A refrigerated truck temperature and humidity anomaly early warning processing method and system

Abstract

This invention relates to the field of intelligent monitoring technology for cold chain logistics, specifically to a method and system for handling abnormal temperature and humidity in refrigerated trucks. The method includes: real-time acquisition of multi-dimensional spatiotemporal temperature and humidity data via a multi-node wireless sensor network deployed within the truck compartment; extraction of temperature and humidity trends, fluctuation characteristics, and spatial distribution differences to construct a thermal environment state feature vector for the refrigerated truck; inputting this state feature vector into an abnormal warning model based on isolated forests and gated loop units to predict and identify potential refrigeration system failures, abnormal cargo respiration heat, or truck compartment sealing failures; and, based on the warning level, driving local audible and visual alarms and automatically adjusting refrigeration unit parameters via an edge computing gateway, and pushing the warning information and handling suggestions to a remote monitoring center and the driver's terminal via an IoT communication module. This invention improves the safety and energy efficiency of goods transported in cold chain logistics.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology for cold chain logistics, specifically to a method and system for early warning and handling of abnormal temperature and humidity in refrigerated trucks. Background Technology

[0002] As the core carrier of cold chain transportation, the precise control of temperature and humidity inside refrigerated trucks directly determines the quality and safety of goods. For example, COVID-19 vaccines need to be transported in ultra-low temperature or constant temperature environments, and even slight deviations can render them ineffective; fresh fruits and vegetables are extremely sensitive to humidity and temperature fluctuations, and improper environments can accelerate spoilage.

[0003] However, in actual transportation, refrigerated trucks face many uncontrollable factors: refrigeration system failures lead to a decrease or loss of refrigeration capacity; goods are not sufficiently pre-cooled, improper stacking methods hinder cold air circulation, and some fruits and vegetables have excessive respiration heat; frequent opening and closing of doors leads to cold air leakage, and aging and damage to the truck compartment seals introduce external hot and humid air; traditional single or a few temperature and humidity sensors may fail or be maliciously blocked, resulting in data distortion.

[0004] Currently, temperature and humidity monitoring in the industry generally remains at the threshold alarm stage: an alarm is triggered when the sensor reading exceeds the set upper or lower limits. This approach has a fundamental flaw: by the time an alarm occurs, the goods have often already been exposed to the abnormal environment for a considerable period, resulting in irreversible losses—a typical case of reactive detection. There is an urgent need for an intelligent solution capable of providing early warnings, diagnosing causes, and proactively addressing issues.

[0005] Among existing technologies, the most similar solution is a refrigerated truck temperature and humidity monitoring system based on wireless sensor networks and threshold comparison. This system deploys multiple wireless temperature and humidity sensor nodes inside the truck compartment, collecting data to an onboard terminal via ZigBee or LoRa protocols. The onboard terminal compares the real-time value of each sensor with a preset static threshold. When any sensor value exceeds the threshold, the system activates a buzzer alarm and uploads the alarm information and location to the monitoring center via a GPS / GPRS module.

[0006] The existing technical solution has the following technical defects and limitations: Early warning is severely delayed, essentially a reactive measure, only triggering an alarm when the temperature has already exceeded the limit and the goods have begun to deteriorate, failing to provide early warnings at or before an anomaly occurs; it has a high false alarm rate and lacks intelligent diagnostics, unable to distinguish between occasional noise interference, sensor malfunction, and genuine refrigeration failure. For example, a brief temperature rise caused by opening the door to retrieve goods can trigger a false alarm, while a gradual temperature rise caused by a slow decline in refrigeration efficiency will not trigger an alarm before exceeding the threshold, but has already had a continuous impact on the goods; the information dimension is limited and lacks traceability, relying solely on a single temperature value without analyzing the trends, rates, spatial distribution differences, and coupling relationships between temperature and humidity, thus failing to determine the root cause of the anomaly, and preventing drivers and maintenance personnel from quickly taking targeted measures; it lacks closed-loop control, only providing alarms without linkage with refrigeration units, ventilation dampers, and other actuators, failing to automatically intervene when an anomaly occurs; and it has poor adaptability, with static thresholds unable to adapt to the significant differences in the temperature and humidity sensitivity of different goods, as well as dynamic heat load changes under different seasons and driving conditions. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a method and system for handling abnormal temperature and humidity in refrigerated trucks. It aims to achieve a leap from post-event alarms to pre-event warnings and from single-point judgments to intelligent diagnosis through a technical chain of multi-source sensing, feature mining, intelligent early warning, and collaborative handling, thereby significantly improving the safety and energy efficiency of goods transported in the cold chain.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] On one hand, embodiments of the present invention provide a method for handling abnormal temperature and humidity in refrigerated trucks, the method comprising the following steps:

[0010] S100: Acquire multidimensional temperature and humidity spatiotemporal sequence data inside the refrigerated truck compartment and a trained hybrid anomaly early warning model. The hybrid anomaly early warning model includes a feature extraction module, an isolated forest screening module, a gated loop unit confirmation module, and a root cause diagnosis module.

