AI-based multimodal transport collaborative optimization method and system
By constructing temperature sampling sequences and internal loss dynamics models, the temperature deviation trend of cold chain goods is identified, and the matching of refrigerated truck capacity is optimized. This solves the problems of scientific formulation of temperature control strategies and capacity matching in multimodal transport of cold chain goods, and realizes efficient and dynamic temperature control and risk management of cold chain goods.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TOP XINGDA
- Filing Date
- 2025-08-11
- Publication Date
- 2026-05-26
AI Technical Summary
In the process of multimodal transport of cold chain goods, existing technologies have failed to effectively quantify the comprehensive impact of the duration and magnitude of temperature deviation on quality, making it difficult to scientifically formulate transportation timeliness and temperature control strategies. Furthermore, the temperature sensitivity coefficients of refrigerated trucks and goods have not been dynamically matched, leading to the risk of exceeding quality standards or idle high-quality transport capacity.
By acquiring the temperature of non-temperature-controlled areas during flight delays, a temperature sampling sequence is constructed and peak factors are identified. Peak deviation analysis is performed, an internal loss dynamic model is established, intermodal transport capacity units are matched, temperature deviation correction and capacity adjustment are carried out, and AI optimization strategies are used to minimize the cumulative temperature deviation.
It enables multi-dimensional capture of temperature deviation events in cold chain goods, reduces cargo damage risk, enhances front-end monitoring capabilities in transportation, dynamically assesses transportation risks, optimizes capacity utilization, and achieves dynamic temperature control and cargo damage risk control throughout the entire supply chain.
Smart Images
Figure CN120975678B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent transportation technology, specifically to an AI-based multimodal transport collaborative optimization method and system. Background Technology
[0002] In the process of multimodal transport of cold chain goods, the goods need to go through multiple links such as air, road and warehousing. The temperature control environment of each link is significantly different. In particular, the risk of deterioration of goods quality due to flight delays and non-temperature control areas is difficult to control.
[0003] Existing technologies do not take into account the thermal inertia of goods and the lag of biochemical reactions, and only use linear models to estimate the safety window. They cannot quantify the comprehensive impact of the duration and magnitude of temperature deviation on quality, making it difficult to support the scientific formulation of transportation timeliness and temperature control strategies.
[0004] There is no dynamic matching mechanism between the thermal inertia capacity of refrigerated trucks and other transport units and the temperature sensitivity coefficient of goods. Traditional scheduling strategies are mostly based on distance or cost priority principles, which leads to highly sensitive goods being assigned to vehicles with insufficient refrigeration performance, causing the risk of exceeding quality standards or resulting in idle high-quality transport capacity. Summary of the Invention
[0005] The purpose of this invention is to provide an AI-based multimodal transport collaborative optimization method and system to solve at least one of the aforementioned problems in the prior art.
[0006] The AI-based multimodal transport collaborative optimization method and system includes the following steps:
[0007] The temperature of cold chain cargo in the non-temperature-controlled area of flight delay is obtained and a temperature sampling sequence is constructed. A temperature adaptive baseline is established and the temperature peak factor of the temperature sampling sequence is identified. Peak deviation analysis is performed to identify the peak change trend.
[0008] An inertial cumulative effect analysis is performed on the cumulative amount of temperature deviation with an upward risk trend. An internal loss dynamic model is constructed to obtain the preset maximum quality loss amount. Then, a safety window boundary function and curve are constructed to determine the safety window of temperature deviation.
[0009] Intermodal transport risk analysis is performed based on the temperature deviation safety window to obtain intermodal transport risk values. Based on the intermodal transport risk values, a matching algorithm is established with the goal of minimizing the cumulative temperature deviation to match intermodal transport capacity units.
[0010] Based on the determined intermodal transport capacity unit, a temperature deviation correction analysis is performed on the intermodal transport capacity unit to determine whether the temperature deviation of the intermodal transport capacity unit meets the requirements. If it does not meet the requirements, the capacity adjustment strategy of the transport unit is adjusted. Combined with the quality loss after arrival, the internal loss dynamic model is calibrated.
