Multi-zone adaptive dynamic temperature control method and system for refrigerated semi-trailers
By constructing a temperature spatiotemporal causal map and heat flow exchange matrix for temperature zones, the problems of low accuracy in temperature anomaly identification and insufficient global control in the temperature control technology of multi-temperature zone refrigerated semi-trailers are solved, achieving precise control and energy optimization, and reducing the risk of cargo loss.
Patent Information
- Application Number
- CN202611207919.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-11
- Publication Date
- 2026-11-17
AI Technical Summary
Existing temperature control technologies for multi-temperature zone refrigerated semi-trailers suffer from low accuracy in identifying temperature anomalies, high false alarm rates, insufficient global control capabilities, energy waste, and risks of cargo damage. They are also unable to quickly pinpoint the root cause of anomalies and implement precise control measures.
A multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers is constructed. By building a spatiotemporal causal map of temperature zones, integrating temporal causality and spatial heat conduction, temperature anomalies are accurately identified. A heat flow exchange matrix is constructed to optimize the global allocation of cooling capacity, and subdivided into sub-temperature zones for targeted and precise control.
It enables accurate identification and rapid location of temperature anomalies, reduces false alarm rates, optimizes cooling capacity distribution, shortens anomaly recovery time, and reduces energy consumption and cargo loss risks.
Smart Images

Figure CN123114119A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of temperature control technology, and more specifically, to a method and system for multi-temperature zone adaptive dynamic temperature control of refrigerated semi-trailers. Background Technology
[0002] With the rapid development of the cold chain logistics industry, refrigerated semi-trailers have become core equipment for long-distance transportation of temperature-sensitive goods such as fresh food and pharmaceutical reagents. To improve transportation efficiency and reduce transportation costs, multi-temperature zone refrigerated semi-trailers capable of transporting multiple goods with different temperature requirements simultaneously have been widely used. However, the temperature control technology of existing multi-temperature zone refrigerated semi-trailers still has many shortcomings, which seriously affect the safety and economy of cold chain transportation.
[0003] First, the accuracy of temperature anomaly identification is low, with significant issues of false alarms and missed alarms. Most existing technologies judge temperature anomalies by comparing real-time temperature with a fixed set temperature threshold. This cannot distinguish between short-term temperature fluctuations caused by normal operations such as vehicle start-stop, door opening and closing, and loading and unloading, and continuous temperature deviations caused by abnormal factors such as refrigeration system failure, damaged door seals, and heat leakage from the container. This results in a high false alarm rate. At the same time, when a temperature anomaly occurs, existing technologies can only indicate the abnormal temperature zone and cannot quickly locate the root cause of the anomaly. This makes it impossible for maintenance personnel to take targeted solutions, resulting in long anomaly response times and making it easy for high-value goods to spoil and be lost.
[0004] Secondly, the overall control capability is insufficient, and the cold energy distribution is unreasonable. Existing multi-temperature zone temperature control systems generally adopt an independent control mode for each temperature zone, without fully considering the heat conduction effect between adjacent temperature zones and the optimization of the overall cold energy distribution of the refrigeration system. This decentralized control method easily leads to insufficient cold energy and failure to reach the required temperature in some temperature zones, while some temperature zones are over-cooled, resulting in energy waste and freezing damage to goods. Moreover, when a temperature zone is determined to be abnormal, existing technologies usually perform unified overall control of the entire temperature zone, and cannot provide targeted and precise cooling to high-risk areas with severe local temperature deviations within the temperature zone. This results in the most severely abnormal temperature areas not being able to quickly recover to the normal temperature range, while relatively normal temperature areas are over-cooled, which not only prolongs the abnormality recovery time, but also exacerbates energy consumption and the risk of cargo damage.
[0005] In addition, existing technologies, when constructing temperature control models, either focus only on the temperature change trend in the time dimension or only consider the heat conduction relationship in the spatial dimension, failing to organically integrate the multi-factor causal relationship in the time dimension with the heat conduction relationship in the spatial dimension. This makes it impossible for existing models to comprehensively and accurately depict the complex interaction relationship between environmental factors, physical components, travel events and the temperature of each temperature zone in a multi-temperature zone system, resulting in a lack of scientific basis for temperature control decisions and unsatisfactory control effects.
[0006] In view of this, the present invention proposes a multi-temperature zone adaptive dynamic temperature control method and system for refrigerated semi-trailers to solve the above problems. Summary of the Invention
[0007] To overcome the aforementioned deficiencies of the prior art and achieve the above objectives, the present invention provides the following technical solution: A multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers, the method includes: The interior of the refrigerated semi-trailer was divided into multiple temperature zones, and external influence data and internal status data of each temperature zone were collected. The external influence data and internal state data of each temperature zone are analyzed to construct a temperature spatiotemporal causal map of the temperature zone. Based on the temperature spatiotemporal causal map of the temperature zone, the anomaly score of each temperature zone is obtained, and the anomaly score is used to determine whether there is a temperature anomaly in each temperature zone. The temperature regions with abnormal temperatures are identified, and a heat flow exchange matrix is constructed. The temperature of the temperature regions with abnormal temperatures is then regulated based on the heat flow exchange matrix. The temperature zone with temperature anomalies is further divided into multiple sub-temperature zones. Temperature monitoring data and heat flow exchange data of each sub-temperature zone are obtained. The priority coefficient of each sub-temperature zone is calculated based on the temperature monitoring data and heat flow exchange data. The sub-temperature zones are then regulated based on the priority coefficient.
[0008] Furthermore, the method for constructing the spatiotemporal causal map of temperature range is as follows: Four types of nodes are predefined: temperature zone grid nodes, physical component nodes, environmental factor nodes, and travel event nodes. Each node has its own attributes. All nodes are paired to form multiple node pairs. Based on preset physical prior knowledge, multiple candidate node pairs are selected from the multiple node pairs. Correlation analysis is performed on each candidate node pair to select candidate node pairs that meet preset conditions. These are denoted as causal association pairs, and corresponding causal edges are constructed based on the causal association pairs. The causal influence strength of each causal edge is calculated and denoted as the corresponding causal edge weight. The edge connecting adjacent temperature zone grid nodes is denoted as a spatial edge. The thermal conductivity coefficients of the two temperature zone grid nodes corresponding to the spatial edge are calculated and denoted as the corresponding spatial edge weights. By combining the temperature zone grid nodes, physical component nodes, environmental factor nodes, travel event nodes, as well as the causal edge weights and spatial edge weights, a temperature spatiotemporal causal graph of the temperature zone is obtained.
[0009] Furthermore, the calculation methods for the intensity of causal influence and the thermal conductivity coefficient are as follows: Within a preset delay time range, the correlation degree of causal pairs under different delay times is calculated, and the maximum correlation degree within the preset delay time range is taken as the optimal correlation degree; the correlation type of the causal pairs is obtained, and the correlation coefficient is determined; the partial correlation coefficient of the causal pairs is calculated, and the absolute value of the partial correlation coefficient is taken to obtain the direct causal proportion coefficient; historical data of causal pairs is obtained and divided into N equal-length time windows, and the optimal correlation degree in each time window is calculated. The average correlation degree is calculated by averaging the optimal correlation degrees of all time windows; the maximum optimal correlation degree in all time windows is recorded as the maximum correlation degree, and the time stability coefficient is divided by the average correlation degree to obtain the time stability coefficient; the causal influence strength is obtained by multiplying the optimal correlation degree, correlation coefficient, direct causal proportion coefficient, and time stability coefficient together and then normalizing the result. In the spatial boundary, obtain the thermal conductivity between adjacent temperature zone grid nodes and calculate the interface equivalent thermal conductivity between the two nodes; obtain the contact area between adjacent temperature zone grid nodes and the distance between the center points of adjacent temperature zone grid nodes; divide the contact area by the distance and multiply by the interface equivalent thermal conductivity to obtain the thermal conductivity value, and normalize the thermal conductivity value to obtain the thermal conductivity coefficient.
[0010] Furthermore, the method for determining whether there are temperature anomalies in each temperature zone is as follows: Based on the constructed temperature spatiotemporal causal graph of the temperature zone, the counterfactual theoretical temperature of each temperature zone grid node is calculated; the real-time temperature of each temperature zone grid node is obtained, and the absolute value of the difference between the counterfactual theoretical temperature and the actual temperature is calculated and recorded as the deviation value. The duration and spread range of the deviation are obtained. The deviation value, duration and spread range are weighted and accumulated to obtain the anomaly score of each temperature zone grid node. The anomaly score is compared with the preset anomaly score threshold. If the anomaly score exceeds the anomaly score threshold, it is determined that there is a temperature anomaly in the temperature zone corresponding to the temperature zone grid node.
[0011] Furthermore, the method for calculating the counterfactual theory temperature is as follows: The temperature zone grid node to be calculated is denoted as the target temperature zone grid node. All causal relationships containing the target temperature zone grid node are extracted from the temperature zone spatiotemporal causal graph. All nodes upstream of the target temperature zone grid node are extracted from all causal relationships and denoted as upstream nodes. For each causal relationship pair and its corresponding causal edge, a time-series lag regression method is used to traverse multiple lag time windows and calculate the correlation degree between the upstream node and the target temperature zone grid node under different lag time windows. The lag time corresponding to the maximum correlation degree is taken as the causal influence delay time. Obtain the historical state data of each upstream node corresponding to the causal impact delay time back to the current moment, multiply the historical state data by the corresponding causal edge weight to obtain the unit causal impact value of the corresponding upstream node, and accumulate the unit causal impact values of all upstream nodes to obtain the comprehensive causal impact value. Construct an autoregressive inertial model of the grid nodes in the target temperature zone and calculate the temperature inertial reference value of the grid nodes in the target temperature zone; obtain the actual temperature and thermal conductivity coefficient of all grid nodes in the temperature zone adjacent to the grid nodes in the target temperature zone and calculate the spatial thermal conduction influence value of the grid nodes in the target temperature zone; weight and superimpose the temperature inertial reference value, spatial thermal conduction influence value, and causal comprehensive influence value to obtain the counterfactual theoretical temperature.
