acquired raw sensing signals, target modal functions are generated through adaptive decomposition.

[0052] The effective signal is obtained by removing the mode function with the largest total entropy from the target number of mode functions;

[0053] Based on the valid signal, abnormal data is identified by comparing discrete data, and new valid signals are obtained by eliminating and replacing the abnormal data.

[0054] Iterative processing continues until the iteration cutoff condition is met to generate a denoised signal;

[0055] The output of the denoised signal is used to construct the time-series dataset, and the identification results of the outlier data are used to update the evaluation of the value accumulation function.

[0056] Compared with the prior art, the beneficial effects of the present invention are:

[0057] This method, by deploying multiple types of intelligent sensing devices, breaks through the limitations of traditional single-sensor methods, enabling the simultaneous acquisition of cargo weight signals, position signals, and loading / unloading action signals, thus achieving comprehensive capture of multi-dimensional data during cargo loading and unloading. This synchronous acquisition of multi-source signals means that the description of cargo status no longer relies on isolated data, but rather provides a more complete information foundation for subsequent analysis through rich raw signals.

[0058] In the feature extraction stage, by mining the correlation features between cargo weight and location signals, as well as the moderating features of loading and unloading events on these two types of signals, the dynamic changes of cargo during the loading and unloading process can be more accurately depicted. This feature-based analysis method avoids over-reliance on a single signal, making the representation of cargo status more consistent with actual loading and unloading scenarios and reducing judgment bias caused by signal isolation.

[0059] The generation of state features representing the cargo quantity status and the acquisition of the target cargo quantity probability distribution provide a scientific basis for subsequent signal processing. Through the probability distribution model, effective information and interference components in the original sensor signal can be distinguished. Interference suppression processing based on this distinction effectively separates the weight and position waveforms of the target cargo, reducing the impact of environmental interference, equipment noise, and other factors on signal accuracy, and making the extracted cargo parameters closer to the actual situation.

[0060] By combining GPS / BeiDou positioning data with a cloud-based big data analytics platform, real-time tracking and dynamic management of cargo volume information have been achieved. Positioning data provides a spatial dimension reference for cargo volume information, while the cloud platform enables centralized data processing and real-time updates. This allows logistics managers to monitor changes in cargo volume in refrigerated trucks during transportation, promptly detect abnormal fluctuations in cargo status, and provide timely information support for decisions such as cargo scheduling and route optimization. Attached Figure Description

[0061] Figure 1 This is a schematic diagram illustrating the working principle of the fully automated loading and unloading monitoring and real-time cargo tracking method for refrigerated trucks described in this invention.

[0062] Figure 2 A flowchart for extracting action signals corresponding to loading and unloading events;

[0063] Figure 3 A flowchart for locating candidate loading / unloading intervals and marking loading / unloading events;

[0064] Figure 4 A flowchart for secondary processing and judgment of loading and unloading events. Detailed Implementation

[0065] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] Please see Figure 1This invention provides a method for fully automated loading and unloading monitoring and real-time cargo volume tracking of refrigerated trucks, the method comprising:

[0067] The implementation of fully automated loading and unloading monitoring and real-time cargo tracking methods for refrigerated trucks begins with the deployment of various intelligent sensing devices in the loading and unloading area. These devices include, but are not limited to: high-precision weighing sensors, such as strain gauge sensors integrated into the loading and unloading platform or the truck bed; spatial positioning sensors, such as UWB ultra-wideband radar, lidar, or multi-camera vision systems; and sensors for capturing loading and unloading actions, such as inertial measurement units (IMUs), vibration sensors, or acoustic sensors in specific frequency bands. These sensors work together to continuously collect raw sensor signals from the loading and unloading area. From these raw sensor signals, three types of key information need to be extracted: cargo weight signals, reflecting real-time changes in cargo mass; cargo position signals, reflecting the three-dimensional spatial coordinates of the cargo within the truck bed; and action signals corresponding to loading and unloading events, reflecting the operational dynamics of loading and unloading machinery or personnel. Weight signals typically come from weighing sensors, position signals from spatial positioning sensors, while action signals need to be extracted through specific analysis of raw signals, such as vibration, sound waves, or specific image features.

[0068] After acquiring the aforementioned signals, it is necessary to analyze the inherent relationship between the changes in cargo weight and its movement within the cargo compartment during loading and unloading. Simultaneously, it is also necessary to analyze the moderating characteristics of loading and unloading events on the cargo weight or position signals. This refers to how the loading and unloading actions themselves affect the fluctuation patterns of the weight signal, such as how the impact at the moment of loading and unloading causes temporary inaccuracies or how it drives significant changes in the position signal, such as when cargo is moved. Based on the extracted correlation and moderating features, the system can generate state features characterizing the cargo volume status. These state features are a comprehensive quantitative description of the current cargo loading volume, distribution status, and the impact of loading and unloading activities. Using these state features, the system further obtains the target cargo volume probability distribution. This probability distribution model describes the probability that the actual cargo volume within the cargo compartment will be in different states under given sensor observations and characteristic conditions, providing a statistical basis for subsequent processing.