[0011] As a further technical solution of the present invention: the method for identifying the temperature peak factor of the temperature sampling sequence is as follows:
[0012] Within the monitoring period, the temperature of cold chain goods corresponding to non-temperature-controlled areas of flight delays is obtained, and a temperature sampling sequence is constructed.
[0013] A peak identification model is established for the temperature sampling sequence to extract the temperature peak factor of the temperature sampling sequence.
[0014] As a further technical solution of the present invention, the baseline model is established as follows:
[0015] Establish a temperature-adaptive baseline;
[0016] A peak identification model is established using a flow cytometry anomaly detection algorithm. The adaptive baseline and temperature sampling sequence are input into the peak identification model to obtain the temperature peak factor.
[0017] As a further technical solution of the present invention: the method for determining the temperature deviation safety window is as follows:
[0018] The internal loss kinetic model is validated by first-order kinetic reaction. If the internal loss kinetic model conforms to first-order kinetic reaction, the preset maximum quality loss is obtained.
[0019] Based on the maximum quality loss, a safety window boundary function is constructed, and a constraint relationship between temperature deviation and time is established.
[0020] A safety window curve is constructed based on the safety boundary function to determine the temperature deviation safety window.
[0021] As a further technical solution of the present invention, the method for constructing the internal loss dynamics model is as follows:
[0022] If the trend of the peak change is an upward trend of risk, then obtain the cumulative temperature deviation within the current monitoring period;
[0023] An inertial accumulation effect analysis is performed on the temperature deviation accumulation. If an inertial accumulation effect exists, an inertial accumulation correction is performed on the temperature deviation, and a nonlinear correction function is constructed.
[0024] The activation energy, the rate constant at the reference temperature, and the deviation between the current temperature and the reference temperature are obtained, and an internal loss dynamic model is constructed by combining the nonlinear correction function.
[0025] As a further technical solution of the present invention: the method of matching the intermodal transport capacity unit is as follows:
[0026] Obtain the intermodal transport risk value. If the intermodal transport risk value is higher than or equal to the preset intermodal transport risk threshold, conduct intermodal transport scenario analysis and establish a matching algorithm with the goal of minimizing the cumulative temperature deviation.
[0027] The matching function value is used to match the corresponding intermodal transport capacity unit for cold chain goods.
[0028] As a further technical solution of the present invention: the method for obtaining the intermodal transport risk value is as follows:
[0029] Obtain the temperature peak factor in the current monitoring period and the previous monitoring period, calculate the frequency of the temperature peak factor in the current monitoring period and the frequency of the temperature peak factor in the previous monitoring period, and extract the peak rise factor.
[0030] Calculate the percentage of the cumulative temperature deviation in the current monitoring period relative to the safe temperature deviation window to obtain the window deviation ratio;
[0031] The intermodal transport risk value is obtained by summing the window deviation ratio and the peak rise factor.
[0032] As a further technical solution of the present invention: the method for extracting the peak rise factor is as follows:
[0033] Calculate the frequency deviation of the temperature peak factor between the current monitoring cycle and the previous monitoring cycle. If the frequency deviation is positive, the frequency deviation will be used as the peak rise factor.
[0034] As a further technical solution of the present invention: the method for performing the temperature deviation correction analysis is as follows:
[0035] Obtain the cargo sampling temperature of the intermodal transport capacity unit, as well as the preset target temperature;
[0036] The absolute deviation between the sampled temperature of the goods and the preset target temperature is calculated to obtain the absolute value of the temperature deviation.
[0037] Obtain the remaining amount of the safety window and combine it with the accumulated temperature deviation to obtain the correction urgency coefficient;
[0038] Based on the correction urgency coefficient, a correction strategy classification is determined to assess whether the temperature deviation of the intermodal transport capacity unit meets the requirements.
[0039] If the conditions are not met, the capacity adjustment strategy of the transport unit will be adjusted through the proportional integral algorithm.