[0011] S200, input the multidimensional temperature and humidity spatiotemporal sequence data into the feature extraction module, extract trend features, fluctuation features, spatial difference features and coupling features through a sliding time window, and construct a feature vector of the thermal environment state of the refrigerated truck.

[0012] S300, input the feature vector of the thermal environment state of the refrigerated truck into the isolated forest screening module, output the anomaly score through the isolated forest model, compare the anomaly score with the preset screening threshold, and screen out potential anomaly windows;

[0013] S400, input the original temperature and humidity time series data corresponding to the potential anomaly window into the gated loop unit confirmation module, perform anomaly confirmation and trend prediction through the gated loop unit network, and output the anomaly confirmation result and the future temperature and humidity change prediction sequence.

[0014] S500, the anomaly confirmation result is input into the root cause diagnosis module, and the current feature vector is matched with the predefined anomaly pattern feature vector through grey relational analysis, and the anomaly root cause diagnosis result and differentiated treatment strategy are output.

[0015] S600, according to the differentiated handling strategy, drives the local audible and visual alarm, refrigeration unit controller and driver terminal to perform coordinated handling operations through the edge computing gateway, and pushes the abnormal root cause diagnosis results to the remote monitoring center.

[0016] Optionally, in S200, the step of extracting trend features, fluctuation features, spatial difference features, and coupling features through a sliding time window to construct a feature vector of the thermal environment state of the refrigerated truck includes:

[0017] S210, by fitting the temperature and humidity sequences within the sliding time window using linear regression, the slopes of temperature and humidity changes are calculated, and abrupt changes within the sliding time window are identified using a Bayesian change point detection algorithm to generate trend features;

[0018] S220, Calculate the standard deviation of temperature, the standard deviation of humidity, and the maximum fluctuation amplitude of temperature within the sliding time window, and generate fluctuation characteristics;

[0019] S230, calculate the difference between the highest and lowest temperatures inside the carriage at the same time to generate spatial difference features;

[0020] S240 calculates the ratio of temperature change rate to humidity change rate and the correlation coefficient between temperature and humidity, generating coupling features;

[0021] S250, the trend feature, the fluctuation feature, the spatial difference feature and the coupling feature are spliced ​​together to construct the thermal environment state feature vector of the refrigerated truck.

[0022] Optionally, in S210, identifying the abrupt change point within the sliding time window using the Bayesian change point detection algorithm includes:

[0023] S211, Using the temperature and humidity sequences within the sliding time window as input data, a Bayesian probability model is established;

[0024] S212, Calculate the posterior probability of each time point in the input data as a change point;

[0025] S213, identify the time points when the posterior probability exceeds a preset probability threshold as mutation points, and record the number and location information of the mutation points;

[0026] S214, the number and location information of the mutation points are incorporated into the trend features as mutation point detection features.

[0027] Optionally, in S300, the step of outputting anomaly scores through an isolated forest model, comparing the anomaly scores with a preset screening threshold, and filtering out potential anomaly windows includes:

[0028] S310, Input the feature vector of the thermal environment state of the refrigerated truck into an isolated forest model composed of multiple isolated trees;

[0029] S320, For each isolated tree, randomly select a feature and a splitting value, recursively divide the data space, and calculate the partitioning path length required for the refrigerated truck thermal environment state feature vector to be isolated;

[0030] S330, take the average of the partition path lengths of all isolated trees, and calculate the anomaly score based on the average value;

[0031] S340, compare the anomaly score with the preset screening threshold, and when the anomaly score is greater than the preset screening threshold, mark the corresponding sliding time window as a potential anomaly window.

[0032] Optionally, in S400, the step of performing anomaly confirmation and trend prediction through a gated cyclic unit network, and outputting anomaly confirmation results and a predicted sequence of future temperature and humidity changes, includes:

[0033] S410, input the original temperature and humidity time series data corresponding to the potential abnormal window into the gated recurrent unit network, and perform nonlinear transformation on the input data through update gate and reset gate to extract time-dependent features;

[0034] S420, the time-dependent features are input into the fully connected classification layer, the anomaly probability is calculated by the Softmax function, and the anomaly confirmation result is output;

[0035] S430, the time-dependent features are input into a fully connected regression layer, and the predicted temperature and humidity values ​​at multiple future time points are calculated through linear mapping, and the predicted sequence of future temperature and humidity changes is output.