[0069] The system performs interference suppression processing on the original sensor signals based on the probability distribution of the target cargo volume. The core of this step is to use a probabilistic model to identify and filter out noise or interference components in the original signal that are irrelevant to the state of the target cargo, such as slight vehicle shaking, environmental vibration, and sensor background noise. After interference suppression, the target weight waveform and target position waveform of the target cargo can be separated from the original sensor signals. These waveforms more clearly and accurately reflect the weight changes and spatial movement trajectory of the target cargo. Combined with the refrigerated truck's own GPS / BeiDou positioning data, the system provides the vehicle's geographical location and motion status information. All processed signals and feature data are uploaded to a cloud-based big data analysis platform. In the cloud, leveraging powerful computing capabilities and historical data resources, data from multiple refrigerated trucks and multiple loading and unloading operations are integrated, mined, and analyzed to achieve real-time tracking of the refrigerated truck's cargo loading volume, location distribution, and changes, providing real-time decision support for logistics management, inventory optimization, and transportation scheduling.

[0070] Example 1

[0071] See Figure 2 The refrigerated truck is equipped with a high-precision weighing sensor array, distributed in key load-bearing areas of the truck bed, to collect real-time data on pressure changes applied to the cargo, forming the basic source of cargo weight signals. A spatial positioning sensor network deployed on the top and side walls of the truck bed, primarily using ultra-wideband radar technology combined with auxiliary depth vision cameras, continuously scans and acquires the three-dimensional point cloud coordinates of the cargo's outer contour, forming the core data stream of cargo position signals. A broadband vibration sensor array and directional microphones arranged around the loading and unloading area capture physical vibrations and specific frequency sound waves generated by mechanical operations; these raw physical quantities constitute the source information for loading and unloading event analysis. The system uses a unified timestamp to synchronously collect data from all sensor streams, ensuring traceable temporal correlation between different signals. Several minutes before the loading and unloading operation begins, the system automatically performs baseline calibration, recording the zero-point offset of each sensor and the background noise characteristics of the environment while the vehicle is stationary, generating a dynamic background model for subsequent signal extraction.

[0072] During the continuous acquisition of raw sensor data streams, the system's primary task is to separate the mixed physical quantities into three identifiable signal types. For the continuous voltage signal acquired by the pressure sensor array, the system eliminates low-frequency baseline drift caused by vehicle engine idling. Specifically, an adaptive high-pass filter is used to eliminate fluctuation components with frequencies below a specific cutoff value. Peak pulses in the remaining signal that are higher than the average ambient noise usually correspond to instantaneous pressure changes caused by cargo placement or removal. The system identifies these abrupt changes by comparing the signal energy integral differences within adjacent time windows. The key to spatial positioning signal processing is eliminating reflection points from static interference objects. A dynamic clustering algorithm based on multi-frame point cloud data is used to identify valid point groups belonging to the cargo, and position trajectory data is generated by calculating the change in the centroid of the point group in the three-dimensional coordinate system. The raw vibration sensor signal undergoes multi-stage bandpass filtering to retain frequency band energy changes reflecting metal impact and motor operation. The acoustic sensor focuses on analyzing the characteristic frequency bands of cargo collision sounds and conveyor belt friction sounds. When multiple sensor channels detect synchronous changes consistent with loading and unloading characteristics within the same timestamp, the system generates an event marker with a confidence score and records the event trigger time. The correlation analysis between weight and position signals needs to address the issue of differences in physical dimensions. The system introduces a dynamic time warping algorithm to align weight abrupt changes and position jumps on the time axis. The position-weight synchronization correlation coefficient is calculated by statistically analyzing the frequency of successful matching between the two within a preset time window. The moderating effect of loading and unloading events on the signal is reflected in the recovery characteristics of the weight sensor readings from shock fluctuations to stable values. The system sets the time for the signal standard deviation to converge to the baseline level as the weight recovery rate index; position stability is quantified by the ratio of the change in the radius of the cargo point cloud distribution before and after the event.