[0012] Furthermore, the method for temperature control in temperature zones with anomalies based on the heat flow exchange matrix is as follows: The temperature zone with the identified temperature anomaly is taken as the target control vertex. The associated vertices that can establish a heat flow channel with the target control vertex are identified. The associated vertices include the refrigeration system vertex, the phase change energy storage vertex, the external environment vertex, and the adjacent temperature zone vertex. The heat flow system between the target control vertex and each associated vertex is abstracted as a directed weighted temperature difference potential energy graph G=(V,E,W), where V contains the target control vertex and its associated vertices, E represents the heat flow channels that can be established between the target control vertex and each associated vertex, and W represents the weight value of each heat flow channel, defined as the temperature difference potential energy from one vertex to another. Real-time acquisition of target control vertex and associated data of each associated vertex; updating of the directed weighted temperature difference potential energy map G based on the associated data. Construct a target optimization function and combine it with the preset temperature recovery constraint of the target control vertex and the maximum heat flow constraint of the heat flow channel to solve for the optimal heat flow of each heat flow channel in the directed weighted temperature difference potential energy map G. Generate a heat flow exchange matrix based on the optimal heat flow and convert the heat flow exchange matrix into control commands to execute temperature control.
[0013] Furthermore, the method for determining the optimal heat flow rate for each heat flow channel is as follows: Based on the directed weighted temperature difference potential energy graph G, a residual network is constructed. In the residual network, the edges along the original heat flow channel direction are defined as forward edges, and the edges opposite to the original heat flow channel direction are defined as reverse edges. The actual heat flow of each forward edge is initialized to zero, the residual capacity of the forward edge is the maximum heat flow preset for the corresponding heat channel, and the residual capacity of the reverse edge is zero. The temperature recovery constraint of the target control vertex is converted into net heat demand. Define power supply nodes and power receiving nodes. Search for augmenting paths from power supply nodes to power receiving nodes in the residual network. Calculate the total path potential of each augmenting path and select the augmenting path with the largest total path potential as the optimal augmenting path for augmentation. Calculate the residual capacity of all forward edges on the optimal augmenting path, take the minimum residual capacity of the forward edge as the bottleneck capacity, obtain the remaining net heat demand of the target control vertex, correct the bottleneck capacity, and obtain the corrected flow rate. Push the corrected flow along the optimal augmenting path, update the actual heat flow and residual capacity of the forward edge, and update the residual capacity of the reverse edge simultaneously; perform iterative updates until the remaining net heat demand of the target control vertex in the residual network is zero, and map the actual heat flow accumulated by each forward edge in the residual network to the corresponding heat flow channel in the directed weighted temperature difference potential energy graph G to obtain the optimal heat flow of each heat flow channel.
[0014] Furthermore, the method for regulating each sub-temperature zone based on priority coefficients is as follows: The temperature zone is divided into multiple sub-temperature zones, and time-series temperature monitoring data in each sub-temperature zone is obtained. Based on the time-series temperature monitoring data of all sub-temperature zones, the history offset and state offset of each sub-temperature zone are calculated. Based on the history offset and state offset, the population deviation of each sub-temperature zone is calculated. For each sub-temperature zone, construct a set of directly related sub-temperature zones and a set of spatially adjacent sub-temperature zones. Obtain temperature monitoring data and heat flow exchange data for each sub-temperature zone within the set of directly related sub-temperature zones and the set of spatially adjacent sub-temperature zones, respectively. Calculate the temperature anomaly and heat flow anomaly for each sub-temperature zone based on the temperature monitoring data and heat flow exchange data. Calculate the neighborhood coupling degree for each sub-temperature zone based on the temperature anomaly and heat flow anomaly. The group deviation degree and the neighborhood coupling degree are weighted and accumulated to obtain the priority coefficient of each sub-temperature zone. The sub-temperature zone with the highest priority coefficient is selected for priority control.
[0015] Furthermore, the methods for calculating the group deviation degree and the neighborhood coupling degree are as follows: Acquire time-series temperature monitoring data for each sub-temperature zone under a preset historical time series, and fit a time-series temperature characteristic function for each sub-temperature zone; calculate the temperature integral value of the time-series temperature characteristic function under the historical time series, and calculate the average temperature integral value by averaging all temperature integral values; use the absolute value of the difference between the temperature integral value of each sub-temperature zone and the average temperature integral value as the history offset; acquire the temperature monitoring data of each sub-temperature zone at the current moment, and calculate the average instantaneous temperature by averaging all temperature monitoring data; use the absolute value of the difference between the temperature monitoring data of each sub-temperature zone and the instantaneous temperature average value as the state offset; weightedly sum the history offset and the state offset to obtain the population deviation. The absolute values of the differences between the temperature monitoring data and heat flow exchange data of each sub-temperature zone in the directly associated sub-temperature zone set and the spatially adjacent sub-temperature zone set and their respective corresponding temperature standard values and heat flow exchange ideal values are calculated to obtain the direct temperature disturbance degree, direct heat flow disturbance degree, adjacent temperature disturbance degree, and adjacent heat flow disturbance degree of each sub-temperature zone. The temperature anomaly is obtained based on the direct temperature disturbance degree and the adjacent temperature disturbance degree, and the heat flow anomaly is obtained based on the direct heat flow disturbance degree and the adjacent heat flow disturbance degree. The temperature anomaly and the heat flow anomaly are weighted and accumulated to obtain the neighborhood coupling degree.
[0016] The refrigerated semi-trailer features a multi-zone adaptive dynamic temperature control system, which includes: The data acquisition module is used to divide the interior of the refrigerated semi-trailer into multiple temperature zones and collect external influence data and internal status data for each temperature zone. The anomaly detection module is used to analyze the external influence data and internal state data of each temperature zone, construct a temperature spatiotemporal causal map of the temperature zone, obtain anomaly scores for each temperature zone based on the temperature spatiotemporal causal map, and determine whether there are temperature anomalies in each temperature zone based on the anomaly scores. The temperature control module is used to acquire temperature zones with abnormal temperatures and construct a heat flow exchange matrix. Based on the heat flow exchange matrix, the temperature of the temperature zones with abnormal temperatures is controlled. The control implementation module is used to further construct multiple sub-temperature zones from the temperature zone with temperature anomalies, acquire temperature monitoring data and heat flow exchange data of each sub-temperature zone, calculate the priority coefficient of each sub-temperature zone based on the temperature monitoring data and heat flow exchange data, and control each sub-temperature zone based on the priority coefficient.
[0017] The technical effects and advantages of the multi-temperature zone adaptive dynamic temperature control method and system for refrigerated semi-trailers of this invention are as follows: This invention constructs a spatiotemporal causal map of temperature zones that integrates temporal causality and spatial heat conduction. Using counterfactual theoretical temperature after eliminating abnormal interference as the judgment benchmark, it replaces the traditional fixed temperature threshold, accurately distinguishing between normal temperature fluctuations and systematic anomalies, fundamentally solving the problem of high false alarm rates in traditional methods. Simultaneously, by integrating three dimensions of features—temperature deviation value, deviation duration, and spatial diffusion range—a comprehensive anomaly score is achieved, enabling quantitative assessment and three-level classification of the severity of temperature anomalies. Relying on the causal reasoning capabilities of the spatiotemporal causal map, the root cause of anomalies can be quickly located and its impact quantified while anomalies are identified, thereby shortening anomaly response time and effectively reducing the risk of spoilage and loss of high-value cold chain goods.
[0018] This invention constructs a directed weighted temperature difference potential energy diagram, unifying the heat flow system between the target control vertex and the refrigeration system vertex, phase change energy storage vertex, external environment vertex, and adjacent temperature zone vertex. A residual network algorithm is used to solve for the optimal heat flow of each heat flow channel, generating a heat flow exchange matrix and converting it into control commands for execution, thus achieving optimized allocation of global cooling capacity in the refrigeration system. Simultaneously, for temperature zones determined to be abnormal, they are further subdivided into multiple independently monitorable and controllable sub-temperature zones. By calculating the priority coefficient of each sub-temperature zone, priority is given to ensuring the fastest recovery of the sub-temperature zone with the most severe temperature anomaly and the highest risk of diffusion to the normal temperature range. This targeted and precise control method not only significantly shortens the anomaly recovery time but also avoids over-cooling of normal temperature areas, effectively preventing cargo freezing damage and achieving the best balance between control effectiveness and economy. Attached Figure Description
[0019] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system module block diagram of the present invention. Detailed Implementation
[0020] 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.
[0021] In one embodiment, a multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers is disclosed, such as... Figure 1 As shown, the method includes: The interior of the refrigerated semi-trailer was divided into multiple temperature zones, and external influence data and internal status data for each temperature zone were collected.
[0022] Specifically, the interior of the refrigerated semi-trailer's cargo compartment is divided into multiple independent temperature zones according to physical partitions and functional requirements. External influence data and internal status data for each temperature zone are collected. External influence data includes environmental data (such as ambient temperature, solar radiation intensity, and outside wind speed, which can be collected using sensors such as temperature sensors, radiation sensors, and wind speed sensors installed outside the refrigerated semi-trailer), travel data (such as vehicle speed, altitude, door opening / closing status and duration, and vehicle start / stop frequency and duration; vehicle speed and altitude can be obtained through the vehicle's built-in GPS system; door opening / closing status and duration are collected by door magnetic switch sensors installed at corresponding positions on each cargo door frame and door panel, recording the timestamp of the opening / closing action and calculating the duration using the time difference; vehicle start / stop frequency and duration are obtained by reading the vehicle's engine ECU operating status data), and event data (such as the start time, end time, and loading / unloading volume of loading / unloading events, refrigeration system start / stop events, and phase change energy storage unit charging / discharging events). The refrigeration system start-up and shutdown events, and phase change energy storage unit charging and discharging events are recorded manually through various communication interfaces. Internal status data includes temperature zone status data (such as real-time temperature and set temperature for each zone; real-time temperature is obtained from temperature sensors installed at corresponding internal locations, and set temperatures are preset manually), refrigeration system data (such as compressor power, evaporator inlet and outlet temperatures, condenser inlet and outlet temperatures, and fan speed; these can all be obtained through the refrigeration system's built-in monitoring data), phase change energy storage data (such as the phase change energy storage unit's charging and discharging status, remaining capacity, phase change temperature, and inlet and outlet temperatures; these are obtained through the phase change energy storage unit's built-in monitoring data), and heat exchange component data (such as damper opening, heat exchange valve opening, and heat exchange efficiency; damper and heat exchange valve openings are directly read by the position feedback sensors built into the damper and heat exchange valve actuators; heat exchange efficiency is determined by the corresponding design parameters and heat exchanger calibration performance). It should be noted that the above data is only illustrative and does not include all of these data; the specific data obtained will be selected based on the actual situation. Then, a unified timeline based on vehicle GPS time is established. The collected external influence data and internal state data are standardized and mapped to the corresponding temperature zones. Specifically, a unified timeline based on the UTC time output by the vehicle GPS module is established, the system sampling frequency is set to 1Hz, and all collected data are aligned to the integer second timestamp of this timeline. The collected external influence data and internal state data are standardized and preprocessed before being mapped to the corresponding temperature zones. For example, the Z-score standardization method is used to map data of different dimensions to a standard normal distribution with a mean of 0 and a variance of 1. For missing data, if the missing duration is less than or equal to 5 seconds, linear interpolation is used to fill the missing data. If the missing duration is greater than 5 seconds, the historical average value of the data type in the same time period is used to fill the missing data to facilitate subsequent analysis and calculation.