[0040] The AI-based multimodal transport collaborative optimization system includes the following modules:
[0041] Trend identification module: In cold chain multimodal transport, the module obtains the temperature of cold chain goods corresponding to the non-temperature-controlled area of flight delays and constructs a temperature sampling sequence. It establishes a temperature adaptive baseline and identifies the temperature peak factor of the temperature sampling sequence, and performs peak deviation analysis to identify the peak change trend.
[0042] Window determination module: Based on the identified trend, the cumulative effect analysis of temperature deviation with the risk increase trend is carried out, the internal loss dynamic model is constructed to obtain the preset maximum quality loss, and then the safety window boundary function and curve are constructed to determine the temperature deviation safety window.
[0043] Intermodal matching module: Based on the temperature deviation safety window, intermodal risk analysis is performed to obtain intermodal risk values. Based on the intermodal risk values, a matching algorithm is established with the goal of minimizing the cumulative temperature deviation to match intermodal transport capacity units.
[0044] Model optimization module: Based on the determined intermodal transport capacity unit, temperature deviation correction analysis is performed on the intermodal transport capacity unit to determine whether the temperature deviation of the intermodal transport capacity unit meets the requirements. If it does not meet the requirements, the capacity adjustment strategy of the transport unit is adjusted. Combined with the quality loss after arrival, the internal loss dynamic model is calibrated.
[0045] The beneficial effects of this invention are:
[0046] 1. By collecting temperatures in non-temperature-controlled areas and constructing sampling sequences, combined with adaptive baselines and LSTM neural network peak recognition models, multi-dimensional capture of temperature deviation events in cold chain goods is achieved. By using differentiated peak thresholds bound to cargo type and scenario, normal fluctuations and abnormal risks in cold storage can be distinguished. The generated intermodal transport warning signals or trend analysis results provide real-time data support for subsequent calculation of cumulative temperature deviations and risk classification response, reducing the risk of cargo damage due to missed temperature deviations and improving the front-end monitoring capabilities of cold chain transportation.
[0047] 2. For the cumulative temperature deviation, which indicates an increasing risk due to peak trends, this study investigates the nonlinear problem of quality degradation caused by cargo thermal inertia or biological reaction lag through inertial accumulation effect analysis and nonlinear correction, making the cumulative temperature deviation closer to the actual physical process. Combined with first-order kinetic reaction verification (and the constraint of maximum quality loss), the constructed safety window boundary provides a theoretical basis for subsequent capacity matching and temperature deviation control.
[0048] 3. By calculating the ratio of peak frequency deviation to cumulative temperature deviation, a multimodal transport risk value is generated, which can dynamically assess whether the current transport risk deviates from expectations. When the risk value exceeds the threshold, the goal is to minimize the cumulative temperature deviation. A matching function is constructed using the cargo temperature sensitivity coefficient, and a reinforcement learning algorithm is used to solve for the optimal capacity allocation scheme. This reduces the cumulative temperature deviation during transport while satisfying the safety window constraint, improving capacity utilization and reducing cargo damage risk. It is suitable for resource collaborative optimization across regions and transport modes in multimodal transport.
[0049] 4. Based on the graded control refrigeration strategy with modified urgency coefficient, the response of temperature deviation control is realized. That is, the refrigeration state is maintained when the risk is low, and the refrigeration intensity is gradually increased when the risk is medium and high, so as to realize the rapid return of temperature deviation to the safe window. Combined with the biological activity detection data after the goods arrive, the parameters of the internal loss dynamic model are calibrated by calculating the attenuation deviation ratio, forming a self-evolving closed loop of monitoring, control, feedback and optimization, so as to realize the dynamic control of the risk of cargo damage in cold chain transportation. Attached Figure Description
[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0051] Figure 1 This is a flowchart of the AI-based multimodal transport collaborative optimization method of the present invention;
[0052] Figure 2 This is a flowchart of the peak recognition model establishment method of the present invention;
[0053] Figure 3 This is a schematic diagram of the AI-based multimodal transport collaborative optimization system of the present invention. Detailed Implementation
[0054] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.
[0055] Example 1: As Figure 1 As shown, the AI-based multimodal transport collaborative optimization method includes the following steps:
[0056] Step 1: In cold chain multimodal transport, obtain the temperature of cold chain cargo corresponding to the non-temperature-controlled area of flight delay and construct a temperature sampling sequence, establish a temperature adaptive baseline and identify the temperature peak factor of the temperature sampling sequence, and perform peak deviation analysis to identify the peak change trend.