[0073] The application of correlation features directly affects the construction process of the target cargo volume probability model. When the location-weight synchronization coefficient is consistently below a threshold, the system automatically reduces the weight of the location signal in the joint probability distribution model; conversely, it increases the confidence parameter of the location dimension. The recovery rate parameter in the adjustment features controls the adjustment of the probability distribution function's attenuation coefficient. Periods of slow recovery after loading and unloading correspond to the broadening correction of the distribution function, enhancing tolerance to potential residual weight interference. When generating cargo volume status features, the system employs a multi-dimensional feature fusion mechanism to encode the mean weight, location distribution entropy, and loading / unloading action intensity into a fixed-dimensional vector descriptor. The status feature vector is input to a cargo volume probability estimator based on a Gaussian mixture model, generating the probability density distribution of the current cargo status in the cargo volume space based on the cluster centers of historical training data. This distribution guides the original signal filtering using a variable bandwidth suppression strategy: narrowband filtering is used in high-probability-density regions to preserve details, while wideband filtering is used in low-probability intervals to eliminate suspicious fluctuations. The key to separating the target weight waveform lies in constructing adaptive filter coefficients weighted by the probability distribution, effectively suppressing atypical weight change patterns. The optimization of the position waveform relies on the probability distribution's assessment of the rationality of the movement trajectory. During the position coordinate update stage, a constraint on cargo volume change is introduced to eliminate physically inconsistent drift points. After all signal processing, a time-stamped feature data packet is finally generated at the vehicle gateway, containing target weight waveform sampling points, target position coordinate sequence, and key feature statistics.

[0074] The energy gradient analysis of the raw sensor signal needs to balance the relationship between detection sensitivity and noise resistance. The system adopts a dual dynamic threshold mechanism to handle the energy change of the sliding window: the first threshold is set as a multiple of the root mean square of the environmental noise, used for preliminary screening of suspected event segments; the second threshold is dynamically adjusted based on the smoothness of the signal over several consecutive seconds, automatically raising the event trigger threshold in bumpy road sections. The time window size selection needs to cover the baseline data of the complete loading and unloading cycle, setting the initial value of the window for the common rhythm of forklift operations and automatically expanding and contracting according to working conditions. When the energy change of a specific channel signal exceeds the dual thresholds, the system records the start timestamp and continues to monitor until the signal fluctuation falls back to the stable range, at which point the determination ends. When the event duration data is written to the dynamic database, an abnormal state flag is marked to facilitate subsequent process judgment on whether to trigger the secondary processing mechanism. This flag information also drives the modification of the preprocessing configuration parameters of the next stage of signal decomposition, determining the spatial dimension sampling accuracy of the non-stationary feature decoupling algorithm. The core of non-stationary feature decoupling lies in separating the intrinsic signal components representing the loading and unloading action.

[0075] For multi-source sensor data streams during the marked time period, the system employs variational mode decomposition to decompose them into a finite number of narrowband oscillation components. The clustering results of the center frequencies of each component correspond to typical operational acoustic characteristics. For example, components with frequencies concentrated in the 250-450 Hz range are identified as resonances between the fork teeth and the shelf; high-frequency components are used to identify the stretching characteristics of plastic wrapping film. The motion trajectory discrimination process uses the original phase change matrix of millimeter-wave radar to reconstruct the cargo displacement vector and detects abnormal motion patterns by comparing it with a preset coordinate system reference grid. The dynamic trajectory point set of spatial location information needs to be matched with a preset standard loading and unloading path library to evaluate its matching degree and calculate the orientation angle deviation and velocity fluctuation index of each path node. Spatial density clustering is performed on continuous abnormal trajectory points to identify local loading and unloading event hotspots. The output value of the hotspot position coordinates is used to correct the regional weighting coefficient in the position stability assessment algorithm. The generation of motion intensity waveforms realizes the quantitative mapping from physical signals to operational behavior intensity. The instantaneous energy integral values ​​of each frequency band component are weighted and fused into a unified intensity index through sensor spatial position coefficients. Spatial coordinate calibration ensures the accurate positioning and mapping of the moving target in the three-dimensional scene. The waveform data periodically updates the weight recovery rate model parameters in the state characteristic analysis module. The system establishes a relationship matrix between peak intensity of the action and weight stabilization time through regression analysis for real-time correction and adjustment of the characteristics. The output data also drives the cloud-based operation mode analysis engine to automatically build an optimal operation benchmark library. The long-term accumulated waveform characteristic data provides key decision-making basis for optimizing loading and unloading procedures.