[0023] The external influence data and internal state data of each temperature zone are analyzed to construct a temperature spatiotemporal causal map of the temperature zone. Based on the temperature spatiotemporal causal map of the temperature zone, the anomaly score of each temperature zone is obtained, and the anomaly score is used to determine whether there is a temperature anomaly in each temperature zone. The method for constructing the temperature spatiotemporal causal graph of the temperature zone is as follows: four types of nodes are predefined: temperature zone grid nodes, physical component nodes, environmental factor nodes, and travel event nodes. Each node has its own attributes. All nodes are paired to form multiple node pairs. Based on preset physical prior knowledge, multiple candidate node pairs are selected from the multiple node pairs. Correlation analysis is performed on each candidate node pair to select candidate node pairs that meet preset conditions. These are denoted as causal association pairs, and corresponding causal edges are constructed based on the causal association pairs. The causal influence strength of each causal edge is calculated and denoted as the corresponding causal edge weight. The edge connecting adjacent temperature zone grid nodes is denoted as a spatial edge. The thermal conductivity coefficients of the two temperature zone grid nodes corresponding to the spatial edge are calculated and denoted as the corresponding spatial edge weights. By combining the temperature zone grid nodes, physical component nodes, environmental factor nodes, travel event nodes, as well as the causal edge weights and spatial edge weights, a temperature spatiotemporal causal graph of the temperature zone is obtained. The calculation methods for causal influence intensity and heat conduction coefficient are as follows: Within a preset delay time range, calculate the correlation degree of causal pairs under different delay times, and take the maximum correlation degree within the preset delay time range as the optimal correlation degree; obtain the correlation type of the causal pairs and determine the correlation coefficient; calculate the partial correlation coefficient of the causal pairs, and take the absolute value of the partial correlation coefficient to obtain the direct causal proportion coefficient; obtain the historical data corresponding to the causal pairs, divide them into N equal-length time windows, calculate the optimal correlation degree in each time window, and calculate the average correlation degree by averaging the optimal correlation degrees of all time windows; take the maximum optimal correlation degree in all time windows as the maximum correlation degree, and divide the average correlation degree by the maximum correlation degree to obtain the time stability coefficient; multiply the optimal correlation degree, correlation coefficient, direct causal proportion coefficient, and time stability coefficient together and then normalize to obtain the causal influence intensity. In the spatial boundary, obtain the thermal conductivity between adjacent temperature zone grid nodes and calculate the interface equivalent thermal conductivity between the two nodes; obtain the contact area between adjacent temperature zone grid nodes and the distance between the center points of adjacent temperature zone grid nodes; divide the contact area by the distance and multiply by the interface equivalent thermal conductivity to obtain the thermal conductivity value, and normalize the thermal conductivity value to obtain the thermal conductivity coefficient.
[0024] Four types of nodes are predefined: temperature zone grid nodes, physical component nodes, environmental factor nodes, and travel event nodes. Each node has its own attributes. Temperature zone grid nodes are defined as follows: each temperature zone corresponds to one temperature zone grid node, and its attributes include: temperature zone number, spatial coordinates, set temperature, cargo type, cargo weight, and real-time temperature. Physical component nodes include refrigeration system nodes, phase change energy storage unit nodes, fan nodes, damper nodes, and heat exchange valve nodes, and their attributes include: component number, temperature zone, operating status, real-time power, inlet and outlet temperatures, opening degree, and heat exchange efficiency. Environmental factor nodes include ambient temperature nodes. Nodes include solar radiation nodes, wind speed nodes, and altitude nodes, whose attributes include real-time measurements and predicted values for the next 5-15 minutes. Travel event nodes include door opening / closing nodes, loading / unloading nodes, vehicle start / stop nodes, and fault event nodes, whose attributes include event type, occurrence time, duration, and event intensity (e.g., loading / unloading volume, door opening degree). All nodes are paired to form multiple node pairs. Based on pre-defined physical prior knowledge, multiple candidate node pairs are selected from these pairs. This physical prior knowledge can be combined with the mechanical structure design of refrigerated semi-trailers, the working principle of the refrigeration system, and the actual operating rules of cold chain transportation. Pre-determined objective causal relationships and physical constraints that can be verified without statistical data analysis are used. For example, heat conduction is guaranteed between temperature zone grid nodes and all adjacent temperature zone grid nodes, which are considered candidate node pairs. Non-adjacent temperature zone grid nodes that do not have direct heat conduction are directly eliminated. Correlation analysis is performed on each candidate node pair to select those that meet the preset conditions, which are then denoted as causal association pairs. For node pairs consisting of two continuous variables (such as temperature zone temperature and ambient temperature), the maximum information coefficient (MIC) is calculated, and a significance test is performed to obtain the corresponding significance (p-value). For node pairs containing discrete variables... For node pairs of event variables (such as temperature zone and door opening / closing event), mutual information (MI) is calculated, and a significance test is performed to obtain the corresponding significance. The preset conditions are: the maximum information coefficient is greater than the preset information coefficient threshold (set empirically, such as 0.6) and the corresponding significance is less than the preset significance threshold (set empirically, such as 0.01), or the mutual information is greater than the preset mutual information threshold (set empirically, such as 0.3) and the corresponding significance is less than the preset significance threshold (set empirically, such as 0.01). The calculation of maximum information coefficient, mutual information, and significance are all existing techniques in the field of statistics and will not be described in detail here. The obtained causal relationships are used to construct corresponding causal edges, and the causal influence strength of each causal edge is calculated and recorded as the corresponding causal edge weight. Specifically, within a preset delay time range, the preset delay time range can be determined according to the time characteristics of heat conduction in refrigerated semi-trailers, such as taking 0-60 seconds as the preset delay time range. The correlation degree of the causal relationship pairs under different delay times is calculated. Here, the correlation degree is the obtained maximum information coefficient or mutual information. The maximum correlation degree within the delay time range is taken as the optimal correlation degree. The optimal correlation degree is used to measure the upper limit of the causal influence strength of the cause node on the result node, reflecting the strongest nonlinear correlation between the two variables. The larger the value, the stronger the causal influence of the cause node on the result node, and the more the change of the result node can be explained by the change of the cause node. The association type of causal pairs is determined, and the association coefficient is determined. The determination of the association type and coefficient employs existing correlation analysis methods. For continuous variable pairs, the Pearson correlation coefficient or Spearman rank correlation coefficient is calculated to determine the corresponding association coefficient. For discrete event-continuous variable pairs, the sign of the change in the mean of the outcome variable before and after the event is calculated. If the sign is positive, the association coefficient is determined to be 1 (positive correlation); if the sign is negative, the association coefficient is determined to be −1 (negative correlation). This is used to distinguish between positive and negative causal relationships. Absolute values are used in subsequent calculations of causal influence strength because the association coefficient only indicates direction and does not affect the magnitude of the influence. The partial correlation coefficients of causal pairs are calculated, and the absolute value of the partial correlation coefficients is taken to obtain the direct causal proportion coefficient. In this embodiment, in the construction of the temperature spatiotemporal causal map of the temperature region, for any pair of causal pairs (cause node X, result node Y), all upstream nodes pointing to Y except X are used as control variables. The partial correlation coefficients between X and Y are calculated, and the absolute value is taken to obtain the direct causal proportion coefficient. The calculation of the partial correlation coefficient adopts the recursive calculation method based on the covariance matrix, which is well known in the art. Its core principle is to calculate the pure correlation coefficient between the cause node and the result node after controlling all other upstream variables that may affect the result node. This is used to eliminate the interference of indirect causal relationships and accurately measure the direct causal strength between the two nodes. This calculation method has been widely used in the field of statistics, and its specific calculation process will not be elaborated here. The larger the value, the more direct the causal relationship between the two nodes, and the smaller the indirect influence produced by other intermediate nodes. Acquire historical data corresponding to causal relationships. Historical data refers to the time-series data collection of the cause node and result node constituting the causal relationship during a continuous period of normal operation, aligned to a unified GPS timeline. The historical data varies depending on the node type of the causal relationship. For example, for a causal relationship (ambient temperature → temperature zone), the historical data is time-series temperature data. For a causal relationship (door opening / closing → temperature zone), the historical data is the historical event time-series data of the cause node (recording the start / end timestamps of the event) + the historical time-series numerical data of the result node before and after the event. Divide the historical data into... Divide the time into N equal-length time windows (e.g., N is 10, and each window is 10 minutes long). Calculate the optimal correlation degree within each time window. Calculate the average correlation degree for all time windows. Take the maximum optimal correlation degree among all time windows as the maximum correlation degree. Divide the average correlation degree by the maximum correlation degree to obtain the time stability coefficient. The time stability coefficient measures the stability and reliability of the causal relationship over time, reflecting whether the causal relationship remains consistent under different operating conditions. The larger the value, the more stable the causal relationship is over time, the less affected by changes in operating conditions, and the more universally applicable it is. The causal influence strength is obtained by multiplying the optimal correlation degree, correlation coefficient, direct causal proportion coefficient and time stability coefficient and then normalizing the result. The greater the causal influence strength, the stronger the causal effect between the two nodes. Therefore, the causal influence strength is denoted as the corresponding causal edge weight.