[0036] S440, compare the predicted sequence of future temperature and humidity changes with a preset safety threshold, and trigger a forward-looking warning when the predicted temperature exceeds the preset safety threshold.

[0037] Optionally, in S500, the step of matching the current feature vector with the predefined abnormal pattern feature vector through grey relational analysis to output the abnormal root cause diagnosis result and differentiated treatment strategy includes:

[0038] S510, perform gray relational calculations on the current feature vector and the predefined refrigeration system failure mode feature vector, the compartment sealing failure mode feature vector, the frequent door opening and closing mode feature vector, and the cargo self-heating mode feature vector, respectively.

[0039] S520, For each abnormal mode, calculate the difference sequence of features in each dimension between the current feature vector and the feature vector of the abnormal mode;

[0040] S530, calculate the correlation coefficient based on the difference sequence, and take the average of the correlation coefficients of all dimensions to obtain the gray correlation degree between the current feature vector and each abnormal pattern;

[0041] S540, the abnormal pattern with the highest gray correlation degree is determined as the abnormal root cause diagnosis result, and the differentiated handling strategy is generated according to the warning level and handling rules corresponding to the abnormal root cause diagnosis result.

[0042] Optionally, in S600, the step of driving the local audible and visual alarm, the refrigeration unit controller, and the driver terminal to perform coordinated handling operations via the edge computing gateway includes:

[0043] When the differentiated handling strategy corresponds to a general warning level, the edge computing gateway sends instructions to the refrigeration unit controller through the controller local area network bus to increase the compressor speed or increase the opening of the electronic expansion valve.

[0044] When the differentiated handling strategy corresponds to a severe warning level, the edge computing gateway drives the local audible and visual alarm and displays abnormal information on the driver's terminal;

[0045] When the differentiated handling strategy corresponds to the emergency warning level, the edge computing gateway controls the ventilation damper to open and pushes the warning event, root cause diagnosis results and vehicle location information to the remote monitoring center through the Internet of Things communication module.

[0046] On the other hand, embodiments of the present invention provide a refrigerated truck temperature and humidity abnormality early warning and processing system, including:

[0047] At least one processor;

[0048] At least one memory for storing at least one program;

[0049] When the at least one program is executed by the at least one processor, the at least one processor performs the method described above.

[0050] On the other hand, embodiments of the present invention provide a computer-readable storage medium storing a processor-executable program, which, when executed by a processor, is used to perform the above-described method.

[0051] On the other hand, embodiments of the present invention provide a computer program product, including a computer program or computer instructions, the computer program or computer instructions being stored in a memory, a processor of a computer device reading the computer program or computer instructions from the memory, and the processor executing the computer program or computer instructions to cause the computer device to perform the above-described method.

[0052] The embodiments of the present invention have the following beneficial effects:

[0053] The present invention proposes a method for characterizing the thermal environment of refrigerated trucks based on multi-dimensional temporal feature fusion. This method integrates temperature and humidity trends, fluctuations, spatial differences, and temperature-humidity coupling features to construct a high-dimensional feature vector for anomaly detection. This overcomes the technical defects of existing technologies, such as single information dimension and lack of traceability, and provides a rich feature foundation for accurate anomaly detection.

[0054] The present invention designs a hybrid anomaly early warning model based on isolated forest and gated loop unit. It uses isolated forest for rapid initial screening and then uses gated loop unit for accurate anomaly confirmation and future trend prediction. This achieves low false alarm and forward-looking early warning, effectively solving the technical problems of serious lag and high false alarm rate in existing technologies. It can predict the upcoming over-temperature event in advance, leaving valuable intervention time for the driver or automatic control system.

[0055] The present invention proposes an intelligent diagnostic method for abnormal temperature and humidity in refrigerated trucks based on grey relational analysis. By matching real-time features with a typical fault mode library using grey relational analysis, the method automatically identifies the source of the abnormality. This overcomes the shortcomings of existing technologies, such as lack of intelligent diagnosis and inability to determine the root cause of the abnormality, and significantly reduces manual judgment and response time.

[0056] The integrated closed-loop system for early warning and response constructed in this invention realizes a multi-level collaborative response closed loop based on the early warning level and root cause type, including local audible and visual alarms, driver terminal notifications, remote monitoring center push notifications, and automatic adjustment of refrigeration unit parameters. This overcomes the technical defects of existing technologies that lack closed-loop control and only provide alarms without taking action.

[0057] The proposed architecture for continuous model evolution based on digital twins and incremental learning utilizes feedback from cloud-based digital twin models and actual operational data to periodically incrementally train and update the early warning model at the edge using over-the-air (OTA) technology. This enables the system to have adaptive and continuous optimization capabilities, overcoming the technical defects of existing models that are fixed and unable to evolve. It can adapt to changes such as vehicle aging, seasonal changes, and new cargo types without the need for frequent manual parameter adjustments. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 This is a flowchart illustrating the method for handling abnormal temperature and humidity in refrigerated trucks in an embodiment of the present invention.