[0076] Example 2

[0077] See Figure 3The millimeter-wave radar array deployed in the refrigerated truck loading and unloading area forms a three-dimensional spatial perception network. Each radar node continuously transmits frequency-modulated continuous waves and receives reflected signals. The raw echo signals contain static environmental reflections and dynamic target information. The system separates the frequency domain feature points corresponding to the moving targets through range-Doppler processing. For each potential moving point detected, the system records its range-direction frequency offset and azimuth-direction phase difference, and calculates the instantaneous three-dimensional coordinates of the target in the radar coordinate system based on the multi-antenna geometry. The coordinate points of adjacent moments are connected in chronological order to form preliminary motion trajectory segments. The system automatically eliminates jump points caused by multipath effects and smooths the trajectory data through Kalman filtering. Vibration signals generated during loading and unloading operations are collected by accelerometers arranged in the truck body structure. The raw vibration waveforms are bandpass filtered to retain the frequency bands reflecting mechanical impact, and the Teager energy operator values ​​of the signals within each window are calculated using a sliding time window. The Teager energy difference sequence between adjacent time windows is calculated in real time. When the difference exceeds a threshold dynamically adjusted based on the environmental vibration level, an event start marker is triggered. The system continues to monitor until the difference falls back below the threshold to determine the event duration, generating a loading / unloading event marker segment containing start and end timestamps. This marker segment controls the resource allocation for subsequent signal processing, activating computationally complex feature extraction algorithms only during the marked period.

[0078] During the marked period, the system reconstructs the three-dimensional motion trajectory from the raw millimeter-wave radar data. By solving the phase difference matrix ΔΦ between each receiving antenna and combining it with the carrier wavelength λ and the antenna spacing d, the target azimuth angle θ is calculated. ;

[0079] Where: θ represents the target azimuth angle, λ is the carrier wavelength of the radar transmitted signal, ΔΦ is the phase difference matrix between the receiving antennas, and d is the antenna spacing. By integrating spatial points from consecutive frames using a multi-target tracking algorithm, a time-varying trajectory point set P(t) = [x(t), y(t), z(t)] is formed. To distinguish between local loading / unloading actions and the overall vehicle motion, the system establishes a historical trajectory benchmark model: during periods without loading / unloading operations, trajectory data under typical driving conditions is collected, and the first three principal component directions are extracted through principal component analysis to construct a benchmark subspace B. For a new trajectory point P(t), its Mahalanobis distance D to the benchmark subspace B is calculated. M (P): ;

[0080] in: The mean vector of the baseline trajectory. Let P be the covariance matrix of the baseline trajectory and P be the new trajectory point. When the cumulative Mahalanobis distance of five consecutive trajectory points exceeds the dynamic threshold, it is determined to be a valid loading / unloading action, and the spatial positioning coordinates of this action are output as C=[x]. c ,y c ,z cThe vibration sensor signals are processed synchronously, and variational mode decomposition is performed on the acceleration data within the marked time period to separate specific frequency band components characterizing the loading and unloading actions. The instantaneous energy of each component is spatiotemporally aligned with the spatial positioning coordinates C to generate a time-domain continuous motion intensity waveform. The dynamic adjustment mechanism of key parameters involved in the generation of the motion intensity waveform is based on real-time operating condition feedback.

[0081] The spatial positioning coordinates C are updated in real time to the regional weight coefficients in the loading / unloading position coordination criterion. When loading / unloading actions are detected to be concentrated in the front area of ​​the truck bed, the position stability assessment algorithm automatically increases the data sampling frequency of the front sensors. The data stream of the action intensity waveform is input into the weight recovery analysis module. The system establishes a regression model between the intensity peak and the weight signal stabilization time. When the monitored intensity peak exceeds twice the standard deviation of the historical mean, the calculation window for the weight recovery rate is automatically extended. The accuracy of the 3D motion trajectory map is improved through multi-sensor fusion. UWB anchor points added at the four corners of the truck bed provide auxiliary positioning information, and the trajectory coordinate accuracy is optimized by minimizing reprojection error. The benchmark model required for Mahalanobis distance calculation is automatically updated every ten minutes to adapt to changes in vehicle vibration patterns caused by changes in road conditions. The spatiotemporal alignment process uses dynamic time warping technology to solve the time delay problem between vibration signals and spatial coordinates, ensuring that the action intensity waveform accurately reflects the spatiotemporal characteristics of physical events. The output data simultaneously drives the cloud analysis platform to build a digital twin model of the loading / unloading operation. The mapping relationship between the action intensity waveform and spatial coordinates provides a quantitative basis for optimizing loading / unloading path planning.

[0082] Example 3

[0083] See Figure 4 After initial signal feature extraction, the refrigerated truck loading and unloading monitoring system enters the complex condition judgment stage. The system presets a standard duration threshold reference value for loading and unloading events, derived from the average time taken for a single routine cargo handling operation recorded in the historical operation statistics database. When the duration of a real-time detected loading and unloading event exceeds 1.5 times the reference value, the system automatically marks it as an abnormally long event. The motion intensity waveform analysis module synchronously outputs a spatial heatmap. When a specific area continuously exhibits high-intensity pulses within the event period and the spatial coordinate fluctuation range is less than 0.3 meters, it is determined that there is a multi-target reflection cross-section overlap phenomenon. The system confirms the judgment result through a dual-channel verification mechanism: the weight signal channel detects a continuous stepped change pattern, and the position signal channel shows multi-point cloud aggregation characteristics. The judgment result triggers a secondary processing flag update, which directly controls the branch selector of the signal processing pipeline.