[0057] The method for identifying the temperature peak factor in the temperature sampling sequence is as follows:
[0058] Preferably, in cold chain multimodal transport, the temperature of cold chain cargo in non-temperature-controlled areas is obtained by temperature sensors during the monitoring period when there is a sudden flight delay, and a temperature sampling sequence T(i) is constructed.
[0059] Where T(i) represents the actual temperature value collected at time point i in the past monitoring period, and i is the number of the time point;
[0060] It is understandable that if cold chain goods are temporarily stored in non-temperature-controlled areas such as tarmacs or general warehouses due to sudden flight delays, the ambient temperature will continuously deviate from the set value, resulting in a persistent average temperature deviation. In extreme weather conditions, instantaneous temperature peaks may also occur. When goods are exposed to non-temperature-controlled environments for 5–15 minutes during loading and unloading, short-term temperature deviation pulses will occur (e.g., a single exposure accumulating 0.3°C·h). Therefore, it is necessary to use high-frequency sensors to simultaneously collect instantaneous temperature (to capture sudden fluctuations) and average temperature deviation over time (to calculate cumulative effects), combined with circulation scenario tags (cold storage operations, tarmac exposure, etc.), to provide multi-dimensional data support for cargo damage modeling.
[0061] A peak identification model is established for the temperature sampling sequence to extract the temperature peak factor of the temperature sampling sequence.
[0062] like Figure 2 As shown, the method for establishing the spike recognition model is as follows:
[0063] A1. Establish a temperature adaptive baseline;
[0064] The preferred method is based on the temperature sampling sequence T(i), using the formula: Construct a temperature-adaptive baseline B(t);
[0065] Where B(t) represents the adaptive temperature baseline value at time t, W represents the size of the time window for calculating the moving average, and t represents the current time point;
[0066] Understandably, the time window is a pre-defined positive integer that determines how much data is used to calculate the average. For example, if W=60, it means that data from the past 60 time points is used to calculate the current baseline;
[0067] A2. Based on the adaptive baseline, temperature peak factors are identified through a peak identification model.
[0068] A peak identification model is established using a flow cytometry anomaly detection algorithm. An adaptive baseline and temperature sampling sequence are input into the peak identification model to obtain the temperature peak factor.
[0069] Those skilled in the art will understand that the LSTM neural network in the flow cytometry anomaly detection algorithm is used to construct a real-time monitoring model, with the adaptive baseline as the dynamic reference boundary for normal temperature fluctuations and the temperature sampling sequence as the real-time input stream. The algorithm calculates the deviation of real-time data from the baseline, and combines the preset peak thresholds (such as instantaneous temperature peak amplitude and duration) for cargo type and flow scenario to identify temperature peak events that exceed the baseline fluctuation range, and outputs a temperature peak factor containing features such as peak amplitude, duration, and frequency of occurrence.
[0070] The temperature peak factor is compared with the preset peak threshold. If the temperature peak factor is higher than or equal to the preset peak threshold, an intermodal warning signal is generated.
[0071] It should be explained that the preset peak thresholds correspond one-to-one with the type of goods and the flow scenario of the goods;
[0072] If the temperature peak factor is lower than the preset peak threshold, peak deviation analysis is performed to obtain the peak deviation degree and identify the trend of peak change.
[0073] The method for performing peak deviation analysis is as follows:
[0074] The difference ratio between the temperature peak factor and the preset peak threshold is calculated to obtain the peak deviation.
[0075] It should be explained that the method of identifying the peak change trend through trend slope analysis and statistical test within the sliding window is as follows: take the deviation data of the current and previous multiple monitoring periods to form a time series, use linear regression to calculate the trend slope, if the slope is negative and the absolute value exceeds the deviation decrease per unit time ≥ 5%), it is determined that the peak intensity is approaching the risk of an upward trend, triggering the calculation of the cumulative temperature deviation;
[0076] If the slope is positive or close to zero, it is judged as a risk mitigation trend. Regular monitoring is maintained. The trend identification result directly determines whether to enter the subsequent inertial cumulative effect analysis and temperature deviation safety window construction process, so as to realize the graded response and dynamic tracking of risk.