[0060] Figure 2 This is a schematic diagram of the overall architecture of the refrigerated truck temperature and humidity abnormality early warning and processing system in this embodiment of the invention. Detailed Implementation

[0061] The following will provide a clear and complete description of the concept, specific structure, and technical effects of the present invention in conjunction with embodiments and accompanying drawings, so as to fully understand the purpose, solution, and effects of the present invention. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0062] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of this invention; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this invention as detailed in the appended claims.

[0063] It is understood that the terms "first," "second," etc., used in this invention can be used to describe various concepts, but unless specifically stated otherwise, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of embodiments of this invention, first information can also be referred to as second information, and similarly, second information can also be referred to as first information. Depending on the context, words such as "if" and "if" as used herein can be interpreted as "at" or "when" or "in response to determination."

[0064] The terms used in this invention, such as at least one, multiple, each, any, etc., include at least one, two or more, multiple, each, each, and any, where at least one includes one, two, or more, multiple includes two or more, each refers to each of the corresponding multiple, and any refers to any one of the multiple.

[0065] Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this invention is for descriptive purposes only and is not intended to limit the invention.

[0066] refer to Figure 1 ,like Figure 1 The image shows a method for handling abnormal temperature and humidity in refrigerated trucks according to an embodiment of the present invention. The method includes the following steps:

[0067] S100: Acquire multidimensional temperature and humidity spatiotemporal sequence data inside the refrigerated truck compartment and a trained hybrid anomaly early warning model. The hybrid anomaly early warning model includes a feature extraction module, an isolated forest screening module, a gated loop unit confirmation module, and a root cause diagnosis module.

[0068] S200, input the multidimensional temperature and humidity spatiotemporal sequence data into the feature extraction module, extract trend features, fluctuation features, spatial difference features and coupling features through a sliding time window, and construct a feature vector of the thermal environment state of the refrigerated truck.

[0069] S300, input the feature vector of the thermal environment state of the refrigerated truck into the isolated forest screening module, output the anomaly score through the isolated forest model, compare the anomaly score with the preset screening threshold, and screen out potential anomaly windows;

[0070] S400, input the original temperature and humidity time series data corresponding to the potential anomaly window into the gated loop unit confirmation module, perform anomaly confirmation and trend prediction through the gated loop unit network, and output the anomaly confirmation result and the future temperature and humidity change prediction sequence.

[0071] S500, the anomaly confirmation result is input into the root cause diagnosis module, and the current feature vector is matched with the predefined anomaly pattern feature vector through grey relational analysis, and the anomaly root cause diagnosis result and differentiated treatment strategy are output.

[0072] S600, according to the differentiated handling strategy, drives the local audible and visual alarm, refrigeration unit controller and driver terminal to perform coordinated handling operations through the edge computing gateway, and pushes the abnormal root cause diagnosis results to the remote monitoring center.

[0073] This invention provides a method and system for handling abnormal temperature and humidity in refrigerated trucks. It comprehensively collects real-time operational status data of refrigerated trucks through a multi-source sensing layer, utilizes a feature extraction module and a hybrid abnormality warning model in an edge computing layer for deep feature mining and intelligent abnormality detection, then achieves integrated closed-loop control of early warning and handling through a collaborative processing layer, and finally endows the system with continuous learning and evolution capabilities through a digital twin model and incremental learning module in a cloud platform layer. This invention effectively addresses the core challenges of existing technologies, such as delayed early warning, high false alarm rate, inability to diagnose the root cause of abnormalities, lack of closed-loop control, and model rigidity. By constructing a technical chain of multi-source sensing, feature mining, intelligent early warning, and collaborative handling, it significantly improves the safety and energy efficiency of goods transported in cold chain transportation.

[0074] The core of this embodiment lies in the network framework of the hybrid anomaly early warning model, which realizes intelligent early warning and handling of abnormal temperature and humidity in refrigerated trucks through four stages: feature extraction, rapid screening, accurate confirmation, and root cause diagnosis.

[0075] Data acquisition and model preparation: Acquire multidimensional temperature and humidity spatiotemporal sequence data inside the refrigerated truck compartment and load a pre-trained hybrid anomaly early warning model.

[0076] like Figure 2 As shown, the overall system architecture comprises four core layers: a multi-source sensing layer, an edge computing layer, a collaborative processing layer, and a cloud platform layer. The multi-source sensing layer is responsible for comprehensively and in real-time collecting refrigerated truck operating status data; the edge computing layer is the system's core early warning processing unit; the collaborative processing layer executes differentiated actions based on the early warning level and root cause type; and the cloud platform layer provides digital twin models and incremental learning capabilities. The layers interact via IoT communication protocols such as 5G or Cat.1.