[0025] The edges connecting adjacent temperature zone grid nodes are denoted as spatial edges. The thermal conductivity coefficients of the two temperature zone grid nodes corresponding to the spatial edge are calculated and denoted as the corresponding spatial edge weights. Specifically: the thermal conductivity coefficients between adjacent temperature zone grid nodes are obtained, and the equivalent thermal conductivity of the interface between the two nodes is calculated based on the principle of thermal resistance series connection. Since the thermal resistance between adjacent temperature zone grid nodes is composed of the thermal resistance of the partition material and the interface contact thermal resistance in series, the equivalent thermal conductivity of the interface is obtained by adding the thermal resistance of the partition material and the interface contact thermal resistance and taking the reciprocal. The thermal resistance of the partition material is obtained by dividing the partition thickness by the inherent thermal conductivity of the material, and the interface contact thermal resistance is obtained by direct measurement. The contact area between adjacent temperature zone grid nodes and the distance between the center points of adjacent temperature zone grid nodes are obtained. The contact area is divided by the distance and then multiplied by the equivalent thermal conductivity of the interface to obtain the thermal conductivity value. The thermal conductivity value is normalized to obtain the thermal conductivity coefficient, which is denoted as the corresponding spatial edge weight. The larger the value of the thermal conductivity coefficient, the stronger the heat exchange capacity between the two adjacent temperature zones and the more significant the temperature interaction. Finally, by combining the temperature zone grid nodes, physical component nodes, environmental factor nodes, travel event nodes, as well as the causal edge weights and spatial edge weights, a spatiotemporal causal graph of temperature zones is obtained. In this way, the causal relationship in the time dimension and the heat conduction relationship in the spatial dimension are integrated, which can accurately depict the mutual influence relationship between various factors in the multi-temperature zone system of refrigerated semi-trailers. Based on this graph, the root cause of temperature anomalies can be accurately identified, providing accurate basic data for subsequent global heat flow optimization and control, realizing early warning and precise control of temperature anomalies, and reducing the risk of cargo loss and system energy consumption.
[0026] The method for determining whether there are temperature anomalies in each temperature zone is as follows: based on the constructed temperature spatiotemporal causal map of the temperature zone, calculate the counterfactual theoretical temperature of each temperature zone grid node, obtain the real-time temperature of each temperature zone grid node, and calculate the absolute value of the difference between the counterfactual theoretical temperature and the actual temperature, which is recorded as the deviation value. The duration and spread range of the deviation are obtained. The absolute value of the difference, the duration of the deviation, and the spread range are weighted and accumulated to obtain the anomaly score of each temperature zone grid node. The anomaly score is compared with the preset anomaly score threshold. If the anomaly score exceeds the anomaly score threshold, it is determined that there is a temperature anomaly in the temperature zone corresponding to the temperature zone grid node. The method for calculating the counterfactual theory temperature is as follows: the grid node of the temperature region to be calculated is denoted as the target temperature region grid node. All causal association pairs containing the target temperature region grid node are extracted from the spatiotemporal causal graph of the temperature region. All nodes upstream of the target temperature region grid node are extracted from all causal association pairs and denoted as upstream nodes. For each causal relationship pair and its corresponding causal edge, a time-series lag regression method is used to traverse multiple lag time windows and calculate the correlation degree between the upstream node and the target temperature zone grid node under different lag time windows. The lag time corresponding to the maximum correlation degree is taken as the causal influence delay time. Obtain the historical state data of each upstream node corresponding to the causal impact delay time back to the current moment, multiply the historical state data by the corresponding causal edge weight to obtain the unit causal impact value of the corresponding upstream node, and accumulate the unit causal impact values of all upstream nodes to obtain the comprehensive causal impact value. Construct an autoregressive inertial model of the grid nodes in the target temperature zone and calculate the temperature inertial reference value of the grid nodes in the target temperature zone; obtain the actual temperature and thermal conductivity coefficient of all grid nodes in the temperature zone adjacent to the grid nodes in the target temperature zone and calculate the spatial thermal conduction influence value of the grid nodes in the target temperature zone; weight and superimpose the temperature inertial reference value, spatial thermal conduction influence value, and causal comprehensive influence value to obtain the counterfactual theoretical temperature.
[0027] Specifically, the counterfactual theoretical temperature of each temperature zone grid node is first calculated. The temperature zone grid node to be calculated is denoted as the target temperature zone grid node. All causal association pairs containing the target temperature zone grid node are extracted from the temperature spatiotemporal causal graph of the temperature zone. All nodes upstream of the target temperature zone grid node are extracted from all causal association pairs and denoted as upstream nodes. Here, all nodes upstream of the target temperature zone grid node refer to the direction of the causal edge. All corresponding nodes pointing to the target temperature zone grid node are filtered out. It should be noted that the upstream nodes here refer to the upstream nodes after key anomaly elimination. Key anomaly elimination is: traversing all upstream nodes and their corresponding causal edges. If an upstream node is marked as an abnormal state, or a causal edge is marked as an abnormal causal edge (such as the heat leakage edge corresponding to the broken door seal), then the node and its corresponding causal edge are excluded from the calculation to ensure that the counterfactual reasoning is based only on the causal relationship of the normal state. For each retained normal causal edge, a time-series lag regression method is used to calculate its causal impact delay time. First, multiple lag time windows are preset (the length is determined empirically). Each lag time window contains multiple lag times. The correlation degree of each lag time within each lag time window is obtained, and the lag time corresponding to the highest correlation degree is taken as the causal impact delay time of the corresponding causal edge. This method can accurately quantify the time difference of upstream factor changes transmitting to the target temperature zone, providing a basis for subsequent data time alignment. Then, the historical state data corresponding to the causal impact delay time of each upstream node at the current moment is obtained. Here, the data type of the historical state data is determined according to the type of the upstream node. If the upstream node is a refrigeration system, the historical state data is the compressor's output power. If the upstream node is a phase change energy storage unit, the historical state data is the cooling power. If the upstream node is an adjacent temperature zone, the historical state data is the temperature. Multiplying this historical state data by the corresponding causal edge weights yields the unit causal impact value of the corresponding upstream node. The unit causal impact values of all upstream nodes are then weighted and summed, with the weights being the causal edge weights of the corresponding upstream nodes, to obtain the comprehensive causal impact value. The larger the unit causal impact value, the more significant the influence of the upstream node on the target temperature zone. Therefore, the larger the comprehensive causal impact value, the stronger the combined effect of all upstream nodes on the target temperature zone. Since the temperature of a temperature zone has significant thermal inertia, the current temperature largely depends on the temperature of the previous few moments. Therefore, an autoregressive inertial model of the grid nodes of the target temperature zone is constructed. Here, a third-order autoregressive inertial model can be preferred. For example, the actual temperature of the grid nodes of the target temperature zone in the previous 1 second, 2 seconds, and 3 seconds can be obtained. The grid nodes of the target temperature zone are trained offline using normal temperature data of the grid nodes in the past 24 hours to obtain the autoregressive coefficients of the previous 1 second, 2 seconds, and 3 seconds. The weighted summation is then used to obtain the temperature inertial reference value of the grid nodes of the target temperature zone. The temperature inertial reference value quantifies the thermal inertia of each grid node of the target temperature zone, that is, the natural trend of temperature change, reflecting the temperature trend of the temperature zone when there are no external factors. Within the corresponding spatial boundary, the actual temperature and thermal conductivity coefficient of adjacent temperature zone grid nodes are obtained. The spatial thermal conduction influence value of the target temperature zone grid node is calculated. The calculation method for the spatial thermal conduction influence value is as follows: the absolute value of the difference between the actual temperature of each adjacent temperature zone grid node and the actual temperature of the target temperature zone grid node is calculated, and then multiplied by the corresponding thermal conductivity coefficient to obtain the spatial conduction value of a single adjacent temperature zone grid node; the spatial conduction values of all adjacent temperature zone grid nodes that are adjacent to the target temperature zone grid node are arithmetically summed to obtain the spatial thermal conduction influence value; the spatial thermal conduction influence value is used to quantify the influence of adjacent temperature zones on the target temperature zone temperature through thermal conduction. The larger the value, the greater the temperature difference between the adjacent temperature zone and the target temperature zone, or the higher the thermal conduction efficiency, and the more obvious the influence on the target temperature zone temperature. Finally, the temperature inertia benchmark value, spatial heat conduction influence value, and causal comprehensive influence value are weighted and superimposed to obtain the counterfactual theoretical temperature. The weight coefficients of the temperature inertia benchmark value, spatial heat conduction influence value, and causal comprehensive influence value can be obtained by collecting a large amount of historical normal operation data of refrigerated semi-trailers and using multiple linear regression or least squares method for offline training and optimization, or can be formulated based on the experience of people in this field, which will not be described in detail here. Among them, the counterfactual theoretical temperature is the theoretical temperature that the target temperature zone should reach under the current normal factors assuming that there are no abnormal factors (such as broken door seals, refrigeration failure, heat leakage, etc.). It can be used as the benchmark for subsequent temperature anomaly judgment. After obtaining the counterfactual theoretical temperature for each temperature zone grid node, the real-time temperature of each temperature zone grid node is acquired. The absolute value of the difference between the counterfactual theoretical temperature and the actual temperature is calculated and recorded as the deviation value. The deviation value reflects the degree to which the current temperature deviates from the normal state. The duration and diffusion range of the deviation are obtained. The deviation duration is the continuous duration for which the deviation value exceeds a preset deviation threshold based on experience. It reflects the persistence of the temperature anomaly in the time dimension and can effectively filter out short-term temperature fluctuations caused by opening and closing doors, loading and unloading, etc., reducing the false alarm rate. The diffusion range is: taking the target temperature zone grid node as the seed point, traversing all target temperature zone grid nodes connected to it through spatial edges, and recording them as connected temperature zone grid nodes, the number of connected temperature zone grid nodes that also have a deviation value greater than the deviation threshold is counted. The number of temperature zone grid nodes reflects the extent of temperature anomaly spread in the spatial dimension, distinguishing between local single-point anomalies and systemic global anomalies. Anomaly scores for each temperature zone grid node are obtained by weighted summation of the absolute value of the difference, the duration of the deviation, and the diffusion range. Here, the absolute value of the difference, the duration of the deviation, and the diffusion range are normalized values, and their weights can be determined based on the experience of those in the field. A higher anomaly score indicates a more severe temperature anomaly at the corresponding temperature zone grid node. Therefore, an anomaly score threshold is set based on experience, and the anomaly score is compared with this threshold. If the anomaly score exceeds the threshold, it is determined that the temperature zone corresponding to the temperature zone grid node has a temperature anomaly. This solution utilizes a temperature anomaly identification method based on a spatiotemporal causal graph of temperature zones. It uses the counterfactual theoretical temperature after eliminating anomaly interference as the judgment benchmark, replacing the single set temperature threshold in traditional methods. This method can accurately distinguish between normal temperature fluctuations caused by vehicle start-stop, door opening and closing, and loading / unloading, and systemic anomalies caused by door seal damage or refrigeration failure. This fundamentally solves the problems of high false alarm rates and severe underreporting in traditional temperature control methods. Simultaneously, by integrating three dimensions of features—absolute value of temperature deviation, duration of deviation, and spatial diffusion range—a weighted anomaly score is applied, enabling a quantitative assessment of the severity of temperature anomalies. This allows for early identification and warning of potential minor anomalies. Furthermore, leveraging the causal reasoning capabilities of the spatiotemporal causal graph, the root cause of anomalies can be quickly located and its impact quantified while anomalies are identified. This provides precise anomaly information input for subsequent global heat flow optimization and control, effectively shortening anomaly response time, significantly reducing the risk of spoilage and loss of high-value cold chain goods, and improving the reliability of multi-temperature zone temperature control systems for refrigerated semi-trailers.