[0077] Step 2: Based on the identified trend, perform inertial cumulative effect analysis on the cumulative amount of temperature deviation with the risk increasing trend, construct an internal loss dynamic model to obtain the preset maximum quality loss, and then construct a safety window boundary function and curve to determine the temperature deviation safety window.
[0078] The method for obtaining the accumulated temperature deviation is as follows:
[0079] If the trend of the peak change is an upward trend of risk, then obtain the cumulative temperature deviation Wp within the current monitoring period;
[0080] Preferably, by formula: Obtain the cumulative temperature deviation Wp within the current monitoring period;
[0081] Where T(t) is the real-time temperature value collected at time point t within the current monitoring period, B(t) is the temperature adaptive baseline value, Δt represents the time interval between two temperature samplings, and n is the number of time points;
[0082] An inertial accumulation effect analysis is performed on the temperature deviation accumulation Wp. If an inertial accumulation effect exists, an inertial accumulation correction is applied to the temperature deviation, and a nonlinear correction function is constructed. ;
[0083] Those skilled in the art will understand that in cold chain transportation, due to thermal inertia or the lag of biochemical reactions, the cumulative temperature deviation Wp and quality loss may exhibit a non-linear relationship. The inertial cumulative effect analysis can use regression analysis to examine the trend of the cumulative temperature deviation Wp over time. If it is found that the growth rate of the cumulative amount accelerates over time, it is determined that there is an inertial effect.
[0084] By analyzing the second derivative of the cumulative temperature deviation with time, if the growth rate follows an exponential trend, a correction function is constructed by introducing a preset activation energy Ea and the deviation ΔT between the current temperature and the reference temperature. , where R is the gas constant;
[0085] By amplifying the cumulative temperature deviation under the inertial effect through the exponential term, the corrected cumulative temperature deviation Wp is made closer to the actual physical process of accelerated quality loss caused by temperature-time coupling. This provides nonlinear input parameters for the subsequent internal loss dynamics model and ensures the accuracy of the safety window boundary function.
[0086] The internal loss dynamics modeling method is as follows:
[0087] Preferably, by formula: Constructing an internal loss dynamics model ;
[0088] in, Used to represent the internal loss rate constant at the current temperature;
[0089] K0 is the rate constant at the reference temperature T0, and Ea is the preset activation energy. The nonlinear correction function is constructed based on the cumulative temperature deviation, where R is the gas constant and ΔT is the deviation between the current temperature and the reference temperature.
[0090] It should be explained that the preset activation energy corresponds one-to-one with the type of goods and the flow scenario of the goods;
[0091] The internal loss kinetic model is validated using first-order kinetics. If the internal loss kinetic model conforms to first-order kinetics, the preset maximum quality loss is obtained. ;
[0092] Understandably, based on the internal loss kinetic model, quality loss data at different time points are collected, and the data is transformed using logarithmic transformation. Linear regression fitting is performed. If the data points conform to a logarithmic linear relationship, then the integral model equation is used. The preset maximum quality loss Qmax is calculated, where k represents the internal loss rate constant at the current temperature;
[0093] Maximum quality loss Used to characterize the theoretical maximum attenuation limit of cargo quality under extreme temperature deviation conditions, and to construct the boundary constraints of the temperature deviation safety window;
[0094] For example, a hardness tester is used to measure the hardness of the pulp as the quality loss. If the cumulative temperature deviation increases by 10°C·h, the hardness decreases by 5~10N. The decrease in hardness is the quality loss.
[0095] Based on maximum quality loss Construct a safety window boundary function and establish a constraint relationship between temperature deviation and time;
[0096] Preferably, the safety window boundary function is: ;
[0097] Among them, the initial quality loss during the loading of the Q0 intermodal transport capacity unit. K0 represents the transport duration of the intermodal transport capacity unit, and K0 is the quality decay rate constant at the reference temperature T0. The temperature sensitivity coefficient is used to reflect the accelerating effect of unit temperature deviation on the decay rate, and is derived from the Arrhenius equation.