[0077] The multi-source sensing layer specifically includes: a multi-point temperature and humidity sensor array, deploying at least eight digital temperature and humidity sensors at key locations within the vehicle compartment (front, middle, rear, left, right, top, bottom, and return air vents), with a sampling frequency of no less than 1 Hz and an accuracy requirement of ±0.3 degrees Celsius for temperature and ±3% for relative humidity, used to capture temperature stratification and local hotspots; auxiliary sensing information, accessing door open / close status sensors and reading the refrigeration unit's operating parameters via the controller area network bus. All sensor data is collected to the edge computing gateway via RS-485 or wireless means, with data packets including precise timestamps.

[0078] The edge computing layer is deployed on the vehicle edge computing gateway, based on the ARM architecture, and has a computing power of more than 0.5 trillion operations per second.

[0079] In some embodiments, in S200, the step of extracting trend features, fluctuation features, spatial difference features, and coupling features through a sliding time window to construct a feature vector of the thermal environment state of the refrigerated truck includes:

[0080] S210, by fitting the temperature and humidity sequences within the sliding time window using linear regression, the slopes of temperature and humidity changes are calculated, and abrupt changes within the sliding time window are identified using a Bayesian change point detection algorithm to generate trend features;

[0081] S220, Calculate the standard deviation of temperature, the standard deviation of humidity, and the maximum fluctuation amplitude of temperature within the sliding time window, and generate fluctuation characteristics;

[0082] S230, calculate the difference between the highest and lowest temperatures inside the carriage at the same time to generate spatial difference features;

[0083] S240 calculates the ratio of temperature change rate to humidity change rate and the correlation coefficient between temperature and humidity, generating coupling features;

[0084] S250, the trend feature, the fluctuation feature, the spatial difference feature and the coupling feature are spliced ​​together to construct the thermal environment state feature vector of the refrigerated truck.

[0085] Specifically, a 10-minute window (600 sampling points) with a 1-minute sliding step was used to perform feature engineering on the temperature and humidity time-series data within the window. The extracted feature vectors include: trend features (calculating the slope of temperature change by fitting the temperature sequence within the window using linear regression, calculating the slope of humidity change by fitting the humidity sequence within the window using linear regression, and using a Bayesian change point detection algorithm to identify any abrupt changes within the window); fluctuation features (calculating the standard deviation of temperature and humidity within the window, as well as the maximum temperature fluctuation amplitude); spatial difference features (calculating the difference between the highest and lowest temperatures inside the carriage at the same time); and coupling features (calculating the ratio of the temperature change rate to the humidity change rate, and the correlation coefficient between temperature and humidity).

[0086] In some embodiments, S210, identifying abrupt change points within the sliding time window using a Bayesian change point detection algorithm includes:

[0087] S211, Using the temperature and humidity sequences within the sliding time window as input data, a Bayesian probability model is established;

[0088] S212, Calculate the posterior probability of each time point in the input data as a change point;

[0089] S213, identify the time points when the posterior probability exceeds a preset probability threshold as mutation points, and record the number and location information of the mutation points;

[0090] S214, the number and location information of the mutation points are incorporated into the trend features as mutation point detection features.

[0091] In this embodiment, the Bayesian change point detection algorithm establishes a probability model to calculate the posterior probability of each time point as a change point. When the posterior probability exceeds a preset probability threshold, the time point is identified as a change point. This probability-based change point detection method can effectively identify the moments of statistical characteristic abrupt changes in temperature and humidity sequences, providing a reliable basis for determining the starting point of temperature and humidity changes.

[0092] In some embodiments, S300, the step of outputting anomaly scores through an isolated forest model, comparing the anomaly scores with a preset screening threshold, and filtering out potential anomaly windows includes:

[0093] S310, Input the feature vector of the thermal environment state of the refrigerated truck into an isolated forest model composed of multiple isolated trees;

[0094] S320, For each isolated tree, randomly select a feature and a splitting value, recursively divide the data space, and calculate the partitioning path length required for the refrigerated truck thermal environment state feature vector to be isolated;

[0095] S330, take the average of the partition path lengths of all isolated trees, and calculate the anomaly score based on the average value;

[0096] S340, compare the anomaly score with the preset screening threshold, and when the anomaly score is greater than the preset screening threshold, mark the corresponding sliding time window as a potential anomaly window.