[0028] The temperature regions with abnormal temperatures are identified, and a heat flow exchange matrix is constructed. The temperature of the temperature regions with abnormal temperatures is then regulated based on the heat flow exchange matrix. The temperature zone with the identified temperature anomaly is taken as the target control vertex. The associated vertices that can establish a heat flow channel with the target control vertex are identified. The associated vertices include the refrigeration system vertex, the phase change energy storage vertex, the external environment vertex, and the adjacent temperature zone vertex. The heat flow system between the target control vertex and each associated vertex is abstracted as a directed weighted temperature difference potential energy graph G=(V,E,W), where V contains the target control vertex and its associated vertices, E represents the heat flow channels that can be established between the target control vertex and each associated vertex, and W represents the weight value of each heat flow channel, defined as the temperature difference potential energy from one vertex to another. Real-time acquisition of target control vertex and associated data of each associated vertex; updating of the directed weighted temperature difference potential energy map G based on the associated data. Construct a target optimization function and combine it with the preset temperature recovery constraint of the target control vertex and the maximum heat flow constraint of the heat flow channel to solve the optimal heat flow of each heat flow channel in the directed weighted temperature difference potential energy map G. Generate a heat flow exchange matrix based on the optimal heat flow and convert the heat flow exchange matrix into control commands to execute temperature control. The method for finding the optimal heat flow rate for each heat flow channel is as follows: Construct a residual network based on the directed weighted temperature difference potential energy diagram G. In the residual network, the edges along the original heat flow channel direction are defined as forward edges, and the edges opposite to the original heat flow channel direction are defined as reverse edges. Initialize the actual heat flow rate of each forward edge to zero, the residual capacity of the forward edge to the maximum heat flow rate preset for the corresponding heat channel, and the residual capacity of the reverse edge to zero. Then, convert the temperature recovery constraint of the target control vertex into the net heat demand. Define power supply nodes and power receiving nodes. Search for augmenting paths from power supply nodes to power receiving nodes in the residual network. Calculate the total path potential of each augmenting path and select the augmenting path with the largest total path potential as the optimal augmenting path for augmentation. Calculate the residual capacity of all forward edges on the optimal augmenting path, take the minimum residual capacity of the forward edge as the bottleneck capacity, obtain the remaining net heat demand of the target control vertex, correct the bottleneck capacity, and obtain the corrected flow rate. Push the corrected flow along the optimal augmenting path, update the actual heat flow and residual capacity of the forward edge, and update the residual capacity of the reverse edge simultaneously; perform iterative updates until the remaining net heat demand of the target control vertex in the residual network is zero, and map the actual heat flow accumulated by each forward edge in the residual network to the corresponding heat flow channel in the directed weighted temperature difference potential energy graph G to obtain the optimal heat flow of each heat flow channel.
[0029] Specifically, the temperature zone with the abnormal temperature is identified as the target control vertex. The associated vertices that can establish a heat flow channel with the target control vertex are identified. The associated vertices include the refrigeration system vertex, the phase change energy storage vertex, the external environment vertex, and the adjacent temperature zone vertex. The adjacent temperature zone vertex is other temperature zone vertex that is spatially adjacent to the target control vertex and whose temperature state allows for the establishment of heat flow exchange. The heat flow system between the target control vertex and its associated vertices is abstracted as a directed weighted temperature difference potential energy graph G=(V,E,W), where V contains the target control vertex and its associated vertices, E represents the heat flow channels that can be established between the target control vertex and its associated vertices, and W represents the weight value of each heat flow channel, defined as the temperature difference potential energy from one vertex to another; the temperature difference potential energy is calculated as follows: The temperature sensitivity coefficient, cargo value coefficient, and remaining shelf life coefficient of the cargo carried by the target control vertex are obtained, along with the corresponding anomaly score. A weighted calculation is then performed to obtain the heat flow priority factor of the target control vertex. The temperature sensitivity coefficient is obtained as follows: First, based on the category information of the goods carried by the target control vertex, the corresponding benchmark sensitivity coefficient is matched from the preset goods temperature sensitivity database. The database stores benchmark values for different types of goods in advance. For example, biological agents are 5.0, high-end fresh goods are 3.5, ordinary frozen meat is 1.5, and room temperature tolerant goods are 0.5. The category information is entered into the system through electronic waybills or RFID tags during loading. Then, the upper and lower limits of the safe storage temperature of the goods and the current real-time temperature are obtained. The absolute value of the temperature difference between the current real-time temperature and the upper and lower limits is calculated. The minimum absolute value of the temperature difference is taken and normalized as the critical approximation correction factor. The smaller the critical approximation correction factor, the closer the current temperature is to the dangerous critical zone. Therefore, the constant 1 is added to the reciprocal of the critical approximation correction factor and multiplied by the benchmark sensitivity coefficient to obtain the temperature sensitivity coefficient. The temperature sensitivity coefficient characterizes the tolerance of the goods to temperature fluctuations. The larger the coefficient, the more easily the goods are deteriorated in quality due to temperature deviation. The cargo value coefficient is the unit value of cargo after normalization. The cargo value coefficient represents the risk of economic loss of cargo. The larger the coefficient, the more priority the cargo in that temperature zone needs to be protected. The remaining shelf life coefficient is the time from the current time to the expiration date of the goods. After normalization, the duration coefficient is obtained. Subtracting the duration coefficient from 1 gives the remaining shelf life coefficient. The remaining shelf life coefficient represents the time urgency of the goods. The larger the coefficient, the closer the goods are to the end of their shelf life and the more stringent the requirements for temperature stability. The weights of the temperature sensitivity coefficient, cargo value coefficient, and remaining shelf life coefficient are determined using existing technologies such as the analytic hierarchy process or expert experience assignment method. The larger the heat flow priority factor, the higher the urgency of temperature control for the cargo in that temperature zone, and the higher the priority of the corresponding heat flow channel in heat allocation. Then, for any edge (from vertex i to vertex j), obtain the maximum heat flow and heat exchange efficiency of the corresponding vertex i to vertex j. The maximum heat flow and heat exchange efficiency are obtained by conventional thermal calculation methods or experimental measurement methods in this field based on the design parameters of the heat flow channel and the calibration performance of the heat exchanger. This is existing technology and will not be elaborated here. Obtain the real-time temperature of vertex i and vertex j, and subtract the real-time temperature of node j from the real-time temperature of node i to obtain the real-time temperature difference. Multiply the real-time temperature difference, heat flow priority factor, maximum heat flow, and heat exchange efficiency to obtain the weight value of the edge (from vertex i to vertex j), i.e., the temperature difference potential energy. The larger the value, the stronger the comprehensive transmission advantage of this edge (heat flow channel) from vertex i to vertex j, and it should be given priority in temperature control. Real-time acquisition of relevant data such as temperature, heat flow, and operating status parameters of each heat flow channel of the target control vertex and its associated vertices, and updating of the directed weighted temperature difference potential energy graph G, such as updating the vertex attributes and edge weights in the directed weighted temperature difference potential energy graph. Then, an objective optimization function is constructed with the goal of maximizing the utilization rate of the total system temperature difference potential energy and minimizing the auxiliary energy consumption. The utilization rate of the total system temperature difference potential energy is the sum of the products of the temperature difference potential energy and the actual heat flow of each heat flow channel, representing the marginal contribution of a unit heat flow to temperature recovery. The auxiliary energy consumption is the sum of the operating power consumption of the refrigeration system and the phase change energy storage unit. Combining the temperature recovery constraint preset at the target control vertex and the maximum heat flow constraint of the heat flow channel, the optimal heat flow of each heat flow channel in the directed weighted temperature difference potential energy graph G is solved. The temperature recovery constraint is as follows: for the target control vertex, set its temperature recovery target range, calculate the deviation between the current temperature of the target control vertex and the target temperature, and convert it into the net heat that needs to be injected or discharged based on the heat capacity characteristics of the temperature range. The temperature recovery constraint requires that the net algebraic sum of all heat flows into or out of the target control vertex equals the net heat, ensuring that the temperature returns to the target range. Maximum heat flow constraint: For each heat flow channel, the upper limit of its heat flow is set as the maximum heat flow. During the solution process, the actual heat flow of any heat flow channel must be less than or equal to the corresponding maximum heat flow. The objective function is combined with the above constraints to form a constrained extreme value optimization problem. The improved Ford-Fulkerson algorithm is used to solve for the optimal heat flow rate of each heat flow channel, specifically including the following steps: Based on the locally directed weighted temperature difference potential energy graph G, a residual network Gf is constructed. In the residual network, edges along the original heat flow channel direction are defined as forward edges, and edges opposite to the original heat flow channel direction are defined as reverse edges. The actual heat flux of each forward edge is initialized to zero, i.e. The forward edge residual capacity is the maximum heat flow preset for the corresponding heat channel, i.e. The residual capacity of the reverse edge is zero, that is The temperature recovery constraint at the target control peak is converted into net heat demand. The net heat demand is obtained by multiplying the deviation between the current temperature of the target control peak and the target set temperature and the preset equivalent heat capacity of the target control peak. The net heat demand serves as the