[0084] The multi-source data fusion algorithm employs a hybrid architecture of blind source separation and beamforming to process abnormal event signals. The raw data stream from the weight sensor array and the millimeter-wave radar point cloud data stream are input into the joint processing unit. A signal mixing model is established:

[0085] X(t) = [W(t); P(t)] = A·S(t) + N(t); where X(t) represents the observation matrix composed of the weight signal W(t) and the position signal P(t), A is the unknown mixing matrix, S(t) is the source signal matrix to be separated, and N(t) is the environmental noise term. The separation matrix is ​​estimated iteratively through independent component analysis to ensure statistical independence of each component of the output signal. A three-dimensional motion trajectory map provides spatial constraints. The system calculates the spatiotemporal correlation between each separated component and the trajectory hotspot, selecting components with a correlation exceeding 0.7 for signal reconstruction, while suppressing components corresponding to global interference regions in the trajectory map. Adaptive beamforming technology is introduced during the reconstruction process to enhance the signal gain in the target direction in the spatial domain. This processing yields the separated target weight and position signals, and their data quality is quantitatively evaluated by the signal-to-noise ratio improvement.

[0086] After secondary decoupling, the signals enter the feature recalculation process: the position-weight correlation coefficient is calculated using the ratio of covariance to standard deviation, where the weight signal uses the decoupled data and the position signal uses the vertical component. When re-extracting loading and unloading adjustment features, a motion intensity waveform guidance mechanism is introduced: using the peak moment of the motion intensity waveform as a reference point, the rate of change of variance of the position signal and the convergence time of the weight signal are analyzed within 0.5-second windows before and after it. The processed correlation and adjustment features are input to the feature optimizer. This module performs feature weighted fusion using weight coefficients learned from historical data to generate an enhanced state feature vector. The target cargo volume probability distribution model receives this feature vector as conditional input and updates the probability density function parameters. The processed weight and position signals are synchronously input into the data pool, triggering the online learning mechanism of the probability distribution model to adjust the distribution morphology parameters.

[0087] The stability verification of signal energy distribution employs a multi-index cross-validation scheme. Time-domain analysis calculates the fluctuation of the processed weight signal during the loading and unloading interval, while frequency-domain analysis extracts the energy proportion of a specific low-frequency band. A comprehensive stability index is set, which is obtained by weighted summation of the reciprocal of the time-domain fluctuation and the low-frequency energy proportion. Verification is considered successful when the index exceeds an empirical threshold. After successful verification, the system marks the enhanced state features as valid inputs, replacing the original state features for calculating the cargo volume probability distribution. The verification results are simultaneously input to the interference suppression parameter controller. This module selects different levels of filtering strategies based on the range of the index value: basic filtering parameters are selected when the index is in the high-value range; enhanced filtering mode is activated when it is in the middle-value range; and three-level filtering is triggered and the sensor diagnostic program is started when it is below the critical value. Parameter adjustment specifically involves gradually reducing the value of the process noise covariance matrix and increasing the moving average window ratio. The entire processing flow forms a closed-loop quality control to ensure the reliability of signal processing under complex operating conditions.

[0088] Example 4

[0089] After undergoing secondary decoupling processing, the refrigerated truck cargo loading and unloading monitoring system enters the quality verification stage. The decoupled weight signal data stream and position signal data stream are input into the verification module, where the system performs time-domain statistical analysis on the weight signal. Within a stable period between two consecutive loading and unloading events, a continuous sampling sequence of at least five seconds is selected to calculate the dispersion index of the weight signal sampling values ​​during this period. This index reflects the natural fluctuation level of the signal under no operational interference. Simultaneously, frequency-domain energy distribution analysis is performed. A fast Fourier transform is applied to the weight signal to obtain a spectrum, calculating the proportion of vibration energy in the frequency range of 1 Hz to 5 Hz to the total energy of the entire frequency band. An excessively high proportion of low-frequency energy usually indicates the presence of unfiltered mechanical vibration residue. Position signal verification focuses on the convergence of spatial coordinates. Within a three-second window after the loading and unloading action ends, the contraction rate of the cargo point cloud distribution radius is calculated. This rate is obtained by fitting the slope of the distribution radius over time using linear regression. The system sets the verification sampling frequency to half of the original sensor sampling rate to balance computational load and feature fidelity requirements.