[0097] In this embodiment, the isolated forest model consists of multiple isolated trees. Each isolated tree recursively partitions the data space by randomly selecting features and split values. Outliers, due to their small number and large differences in feature values, are easier to isolate and thus quickly identified. An anomaly score is obtained by calculating the path length required to isolate the feature vector and averaging the path lengths of all isolated trees. A higher anomaly score indicates a more isolated data pattern within that window, meaning a greater likelihood of an anomaly. This step is used to quickly filter out potential anomaly windows; it is computationally fast and suitable for real-time operation at edge environments.

[0098] In some embodiments, S400, the step of performing anomaly confirmation and trend prediction through a gated recurrent unit network, and outputting anomaly confirmation results and a future temperature and humidity change prediction sequence, includes:

[0099] S410, input the original temperature and humidity time series data corresponding to the potential abnormal window into the gated recurrent unit network, and perform nonlinear transformation on the input data through update gate and reset gate to extract time-dependent features;

[0100] S420, the time-dependent features are input into the fully connected classification layer, the anomaly probability is calculated by the Softmax function, and the anomaly confirmation result is output;

[0101] S430, the time-dependent features are input into a fully connected regression layer, and the predicted temperature and humidity values ​​at multiple future time points are calculated through linear mapping, and the predicted sequence of future temperature and humidity changes is output.

[0102] S440, compare the predicted sequence of future temperature and humidity changes with a preset safety threshold, and trigger a forward-looking warning when the predicted temperature exceeds the preset safety threshold.

[0103] In this embodiment, the gated recurrent unit network effectively solves the long-term dependency problem of traditional recurrent neural networks through the update gate and reset gate mechanism, and is good at learning and memorizing patterns in long-term series data. For potential abnormal windows screened by the isolated forest, the temperature sequence points within the original window are input into the pre-trained gated recurrent unit network model. The model has two tasks: Task A is binary classification, outputting whether it is a real abnormal event, effectively filtering false alarms caused by instantaneous sensor interference or a single gate opening; Task B is regression prediction, predicting the temperature change sequence within the next 5 minutes. If the predicted temperature will exceed the preset safety threshold within the next 5 minutes, an early warning is issued, realizing true pre-emptive prediction.

[0104] In some embodiments, S500, the step of matching the current feature vector with the predefined abnormal pattern feature vector through grey relational analysis to output the abnormal root cause diagnosis result and differentiated treatment strategy includes:

[0105] S510, perform gray relational calculations on the current feature vector and the predefined refrigeration system failure mode feature vector, the compartment sealing failure mode feature vector, the frequent door opening and closing mode feature vector, and the cargo self-heating mode feature vector, respectively.

[0106] S520, For each abnormal mode, calculate the difference sequence of features in each dimension between the current feature vector and the feature vector of the abnormal mode;

[0107] S530, calculate the correlation coefficient based on the difference sequence, and take the average of the correlation coefficients of all dimensions to obtain the gray correlation degree between the current feature vector and each abnormal pattern;

[0108] S540, the abnormal pattern with the highest gray correlation degree is determined as the abnormal root cause diagnosis result, and the differentiated handling strategy is generated according to the warning level and handling rules corresponding to the abnormal root cause diagnosis result.

[0109] In this embodiment, grey relational analysis is a multi-factor statistical analysis method that measures the degree of correlation between factors based on the similarity or dissimilarity of their development trends, i.e., the grey relational degree. When the model determines a true anomaly, the root cause diagnosis engine is activated. This engine performs grey relational analysis on the current feature vector and the feature vectors of several predefined typical anomaly patterns. Typical anomaly patterns include: a refrigeration system failure mode, characterized by a slow and continuous rise in global temperature, a positive and stable temperature change slope, a small temperature standard deviation, and normal spatial temperature differences; a compartment sealing failure mode, characterized by a rapid and simultaneous rise in temperature and humidity, with both temperature and humidity change slopes being large positive values, and a strong positive correlation coefficient between temperature and humidity; a frequent door opening and closing mode, characterized by multiple V-shaped spikes in the temperature sequence, multiple change points identified by change point detection, and a large temperature standard deviation; and a cargo self-heating mode, characterized by a local sensor temperature consistently higher than other areas, and a continuously increasing spatial temperature difference. The mode with the highest correlation degree is diagnosed as the root cause of the anomaly, and corresponding processing suggestions are output.

[0110] In some embodiments, S600, the step of driving the local audible and visual alarm, the refrigeration unit controller, and the driver terminal to perform coordinated handling operations via the edge computing gateway includes:

[0111] When the differentiated handling strategy corresponds to a general warning level, the edge computing gateway sends instructions to the refrigeration unit controller through the controller local area network bus to increase the compressor speed or increase the opening of the electronic expansion valve.