upper limit constraint and iteration termination condition for the subsequent augmented flow rate. In the residual network Gf, nodes with temperatures higher than the target control vertex are designated as power supply nodes, and nodes with temperatures lower than the target control vertex are designated as power receiving nodes. Augmentation paths are searched from power supply nodes to power receiving nodes. The temperature difference potential energy gradient is introduced as a priority criterion for selecting augmentation paths: the sum of the temperature difference potential energy of each edge on each augmentation path is calculated as the total path potential energy, and the augmentation path with the largest total path potential energy is selected as the optimal augmentation path to ensure that the marginal contribution of unit heat flow to temperature recovery is maximized. When there are multiple augmentation paths with similar total path potential energy, the path with the minimum total auxiliary energy consumption is selected. For the optimal augmenting path P found, calculate the residual capacity of all forward edges on the optimal augmenting path, and take the minimum residual capacity of the forward edge as the bottleneck capacity. Obtain the remaining net heat demand at the target control peak, and take the smaller value between the bottleneck capacity and the remaining net heat demand as the correction flow rate. Here, the initial value of the remaining net heat demand is equal to the net heat demand, and it decreases after each iteration when the corrected flow is pushed, with the decrease amount being the corrected flow for this iteration; the corrected flow is pushed along the optimal augmenting path P. For the forward edge ( ): Update forward edge residual capacity Update actual heat flow For the reverse edge Update the residual capacity of the reverse edge. Repeatedly perform updates, iteratively search for new augmenting paths and push traffic, and iteratively update the actual heat flow, forward edge residual capacity and reverse edge residual capacity. After each iteration, update the remaining net heat demand of the target control vertex until the remaining net heat demand of the target control vertex in the residual network is zero or there is no feasible augmenting path from the power supply node to the power receiving node. At this time, the actual heat flow accumulated by each forward edge in the residual network is mapped to the corresponding heat flow channel in the directed weighted temperature difference potential energy graph G to obtain the optimal heat flow of each heat flow channel. The optimal heat flow of each heat flow channel is filled into a square matrix of |V|×|V| according to the node topology to construct a real-time heat flow exchange matrix. In the matrix, the matrix elements represent the optimal heat flow of each heat flow channel. The real-time heat flow exchange matrix is converted into control commands for each heat flow exchange channel, including start / stop and power adjustment commands for the refrigeration system, charge / discharge switching commands for the phase change energy storage unit, and opening adjustment commands for heat flow valves between temperature zones, so as to realize intelligent temperature control of the target control vertex.
[0030] This approach avoids the computational overhead of full system reconstruction by constructing a locally oriented weighted temperature difference potential energy map for only the temperature anomaly zone and solving the heat flow exchange matrix in real time, thus achieving rapid response. Based on the temperature difference potential energy gradient, it optimizes the augmented path selection and combines temperature recovery, maximum flow rate, and energy conservation constraints to ensure precise heat transfer along high potential energy difference and high-efficiency channels, minimizing auxiliary energy consumption while meeting cargo priority requirements. Using net heat demand as the iteration termination condition effectively prevents over-adjustment or under-adjustment. The final generated heat flow exchange matrix can be directly mapped to equipment control commands, realizing a closed loop from anomaly identification to precise control. This significantly improves the control accuracy, real-time performance, and energy efficiency of the multi-temperature zone temperature control system.
[0031] The temperature zone with temperature anomalies is further constructed into multiple sub-temperature zones. Temperature monitoring data and heat flow exchange data of each sub-temperature zone are obtained. The priority coefficient of each sub-temperature zone is calculated based on the temperature monitoring data and heat flow exchange data. The sub-temperature zones are then regulated based on the priority coefficient. The method for regulating each sub-temperature zone based on the priority coefficient is as follows: the temperature zone is divided into multiple sub-temperature zones, the time-series temperature monitoring data in each sub-temperature zone is obtained, the history offset and state offset of each sub-temperature zone are calculated based on the time-series temperature monitoring data of all sub-temperature zones, and the population deviation of each sub-temperature zone is calculated based on the history offset and state offset. For each sub-temperature zone, construct a set of directly related sub-temperature zones and a set of spatially adjacent sub-temperature zones. Obtain temperature monitoring data and heat flow exchange data for each sub-temperature zone within the set of directly related sub-temperature zones and the set of spatially adjacent sub-temperature zones, respectively. Calculate the temperature anomaly and heat flow anomaly for each sub-temperature zone based on the temperature monitoring data and heat flow exchange data. Calculate the neighborhood coupling degree for each sub-temperature zone based on the temperature anomaly and heat flow anomaly. The group deviation and neighborhood coupling are weighted and accumulated to obtain the priority coefficient of each sub-temperature zone. The sub-temperature zone with the highest priority coefficient is selected for priority control. The calculation methods for population deviation and neighborhood coupling are as follows: Time-series temperature monitoring data of each sub-temperature zone under historical time series are obtained; a time-series temperature characteristic function of each sub-temperature zone is fitted; the temperature integral value of the time-series temperature characteristic function under historical time series is calculated; the average temperature integral value is calculated from the mean of all temperature integral values; the absolute value of the difference between the temperature integral value of each sub-temperature zone and the average temperature integral value is used as the history deviation; the temperature monitoring data of each sub-temperature zone at the current moment is obtained; the instantaneous temperature mean is calculated from the mean of all temperature monitoring data; the absolute value of the difference between the temperature monitoring data of each sub-temperature zone and the instantaneous temperature mean is used as the state deviation; the history deviation and the state deviation are weighted and accumulated to obtain the population deviation. The absolute values of the differences between the temperature monitoring data and heat flow exchange data of each sub-temperature zone within the directly associated sub-temperature zone set and the spatially adjacent sub-temperature zone set and their respective corresponding temperature standard values and heat flow exchange ideal values are calculated to obtain the direct temperature disturbance degree, direct heat flow disturbance degree, adjacent temperature disturbance degree, and adjacent heat flow disturbance degree of each sub-temperature zone. The temperature anomaly is obtained based on the direct temperature disturbance degree and the adjacent temperature disturbance degree, and the heat flow anomaly is obtained based on the direct heat flow disturbance degree and the adjacent heat flow disturbance degree. The temperature anomaly and the heat flow anomaly are weighted and accumulated to obtain the neighborhood coupling degree.
[0032] When a temperature zone is determined to be abnormal, traditional overall control methods can lead to uneven distribution of cooling capacity: areas with large local temperature deviations within the abnormal zone cannot receive sufficient cooling capacity for rapid recovery, while areas with smaller temperature deviations are over-cooled, resulting in energy waste and potentially causing frost damage to some goods. This embodiment further subdivides the abnormal temperature zone into multiple sub-temperature zones, comprehensively considering the temperature deviation of each sub-temperature zone and its coupling effect with surrounding areas, calculating the control priority coefficient of each sub-temperature zone, and achieving targeted and precise distribution of cooling capacity. This ensures priority recovery in high-risk areas while minimizing temperature disturbances to other areas. Specifically: Using the temperature sensors already deployed within the abnormal temperature zone as the core, and combining the physical structure of the temperature zone (such as shelf location and partition distribution) and cargo stacking information, the abnormal temperature zone is further divided into multiple sub-temperature zones. Time-series temperature monitoring data is acquired for each sub-temperature zone. Based on the time-series temperature monitoring data of all sub-temperature zones, the history offset and state offset of each sub-temperature zone are calculated. Based on the history offset and state offset, the group deviation of each sub-temperature zone is calculated. The calculation method is as follows: Acquire time-series temperature monitoring data for each sub-temperature zone in the historical time series. The historical time series can be manually set, such as the 30 seconds before the current moment. Fit the time-series temperature characteristic function for each sub-temperature zone, calculate the temperature integral value of the time-series temperature characteristic function in the historical time series, and calculate the average temperature integral value from the average of all temperature integral values. Use the absolute value of the difference between the temperature integral value of each sub-temperature zone and the average temperature integral value as the history offset. The history offset reflects the long-term deviation trend of the sub-temperature zone in the time dimension. The larger the value, the greater the degree of deviation and the more obvious the long-term abnormal trend. The temperature monitoring data of each sub-temperature zone at the current moment is obtained, and the instantaneous temperature mean is calculated from the mean of all temperature monitoring data. The absolute value of the difference between the temperature monitoring data of each sub-temperature zone and the instantaneous temperature mean is used as the state value offset. The state value offset reflects the instantaneous deviation state at the current moment. The larger the value, the more the temperature of the sub-temperature zone deviates from the overall average level at the current moment, and the more serious the instantaneous abnormal state. The population deviation is obtained by weighting and summing the history offset and state offset. Since all sub-temperature zones are in the same temperature zone, the larger the population deviation, the more severe the temperature anomaly of that sub-temperature zone, the greater its impact on the temperature stability of the entire abnormal temperature zone, and the more it needs to be prioritized for regulation. Here, both the history offset and state offset are normalized values, and the weighting coefficient of the history offset is 0.4, while the weighting coefficient of the state offset is 0.6. This weighting allocation reflects the principle of prioritizing the current instantaneous state and supplementing it with the historical cumulative trend, which is more in line with the urgency of temperature anomalies in cold chain transportation.