[0090] The signal energy distribution stability verification employs a multi-dimensional index fusion strategy. The system constructs a verification matrix including time-domain fluctuation index, frequency-domain energy index, and spatial convergence index. The time-domain fluctuation index is normalized by taking the reciprocal of the variance of the weight signal during stable periods. The frequency-domain energy index directly uses the original value of the energy proportion in the 1-5Hz frequency band. The spatial convergence index takes the absolute value of the rate of change of the distribution radius of the location point cloud. The three indices are multiplied by adaptive weighting coefficients and then summed to generate a comprehensive stability index. The weighting coefficients are dynamically adjusted according to the current compartment temperature and vehicle movement status. The verification threshold is set based on a historical successful case database, and the system automatically updates the threshold baseline every 24 hours. When the comprehensive stability index exceeds the dynamic threshold, the secondary decoupling process is considered successful. The verification result triggers a state feature update command, and the enhanced state feature vector is marked as a valid data source, replacing the state features generated by the original processing channel. When the new feature vector is input into the target cargo volume probability distribution model, the system synchronously records the feature switching timestamp and version identifier, providing data anchors for subsequent quality traceability.

[0091] The interference suppression parameter adjustment mechanism establishes a three-level response mode. The system maintains a parameter configuration table storing the filter parameter sets corresponding to different verification results. This table is periodically expanded as the vehicle's cumulative operating time increases. When the comprehensive stability index is in the high confidence interval, the system selects the basic filter parameter set: the Kalman filter process noise covariance matrix uses the standard configuration value, and the moving average window length is set to the default ten sampling periods. When the medium confidence interval is triggered, the system activates the enhanced filtering mode: the values ​​of each element in the process noise covariance matrix are uniformly reduced by 30%, and the moving average window is expanded to fifteen sampling periods. When the index is below the critical threshold, the system activates the three-level response: the process noise covariance matrix value is halved, the moving average window is extended to twenty-five sampling periods, and a sensor verification request is sent to the on-board diagnostic system. The parameter switching process uses a gradual adjustment algorithm to avoid signal jumps, and the new parameters are gradually transitioned to the correct position through linear interpolation over five sampling periods. All parameter adjustment events are logged, including fields such as timestamp, parameter value before adjustment, target parameter value, and adjustment duration (see Table 1).

[0092] Table 1: Verification index parameters for secondary decoupling signals.

[0093]

[0094] The verification process includes an anomaly handling branch. When the comprehensive stability index fails to verify three times consecutively, the system automatically initiates the signal tracing procedure. This procedure retrieves the original sensor signal cache data from the five most recent loading / unloading events, re-executes the decoupling process, and records intermediate variables. The diagnostic engine compares the differences in characteristic parameters during each processing step, focusing on analyzing key node data such as the estimated value of the mixing matrix, the independent component quantity criterion, and the beamforming steering vector. After the diagnostic report is generated, it is automatically uploaded to the cloud analysis platform, while the local system reverts to the basic processing mode without secondary decoupling. In the revert mode, the system attempts to reactivate the decoupling processing module every two hours until the parameter correction scheme issued by the cloud is successfully loaded. The entire verification mechanism and parameter adjustment system form a closed-loop control, dynamically optimizing the signal processing link configuration through continuous monitoring of processing quality. All operation logs and diagnostic reports are stored in the vehicle's black box memory in chronological order, supporting offline analysis of up to thirty days of operating data.

[0095] Example 5

[0096] The cloud platform of the refrigerated truck real-time cargo tracking system continuously monitors data access and computing resource status. When sensor data traffic exceeds a preset warning value and the central processing unit utilization shows a continuous upward trend, the platform automatically activates the load balancing program. This program calls multi-source sensor data streams from the distributed storage cluster, including target weight waveform sampling sequences, spatial location coordinate time series, vehicle positioning information, and motion intensity waveform data. These heterogeneous data are aligned and stitched together using millisecond-level timestamps to construct a multi-dimensional time-series dataset with a unified time axis. Based on the current number of parallel processing tasks, the data throughput of each task, and the computational complexity model, the platform uses a weighted summation algorithm to generate a global load value, which reflects the overall computing pressure level of the cloud cluster in real time. The global load value is updated every five seconds and written to the resource monitoring log, simultaneously triggering subsequent scheduling decision-making processes.