[0112] When the differentiated handling strategy corresponds to a severe warning level, the edge computing gateway drives the local audible and visual alarm and displays abnormal information on the driver's terminal;

[0113] When the differentiated handling strategy corresponds to the emergency warning level, the edge computing gateway controls the ventilation damper to open and pushes the warning event, root cause diagnosis results and vehicle location information to the remote monitoring center through the Internet of Things communication module.

[0114] In this embodiment, differentiated handling is performed based on the anomaly level and root cause type output by the early warning model. Local linkage includes: general early warning, such as predicting that the limit will be exceeded in 5 minutes, sending instructions via the controller LAN bus to increase the compressor speed of the refrigeration unit or increase the opening of the electronic expansion valve for proactive intervention; severe early warning, where the limit has been exceeded and the trend is worsening, immediately triggering the audible and visual alarm in the driver's cab, and simultaneously displaying specific abnormal information on the driver's terminal; emergency early warning, such as a sharp drop in temperature indicating refrigeration runaway, in addition to the above actions, automatically controlling the opening of some ventilation dampers. Remote push includes: all early warning events, root cause diagnosis results, real-time data, and vehicle location are pushed to the large screen of the remote monitoring center and the mobile terminal of the fleet administrator within 1 second via the IoT communication module.

[0115] The cloud platform layer comprises a digital twin model and an incremental learning module. The digital twin model creates a digital twin for each refrigerated truck in the cloud, continuously synchronizing real-time vehicle data and running a more refined physical model or a large-scale deep learning model to simulate the dynamic evolution of the temperature and humidity field inside the truck compartment. The incremental learning module, through data collection, manual annotation, feature extraction, adding new data to the training set, regular retraining, and model performance evaluation, enables the model to continuously absorb new operational data. Drivers or monitors can confirm or correct warning results. The cloud platform regularly collects new data and corrected labels from all vehicles, using this data to incrementally train the gated recurrent unit classification model and the isolated forest model, updating model parameters. New, performance-enhanced models are silently downloaded and updated to each edge computing gateway via over-the-air (OTA) technology, enabling the system to adapt to long-term drift such as seasonal changes and vehicle aging, achieving increasing accuracy with use.

[0116] This invention also provides a refrigerated truck temperature and humidity abnormality early warning and processing system, including:

[0117] At least one processor;

[0118] At least one memory for storing at least one program;

[0119] When the at least one program is executed by the at least one processor, the at least one processor performs the method described above.

[0120] The content of the above method embodiments is applicable to this embodiment. The specific functions implemented in this embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments. Therefore, they will not be repeated here.

[0121] This invention also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described above. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0122] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0123] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0124] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0125] This invention also provides a computer program product, including a computer program or computer instructions, which are stored in a memory. A processor of a computer device reads the computer program or computer instructions from the memory and executes the computer program or computer instructions, causing the computer device to perform the above-described method.

[0126] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0127] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0128] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically include computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0129] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

Claims

1. A method for handling abnormal temperature and humidity in refrigerated trucks, characterized in that, The method includes the following steps: S100: Acquire multidimensional temperature and humidity spatiotemporal sequence data inside the refrigerated truck compartment and a trained hybrid anomaly early warning model. The hybrid anomaly early warning model includes a feature extraction module, an isolated forest screening module, a gated loop unit confirmation module, and a root cause diagnosis module. S200, input the multidimensional temperature and humidity spatiotemporal sequence data into the feature extraction module, extract trend features, fluctuation features, spatial difference features and coupling features through a sliding time window, and construct a feature vector of the thermal environment state of the refrigerated truck. S300, input the feature vector of the thermal environment state of the refrigerated truck into the isolated forest screening module, output the anomaly score through the isolated forest model, compare the anomaly score with the preset screening threshold, and screen out potential anomaly windows; S400, input the original temperature and humidity time series data corresponding to the potential anomaly window into the gated loop unit confirmation module, perform anomaly confirmation and trend prediction through the gated loop unit network, and output the anomaly confirmation result and the future temperature and humidity change prediction sequence. S500, the anomaly confirmation result is input into the root cause diagnosis module, and the current feature vector is matched with the predefined anomaly pattern feature vector through grey relational analysis, and the anomaly root cause diagnosis result and differentiated treatment strategy are output. S600, according to the differentiated handling strategy, drives the local audible and visual alarm, refrigeration unit controller and driver terminal to perform coordinated handling operations through the edge computing gateway, and pushes the abnormal root cause diagnosis results to the remote monitoring center.