[0033] For each sub-temperature zone (sub-temperature zone k), construct a set of directly related sub-temperature zones and a set of spatially adjacent sub-temperature zones. The set of directly related sub-temperature zones includes all sub-temperature zones that share the same cooling vent, heat exchange valve, or heat flow channel with sub-temperature zone k. These sub-temperature zones have a direct cooling capacity distribution relationship with sub-temperature zone k, and a change in the cooling capacity of one sub-temperature zone will directly affect the cooling capacity supply of other sub-temperature zones. The set of spatially adjacent sub-temperature zones includes all sub-temperature zones that are physically adjacent to sub-temperature zone k (sharing a surface). These sub-temperature zones have a direct heat conduction relationship with sub-temperature zone k, and a temperature change in one sub-temperature zone will affect adjacent sub-temperature zones through heat conduction. Temperature monitoring data and heat flow exchange data of each sub-temperature region within the directly associated sub-temperature region set and the spatially adjacent sub-temperature region set are obtained respectively. Based on the temperature monitoring data and heat flow exchange data, the temperature anomaly and heat flow anomaly of each sub-temperature region are calculated. Based on the temperature anomaly and heat flow anomaly, the neighborhood coupling degree of each sub-temperature region is calculated. The calculation method is as follows: The temperature monitoring data and heat flow exchange data of each sub-temperature zone in the directly associated sub-temperature zone set and the spatially adjacent sub-temperature zone set are compared with their respective temperature standard values and heat flow exchange ideal values. The absolute value of the difference is calculated to obtain the direct temperature disturbance degree, direct heat flow disturbance degree, adjacent temperature disturbance degree, and adjacent heat flow disturbance degree of each sub-temperature zone. Specifically, the real-time temperature of each sub-temperature zone within the directly associated sub-temperature zone set is obtained. Each sub-temperature zone has a corresponding temperature standard value set based on experience. The absolute value of the difference between the real-time temperature and the corresponding temperature standard value is calculated to obtain the direct temperature disturbance degree of each sub-temperature zone. Similarly, the real-time heat flow of each sub-temperature zone within the directly associated sub-temperature zone set is obtained. Each sub-temperature zone has a corresponding ideal heat flow exchange value set based on experience. The absolute value of the difference between the real-time heat flow and the corresponding ideal heat flow exchange value is calculated to obtain the direct heat flow disturbance degree of each sub-temperature zone. The real-time temperature of each sub-temperature zone within the set of adjacent sub-temperature zones in the space is obtained. Each sub-temperature zone has a corresponding temperature standard value set based on experience. The absolute value of the difference between the real-time temperature and the corresponding temperature standard value is calculated to obtain the adjacent temperature disturbance degree of each sub-temperature zone within the set of adjacent sub-temperature zones in the space. Similarly, the real-time heat flow of each sub-temperature zone within the set of adjacent sub-temperature zones in the space is obtained. Each sub-temperature zone has a corresponding ideal heat flow exchange value set based on experience. The absolute value of the difference between the real-time heat flow and the corresponding ideal heat flow exchange value is calculated to obtain the adjacent heat flow disturbance degree of each sub-temperature zone within the set of adjacent sub-temperature zones in the space. The mean value of direct temperature disturbance is calculated by taking the mean value of all direct temperature disturbances within the set of directly related sub-temperature zones. The mean value of direct heat flow disturbance is calculated by taking the mean value of all direct heat flow disturbances within the set of directly related sub-temperature zones. The mean values of adjacent temperature disturbances and adjacent heat flow disturbances within the sets of spatially adjacent sub-temperature zones are calculated separately to obtain the mean values of adjacent temperature disturbances and adjacent heat flow disturbances. The mean values of direct temperature disturbances and adjacent temperature disturbances are weighted and summed to obtain the temperature anomaly. The mean values of direct heat flow disturbances and adjacent heat flow disturbances are weighted and summed to obtain the heat flow anomaly. Temperature anomaly measures the degree of temperature fluctuation in the surrounding associated regions (directly associated + spatially adjacent) of a single sub-temperature zone. A larger value indicates a more unstable temperature environment around the sub-temperature zone, and a greater conductive impact of temperature anomalies in the surrounding areas on the sub-temperature zone. Heat flux anomaly measures the degree of heat flux exchange fluctuation in the surrounding associated regions of a single sub-temperature zone. A larger value indicates a more uneven distribution of cooling capacity around the sub-temperature zone, and a greater impact of heat flux anomalies in the surrounding areas on the cooling capacity supply of the sub-temperature zone. Therefore, the temperature anomaly and heat flux anomaly are weighted and summed to obtain the neighborhood coupling degree. A larger neighborhood coupling degree indicates a closer coupling relationship between the sub-temperature zone and its surrounding regions, and a greater influence on the surrounding areas when regulating the sub-temperature zone. The greater the temperature disturbance in the edge region, the easier it is for anomalies in the surrounding regions to propagate to that sub-temperature region, and the more priority it needs to be controlled. Here, both the temperature anomaly and the heat flux anomaly are normalized values. Finally, the group deviation degree and the neighborhood coupling degree are weighted and accumulated to obtain the priority coefficient of each sub-temperature region. All sub-temperature regions are sorted in descending order of priority coefficient, and the sub-temperature region with the highest priority coefficient is the first priority control target. And so on, the sub-temperature region with the highest priority coefficient is selected for priority control. Here, the weight coefficients set for each calculation step can be determined based on the experience and professional knowledge of those in the field, and will not be described in detail here. This approach addresses the problems of uneven cooling distribution, inability to quickly recover in high-risk areas, and energy waste and cargo frost damage caused by over-cooling in areas with small temperature deviations, inherent in traditional overall control methods for abnormal temperature zones. It further subdivides the abnormal temperature zone into independently monitorable and controllable sub-temperature zones. By fusing historical offset and state offset to calculate the population deviation, it accurately quantifies the long-term cumulative trend and instantaneous urgency of temperature anomalies within each sub-temperature zone. Simultaneously, it constructs a dual correlation system of direct correlation and spatial adjacency, calculating the neighborhood coupling degree through temperature anomaly and heat flow anomaly, comprehensively considering the relationship between the quantum temperature zone and its surrounding areas. The solution addresses the risks of mutual influence and abnormal spread, ultimately achieving targeted and precise allocation and dynamic rescheduling of cooling capacity based on a comprehensive priority coefficient. This approach not only prioritizes ensuring the rapid recovery of the sub-temperature zones with the most severe temperature anomalies and the highest risk of spread to normal temperatures, significantly shortening anomaly response time and effectively reducing the risk of spoilage and loss of high-value cold chain goods, but also avoids over-cooling of areas with normal temperatures, significantly reducing the energy consumption of the refrigeration system. Furthermore, through the constraint effect of neighborhood coupling, it minimizes the temperature disturbance of surrounding areas caused by control operations, comprehensively improving the control accuracy, stability, and intelligence level of the multi-temperature zone temperature control system for refrigerated semi-trailers.
[0034] In one embodiment, a multi-temperature zone adaptive dynamic temperature control system for refrigerated semi-trailers is disclosed, such as... Figure 2 As shown, the system includes: The data acquisition module is used to divide the interior of the refrigerated semi-trailer into multiple temperature zones and collect external influence data and internal status data for each temperature zone. The anomaly detection module is used to analyze the external influence data and internal state data of each temperature zone, construct a temperature spatiotemporal causal map of the temperature zone, obtain anomaly scores for each temperature zone based on the temperature spatiotemporal causal map, and determine whether there are temperature anomalies in each temperature zone based on the anomaly scores. The temperature control module is used to acquire temperature zones with abnormal temperatures and construct a heat flow exchange matrix. Based on the heat flow exchange matrix, the temperature of the temperature zones with abnormal temperatures is controlled. The control implementation module is used to further construct multiple sub-temperature zones from the temperature zone with temperature anomalies, acquire temperature monitoring data and heat flow exchange data of each sub-temperature zone, calculate the priority coefficient of each sub-temperature zone based on the temperature monitoring data and heat flow exchange data, and control each sub-temperature zone based on the priority coefficient.
[0035] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
[0036] All formulas in this manual are dimensionless and calculated numerically. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.
[0037] Although embodiments of the invention have been shown and described, those skilled in the art will understand 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 claims and their equivalents.
Claims
1. A multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers, characterized in that, The methods include: The interior of the refrigerated semi-trailer was divided into multiple temperature zones, and external influence data and internal status data of each temperature zone were collected. The external influence data and internal state data of each temperature zone are analyzed to construct a temperature spatiotemporal causal map of the temperature zone. Based on the temperature spatiotemporal causal map of the temperature zone, the anomaly score of each temperature zone is obtained, and the anomaly score is used to determine whether there is a temperature anomaly in each temperature zone. The temperature regions with abnormal temperatures are identified, and a heat flow exchange matrix is constructed. The temperature of the temperature regions with abnormal temperatures is then regulated based on the heat flow exchange matrix. The temperature zone with temperature anomalies is further divided into multiple sub-temperature zones. Temperature monitoring data and heat flow exchange data of each sub-temperature zone are obtained. The priority coefficient of each sub-temperature zone is calculated based on the temperature monitoring data and heat flow exchange data. The sub-temperature zones are then regulated based on the priority coefficient.
2. The multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers according to claim 1, characterized in that, The method for constructing a temperature spatiotemporal causal map of a temperature region is as follows: Four types of nodes are predefined: temperature zone grid nodes, physical component nodes, environmental factor nodes, and travel event nodes. Each node has its own attributes. All nodes are paired to form multiple node pairs. Based on preset physical prior knowledge, multiple candidate node pairs are selected from the multiple node pairs. Correlation analysis is performed on each candidate node pair to select candidate node pairs that meet preset conditions. These are denoted as causal association pairs, and corresponding causal edges are constructed based on the causal association pairs. The causal influence strength of each causal edge is calculated and denoted as the corresponding causal edge weight. The edge connecting adjacent temperature zone grid nodes is denoted as a spatial edge. The thermal conductivity coefficients of the two temperature zone grid nodes corresponding to the spatial edge are calculated and denoted as the corresponding spatial edge weights. By combining the temperature zone grid nodes, physical component nodes, environmental factor nodes, travel event nodes, as well as the causal edge weights and spatial edge weights, a temperature spatiotemporal causal graph of the temperature zone is obtained.