[0097] When the global load reaches the system's preset peak threshold, the cloud scheduling engine initiates a dynamic resource allocation mechanism. The scheduling engine maintains a task queue containing all running analysis models and calculates a dynamic scheduling coefficient for each model instance. This coefficient is a weighted composite of four dimensions: a task urgency factor dynamically assigned based on the perishability level of the goods and alarm status; a data timeliness factor calculated based on a decay function of the data generation timestamp; a task type weight referencing a preset priority table; and historical resource consumption patterns predicted through machine learning models. The engine periodically evaluates the scheduling coefficient and estimated resource consumption of each model instance, ranking actively running model instances in descending order of their scheduling coefficients. When the global load reaches a critical point, the system automatically migrates the top-ranked high-priority model instances to a high-performance computing node cluster, while simultaneously putting the lowest-ranked low-priority model instances into a dormant state or migrating them to a cold storage area. The model state switching operation is logged, including the migration timestamp and the identifiers of the source and target nodes.

[0098] The cloud platform employs a value-driven resource reclamation and preloading strategy to maintain service continuity. The system constructs a value accumulation function for each model instance, dynamically calculating the expected revenue of the instance's continued operation over time. Calculation parameters include scheduling coefficients, data freshness, and user service level agreement weights. The resource reclamation module periodically scans model instances with low value accumulation, deciding whether to immediately release resources or put them into a suspended state based on the time cost difference between cold and warm starts. The preloading module predicts high-value tasks that may arrive within the next two minutes based on recent trends in the value accumulation function, reserving memory space and processor threads in advance on computing nodes. The reserved resource quota is set according to prediction confidence levels: high-confidence predictions allocate a fixed resource pool, while medium-confidence predictions use elastic resource quotas. Global load values ​​are fed back to the scheduling coefficient calculation engine in real time, automatically increasing the weight of task urgency factors when the load remains high. Reserved resource data streams are connected to the load monitoring system to ensure that high-priority tasks immediately obtain computing resources upon arrival, avoiding tracking delays caused by task queue congestion. All resource scheduling events generate timestamped operation records, supporting end-to-end performance auditing and anomaly diagnosis.

[0099] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0100] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for fully automated loading and unloading monitoring and real-time cargo volume tracking of refrigerated trucks, characterized in that, Includes the following steps: Deploy multiple types of intelligent sensing devices to collect raw sensing signals from the loading and unloading area of ​​refrigerated trucks, and extract cargo weight signals, cargo position signals, and action signals corresponding to loading and unloading events from the raw sensing signals; Obtaining the correlation characteristics between the cargo weight signal and the cargo location signal, as well as the adjustment characteristics of the loading and unloading events on the cargo weight signal and / or the cargo location signal, includes the following steps: Calculate the weight change rate and location-weight correlation coefficient during the loading and unloading cycle to obtain correlation characteristics; Establish loading / unloading-position coordination criteria, analyze the stability index and weight recovery rate of the cargo position signal before and after the loading / unloading event, and obtain the adjustment characteristics; The correlation feature is used to adjust the generation process of the target cargo volume probability distribution, and the adjustment feature is used to optimize the interference suppression process; Extracting the action signal corresponding to the loading / unloading event from the original sensing signal includes the following steps: Energy gradient analysis is performed on adjacent signal frames of the original sensing signal to locate loading / unloading candidate intervals and generate loading / unloading event markers. Non-stationary features are decoupled from the marker signal of the loading and unloading event to separate the frequency band feature components that characterize the loading and unloading action; Based on the motion trajectory discrimination criterion, local loading and unloading actions are distinguished from global interference, and the motion intensity waveform is generated by fusing the frequency band feature components. The motion intensity waveform is used to assist in generating the adjustment feature, and the loading / unloading event flag is used to trigger the calculation of the correlation feature; The generation of loading / unloading event markers includes the following steps: The Teager energy operator difference sequence of adjacent signal frames of the original sensing signal is calculated by using a sliding time window; Based on the real-time signal characteristics, a dynamic judgment threshold is obtained; Based on the dynamic determination threshold, the start time and duration range of the loading and unloading event are obtained; The loading and unloading candidate interval is located based on the start time and duration range of the loading and unloading event, and loading and unloading event markers are generated; The output of the loading / unloading event flag is used to control the separation process of the frequency band characteristic components; Based on the correlation features and the adjustment features, a state feature representing the cargo volume status is generated; and based on the state feature, a target cargo volume probability distribution is obtained. The original sensing signal is subjected to interference suppression processing based on the target cargo quantity probability distribution, so as to separate the target weight waveform and target position waveform of the target cargo from the original sensing signal. By combining GPS / BeiDou positioning data and using a cloud-based big data analysis platform, real-time cargo volume tracking is achieved.