2. The method according to claim 1, characterized in that, In S200, the step of extracting trend features, fluctuation features, spatial difference features, and coupling features through a sliding time window to construct a feature vector of the thermal environment state of the refrigerated truck includes: S210, by fitting the temperature and humidity sequences within the sliding time window using linear regression, the slopes of temperature and humidity changes are calculated, and abrupt changes within the sliding time window are identified using a Bayesian change point detection algorithm to generate trend features; S220, Calculate the standard deviation of temperature, the standard deviation of humidity, and the maximum fluctuation amplitude of temperature within the sliding time window, and generate fluctuation characteristics; S230, calculate the difference between the highest and lowest temperatures inside the carriage at the same time to generate spatial difference features; S240 calculates the ratio of temperature change rate to humidity change rate and the correlation coefficient between temperature and humidity, generating coupling features; S250, the trend feature, the fluctuation feature, the spatial difference feature and the coupling feature are spliced ​​together to construct the thermal environment state feature vector of the refrigerated truck.

3. The method according to claim 2, characterized in that, In S210, identifying abrupt change points within the sliding time window using the Bayesian change point detection algorithm includes: S211, Using the temperature and humidity sequences within the sliding time window as input data, a Bayesian probability model is established; S212, Calculate the posterior probability of each time point in the input data as a change point; S213, identify the time points when the posterior probability exceeds a preset probability threshold as mutation points, and record the number and location information of the mutation points; S214, the number and location information of the mutation points are incorporated into the trend features as mutation point detection features.

4. The method according to claim 1, characterized in that, In S300, the step of outputting anomaly scores through an isolated forest model, comparing the anomaly scores with a preset screening threshold, and filtering out potential anomaly windows includes: S310, Input the feature vector of the refrigerated truck's thermal environment state into an isolated forest model composed of multiple isolated trees; S320, For each isolated tree, randomly select a feature and a splitting value, recursively divide the data space, and calculate the partitioning path length required for the refrigerated truck thermal environment state feature vector to be isolated; S330, take the average of the partition path lengths of all isolated trees, and calculate the anomaly score based on the average value; S340, compare the anomaly score with the preset screening threshold, and when the anomaly score is greater than the preset screening threshold, mark the corresponding sliding time window as a potential anomaly window.

5. The method according to claim 1, characterized in that, In S400, the anomaly confirmation and trend prediction performed through the gated recurrent unit network, and the output of the anomaly confirmation result and the future temperature and humidity change prediction sequence, include: S410, input the original temperature and humidity time series data corresponding to the potential abnormal window into the gated recurrent unit network, and perform nonlinear transformation on the input data through update gate and reset gate to extract time-dependent features; S420, the time-dependent features are input into the fully connected classification layer, the anomaly probability is calculated by the Softmax function, and the anomaly confirmation result is output; S430, the time-dependent features are input into a fully connected regression layer, and the predicted temperature and humidity values ​​at multiple future time points are calculated through linear mapping, and the predicted sequence of future temperature and humidity changes is output. S440, compare the predicted sequence of future temperature and humidity changes with a preset safety threshold, and trigger a forward-looking warning when the predicted temperature exceeds the preset safety threshold.

6. The method according to claim 1, characterized in that, In S500, the step of matching the current feature vector with the predefined abnormal pattern feature vector through grey relational analysis to output the abnormal root cause diagnosis result and differentiated treatment strategy includes: S510, perform gray relational calculations on the current feature vector and the predefined refrigeration system failure mode feature vector, the compartment sealing failure mode feature vector, the frequent door opening and closing mode feature vector, and the cargo self-heating mode feature vector, respectively. S520, For each abnormal mode, calculate the difference sequence of features in each dimension between the current feature vector and the feature vector of the abnormal mode; S530, calculate the correlation coefficient based on the difference sequence, and take the average of the correlation coefficients of all dimensions to obtain the gray correlation degree between the current feature vector and each abnormal pattern; S540, the abnormal pattern with the highest gray correlation degree is determined as the abnormal root cause diagnosis result, and the differentiated handling strategy is generated according to the warning level and handling rules corresponding to the abnormal root cause diagnosis result.

7. The method according to claim 1, characterized in that, In S600, the step of driving the local audible and visual alarm, the refrigeration unit controller, and the driver terminal to perform coordinated handling operations via the edge computing gateway includes: When the differentiated handling strategy corresponds to a general warning level, the edge computing gateway sends instructions to the refrigeration unit controller through the controller local area network bus to increase the compressor speed or increase the opening of the electronic expansion valve. When the differentiated handling strategy corresponds to a severe warning level, the edge computing gateway drives the local audible and visual alarm and displays abnormal information on the driver's terminal; When the differentiated handling strategy corresponds to the emergency warning level, the edge computing gateway controls the ventilation damper to open and pushes the warning event, root cause diagnosis results and vehicle location information to the remote monitoring center through the Internet of Things communication module.

8. A refrigerated truck temperature and humidity abnormality early warning and handling system, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor performs the method as described in any one of claims 1 to 7.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images