3. The multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers according to claim 2, characterized in that, The calculation methods for the intensity of causal influence and the thermal conductivity coefficient are as follows: Within a preset delay time range, the correlation degree of causal pairs under different delay times is calculated, and the maximum correlation degree within the preset delay time range is taken as the optimal correlation degree; the correlation type of the causal pairs is obtained, and the correlation coefficient is determined; the partial correlation coefficient of the causal pairs is calculated, and the absolute value of the partial correlation coefficient is taken to obtain the direct causal proportion coefficient; historical data of causal pairs is obtained and divided into N equal-length time windows, and the optimal correlation degree in each time window is calculated. The average correlation degree is calculated by averaging the optimal correlation degrees of all time windows; the maximum optimal correlation degree in all time windows is recorded as the maximum correlation degree, and the time stability coefficient is divided by the average correlation degree to obtain the time stability coefficient; the causal influence strength is obtained by multiplying the optimal correlation degree, correlation coefficient, direct causal proportion coefficient, and time stability coefficient together and then normalizing the result. In the spatial boundary, obtain the thermal conductivity between adjacent temperature zone grid nodes and calculate the interface equivalent thermal conductivity between the two nodes; obtain the contact area between adjacent temperature zone grid nodes and the distance between the center points of adjacent temperature zone grid nodes; divide the contact area by the distance and multiply by the interface equivalent thermal conductivity to obtain the thermal conductivity value, and normalize the thermal conductivity value to obtain the thermal conductivity coefficient.
4. The multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers according to claim 3, characterized in that, The method for determining whether there are temperature anomalies in each temperature zone is as follows: Based on the constructed temperature spatiotemporal causal graph of the temperature zone, the counterfactual theoretical temperature of each temperature zone grid node is calculated; the real-time temperature of each temperature zone grid node is obtained, and the absolute value of the difference between the counterfactual theoretical temperature and the actual temperature is calculated and recorded as the deviation value. The duration and spread range of the deviation are obtained. The deviation value, duration and spread range are weighted and accumulated to obtain the anomaly score of each temperature zone grid node. The anomaly score is compared with the preset anomaly score threshold. If the anomaly score exceeds the anomaly score threshold, it is determined that there is a temperature anomaly in the temperature zone corresponding to the temperature zone grid node.
5. The multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers according to claim 4, characterized in that, The method for calculating counterfactual temperature is as follows: The temperature zone grid node to be calculated is denoted as the target temperature zone grid node. All causal relationships containing the target temperature zone grid node are extracted from the temperature zone spatiotemporal causal graph. All nodes upstream of the target temperature zone grid node are extracted from all causal relationships and denoted as upstream nodes. For each causal relationship pair and its corresponding causal edge, a time-series lag regression method is used to traverse multiple lag time windows and calculate the correlation degree between the upstream node and the target temperature zone grid node under different lag time windows. The lag time corresponding to the maximum correlation degree is taken as the causal influence delay time. Obtain the historical state data of each upstream node corresponding to the causal impact delay time back to the current moment, multiply the historical state data by the corresponding causal edge weight to obtain the unit causal impact value of the corresponding upstream node, and accumulate the unit causal impact values of all upstream nodes to obtain the comprehensive causal impact value. Construct an autoregressive inertial model of the grid nodes in the target temperature zone and calculate the temperature inertial reference value of the grid nodes in the target temperature zone; obtain the actual temperature and thermal conductivity coefficient of all grid nodes in the temperature zone adjacent to the grid nodes in the target temperature zone and calculate the spatial thermal conduction influence value of the grid nodes in the target temperature zone; weight and superimpose the temperature inertial reference value, spatial thermal conduction influence value, and causal comprehensive influence value to obtain the counterfactual theoretical temperature.
6. The multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers according to claim 1, characterized in that, The method for temperature control in temperature zones with anomalies based on the heat flow exchange matrix is as follows: The temperature zone with the identified temperature anomaly is taken as the target control vertex. The associated vertices that can establish a heat flow channel with the target control vertex are identified. The associated vertices include the refrigeration system vertex, the phase change energy storage vertex, the external environment vertex, and the adjacent temperature zone vertex. The heat flow system between the target control vertex and each associated vertex is abstracted as a directed weighted temperature difference potential energy graph G=(V,E,W), where V contains the target control vertex and its associated vertices, E represents the heat flow channels that can be established between the target control vertex and each associated vertex, and W represents the weight value of each heat flow channel, defined as the temperature difference potential energy from one vertex to another. Real-time acquisition of target control vertex and associated data of each associated vertex; updating of the directed weighted temperature difference potential energy map G based on the associated data. Construct a target optimization function and combine it with the preset temperature recovery constraint of the target control vertex and the maximum heat flow constraint of the heat flow channel to solve for the optimal heat flow of each heat flow channel in the directed weighted temperature difference potential energy map G. Generate a heat flow exchange matrix based on the optimal heat flow and convert the heat flow exchange matrix into control commands to execute temperature control.
7. The multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers according to claim 6, characterized in that, The method for determining the optimal heat flow rate for each heat flow channel is as follows: Based on the directed weighted temperature difference potential energy graph G, a residual network is constructed. In the residual network, the edges along the original heat flow channel direction are defined as forward edges, and the edges opposite to the original heat flow channel direction are defined as reverse edges. The actual heat flow of each forward edge is initialized to zero, the residual capacity of the forward edge is the maximum heat flow preset for the corresponding heat channel, and the residual capacity of the reverse edge is zero. The temperature recovery constraint of the target control vertex is converted into net heat demand. Define power supply nodes and power receiving nodes. Search for augmenting paths from power supply nodes to power receiving nodes in the residual network. Calculate the total path potential of each augmenting path and select the augmenting path with the largest total path potential as the optimal augmenting path for augmentation. Calculate the residual capacity of all forward edges on the optimal augmenting path, take the minimum residual capacity of the forward edge as the bottleneck capacity, obtain the remaining net heat demand of the target control vertex, correct the bottleneck capacity, and obtain the corrected flow rate. Push the corrected flow along the optimal augmenting path, update the actual heat flow and residual capacity of the forward edge, and update the residual capacity of the reverse edge simultaneously. Perform iterative updates until the remaining net heat demand of the target control vertex in the residual network is zero. Map the actual heat flow accumulated by each forward edge in the residual network to the corresponding heat flow channel in the directed weighted temperature difference potential energy graph G to obtain the optimal heat flow of each heat flow channel.
8. The multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers according to claim 1, characterized in that, The method for regulating each sub-temperature zone based on priority coefficients is as follows: The temperature zone is divided into multiple sub-temperature zones, and time-series temperature monitoring data in each sub-temperature zone is obtained. Based on the time-series temperature monitoring data of all sub-temperature zones, the history offset and state offset of each sub-temperature zone are calculated. Based on the history offset and state offset, the population deviation of each sub-temperature zone is calculated. For each sub-temperature zone, construct a set of directly related sub-temperature zones and a set of spatially adjacent sub-temperature zones. Obtain temperature monitoring data and heat flow exchange data for each sub-temperature zone within the set of directly related sub-temperature zones and the set of spatially adjacent sub-temperature zones, respectively. Calculate the temperature anomaly and heat flow anomaly for each sub-temperature zone based on the temperature monitoring data and heat flow exchange data. Calculate the neighborhood coupling degree for each sub-temperature zone based on the temperature anomaly and heat flow anomaly. The group deviation degree and the neighborhood coupling degree are weighted and accumulated to obtain the priority coefficient of each sub-temperature zone. The sub-temperature zone with the highest priority coefficient is selected for priority control.
9. The multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers according to claim 8, characterized in that, The methods for calculating group deviation and neighborhood coupling are as follows: Acquire time-series temperature monitoring data for each sub-temperature zone under a preset historical time series, and fit a time-series temperature characteristic function for each sub-temperature zone; calculate the temperature integral value of the time-series temperature characteristic function under the historical time series, and calculate the average temperature integral value by averaging all temperature integral values; use the absolute value of the difference between the temperature integral value of each sub-temperature zone and the average temperature integral value as the history offset; acquire the temperature monitoring data of each sub-temperature zone at the current moment, and calculate the average instantaneous temperature by averaging all temperature monitoring data; use the absolute value of the difference between the temperature monitoring data of each sub-temperature zone and the instantaneous temperature average value as the state offset; weightedly sum the history offset and the state offset to obtain the population deviation. The temperature monitoring data and heat flow exchange data of each sub-temperature zone in the directly associated sub-temperature zone set and the spatially adjacent sub-temperature zone set are compared with their respective temperature standard values and heat flow exchange ideal values. The absolute value of the difference is calculated to obtain the direct temperature disturbance degree, direct heat flow disturbance degree, adjacent temperature disturbance degree, and adjacent heat flow disturbance degree of each sub-temperature zone. Temperature anomaly is obtained based on direct temperature perturbation and adjacent temperature perturbation, and heat flow anomaly is obtained based on direct heat flow perturbation and adjacent heat flow perturbation. The neighborhood coupling degree is obtained by weighted summation of temperature anomaly and heat flow anomaly.
10. A multi-temperature zone adaptive dynamic temperature control system for refrigerated semi-trailers, implementing the multi-temperature zone adaptive dynamic temperature control method for refrigerated semi-trailers as described in any one of claims 1-9, characterized in that the system... include: The data acquisition module is used to divide the interior of the refrigerated semi-trailer into multiple temperature zones and collect external influence data and internal status data for each temperature zone. The anomaly detection module is used to analyze the external influence data and internal state data of each temperature zone, construct a temperature spatiotemporal causal map of the temperature zone, obtain anomaly scores for each temperature zone based on the temperature spatiotemporal causal map, and determine whether there are temperature anomalies in each temperature zone based on the anomaly scores. The temperature control module is used to acquire temperature zones with abnormal temperatures and construct a heat flow exchange matrix. Based on the heat flow exchange matrix, the temperature of the temperature zones with abnormal temperatures is controlled. The control implementation module is used to further construct multiple sub-temperature zones from the temperature zone with temperature anomalies, acquire temperature monitoring data and heat flow exchange data of each sub-temperature zone, calculate the priority coefficient of each sub-temperature zone based on the temperature monitoring data and heat flow exchange data, and control each sub-temperature zone based on the priority coefficient.