2. The method for fully automated loading and unloading monitoring and real-time cargo tracking of refrigerated trucks according to claim 1, characterized in that, The method for distinguishing local loading / unloading actions from global interference based on motion trajectory discrimination criteria, and generating an action intensity waveform by fusing the frequency band feature components, includes the following steps: A three-dimensional motion trajectory map is constructed based on the Doppler phase change, and the Mahalanobis distance between each trajectory point and the historical motion reference trajectory is calculated. When the cumulative Mahalanobis distance of continuous trajectory points exceeds the dynamic interference threshold, it is determined as a local loading and unloading action event and the spatial positioning coordinates are output. The instantaneous energy of the frequency band characteristic components is spatiotemporally aligned with the spatial positioning coordinates to generate a temporally continuous motion intensity waveform; The spatial positioning coordinates are used to optimize the establishment of the loading / unloading-position coordination criterion, and the data stream of the action intensity waveform is used to update the analysis of the weight recovery rate.

3. The method for fully automated loading and unloading monitoring and real-time cargo tracking of refrigerated trucks according to claim 1, characterized in that, After obtaining the correlation characteristics between the cargo weight signal and the cargo location signal, and the adjustment characteristics of the loading and unloading event on the cargo weight signal and / or the cargo location signal, before generating the state characteristics representing the cargo quantity status based on the correlation characteristics and the adjustment characteristics, the following steps are further included: Determine the duration range of the loading / unloading event and the value of a preset threshold, and determine whether there is overlap of multiple target reflection cross-sections; The decision on whether to perform secondary processing is based on the judgment result. The determination of the overlapping of the multi-target reflection cross sections is based on distinguishing between local loading and unloading actions and global interference, and the output of the action intensity waveform generated by fusing frequency band feature components.

4. The method for fully automated loading and unloading monitoring and real-time cargo tracking of refrigerated trucks according to claim 3, characterized in that, After determining the duration range of the loading / unloading event and the value of a preset threshold, and determining whether there is overlap of multiple target reflection sections, the method further includes the following steps: If the duration of the loading and unloading event exceeds a preset threshold, or if there is an overlap of multiple target reflection sections, a multi-source data fusion algorithm is used to perform secondary decoupling processing on the cargo weight signal and the cargo position signal. Based on the three-dimensional motion trajectory map, the overlapping areas of the reflection sections corresponding to global interference are excluded, and the processed weight signal and processed position signal are re-extracted. Based on the processed weight signal, the processed correlation characteristics of the target individual are obtained; based on the processed position signal, the processed adjustment characteristics are obtained; the processed correlation characteristics and the processed adjustment characteristics are used to optimize the generation of the state characteristics; The processed weight signal and processed position signal are used as data inputs to update the calculation of the target cargo quantity probability distribution.

5. The method for fully automated loading and unloading monitoring and real-time cargo tracking of refrigerated trucks according to claim 4, characterized in that, After performing secondary decoupling processing on the cargo weight signal and the cargo position signal, the method further includes the following steps: Verify the stability of the signal energy distribution after the secondary decoupling process; If the verification is successful, the state features are corrected based on the processed correlation features and the processed adjustment features to obtain the processed features, and the processed features are used as input to generate the target cargo volume probability distribution; The verification results of the signal energy distribution stability are used to control the parameter adjustment of the interference suppression process.

6. The method for fully automated loading and unloading monitoring and real-time cargo tracking of refrigerated trucks according to claim 1, characterized in that, The real-time cargo volume tracking through a cloud-based big data analytics platform includes the following steps: When the sensor data traffic exceeds the threshold and the cloud resource utilization continues to rise, a global load value is generated after calling multi-source sensor data and constructing a time-series dataset. When the global load value reaches the system's preset peak threshold, the priority and resource consumption of the evaluation model are assessed based on the scheduling coefficient, and the high-priority model is assigned to the active state. Resource reclamation or preloading is performed based on the value accumulation function and model switching overhead, and cloud resource quotas are reserved based on recent value accumulation function prediction results; The output of the global load value is used to dynamically adjust the calculation of the scheduling coefficient, and the data of the reserved cloud resource quota is used to support the continuity of the real-time cargo volume tracking.

7. The method for fully automated loading and unloading monitoring and real-time cargo tracking of refrigerated trucks according to claim 6, characterized in that, Before generating the global load value after calling multi-source sensor data and constructing a time-series dataset, the following steps are also included: Based on the acquired raw sensing signals, target modal functions are generated through adaptive decomposition. The effective signal is obtained by removing the mode function with the largest total entropy from the target number of mode functions; Based on the valid signal, abnormal data is identified by comparing discrete data, and new valid signals are obtained by eliminating and replacing the abnormal data. Iterative processing continues until the iteration cutoff condition is met to generate a denoised signal; The output of the denoised signal is used to construct the time-series dataset, and the identification results of the outlier data are used to update the evaluation of the value accumulation function.

Images