[0098] Understandably, the intermodal transport capacity unit is a refrigerated transport vehicle;
[0099] A safety window curve is constructed based on the safety boundary function to determine the temperature deviation safety window;
[0100] It needs to be explained that the safety window curve is constructed based on the safety boundary function and the temperature deviation safety window is determined. Based on parameters such as the maximum quality loss and activation energy, the boundary function is substituted, and a nonlinear curve is plotted with time t as the horizontal axis and the cumulative temperature deviation Wp as the vertical axis. The area below the curve is the safety window.
[0101] The technical solution of this embodiment is as follows: obtain the temperature of cold chain cargo corresponding to the non-temperature-controlled area of flight delay and construct a temperature sampling sequence; establish a temperature adaptive baseline and identify the temperature peak factor of the temperature sampling sequence; perform peak deviation analysis to identify the peak change trend; perform inertial cumulative effect analysis on the cumulative amount of temperature deviation with an upward risk trend; construct an internal loss dynamic model to obtain the preset maximum quality loss; and then construct a safety window boundary function and curve to determine the temperature deviation safety window; the constructed safety window boundary provides a theoretical basis for subsequent capacity matching and temperature deviation control.
[0102] Example 2: Figure 1 As shown, the AI-based multimodal transport collaborative optimization method also includes the following steps:
[0103] Step 3: Conduct intermodal transport risk analysis based on the temperature deviation safety window to obtain intermodal transport risk values. Based on the intermodal transport risk values, establish a matching algorithm with the goal of minimizing the cumulative temperature deviation to match intermodal transport capacity units.
[0104] The method for conducting intermodal transport risk analysis is as follows:
[0105] Obtain the temperature peak factor in the current monitoring period and the previous monitoring period, and calculate the frequency of the temperature peak factor in the current monitoring period and the frequency of the temperature peak factor in the previous monitoring period.
[0106] Calculate the frequency deviation of the temperature peak factor between the current monitoring cycle and the previous monitoring cycle. If the frequency deviation is positive, the frequency deviation will be used as the peak rise factor.
[0107] It should be explained that the preceding monitoring cycle refers to the monitoring cycle that is adjacent to the current monitoring cycle in the time dimension and is earlier in time.
[0108] Calculate the percentage of the deviation between the current temperature deviation accumulation Wp and the temperature deviation safety window to obtain the window deviation ratio;
[0109] The intermodal transport risk value is obtained by summing the window deviation ratio and the peak rise factor.
[0110] It is understandable that the physical meaning of the intermodal transport risk value is a comprehensive risk measure of the changing trend of the frequency of temperature deviation events and the cumulative deviation of temperature deviation from the safety window during the current transportation process.
[0111] Peak rise factor: Reflects the increasing trend of the frequency of temperature peak factors in the current monitoring period compared with the previous period, characterizing the risk of increased frequency of temperature deviation events;
[0112] The window deviation ratio is used to characterize the degree of risk when the cumulative temperature deviation approaches the safety threshold.
[0113] The higher the intermodal risk value, the more significant the deviation of the temperature deviation risk from the expectation, which requires triggering a capacity matching algorithm with the goal of minimizing the cumulative temperature deviation, so as to achieve intelligent risk control and dynamic optimization of capacity resources;
[0114] The intermodal transport risk value is compared with the preset intermodal transport risk threshold. If the intermodal transport risk value is higher than or equal to the preset intermodal transport risk threshold, an intermodal transport scenario analysis is performed, and a matching algorithm is established with the goal of minimizing the cumulative temperature deviation.
[0115] The matching algorithm with the objective of minimizing the cumulative temperature deviation is established as follows:
[0116] Obtain the internal loss rate of cold chain goods per unit temperature deviation, and the internal loss rate at the baseline temperature;
[0117] The internal loss rate per unit temperature deviation is compared with the internal loss rate at the baseline temperature to obtain the cargo temperature sensitivity coefficient. ;
[0118] Through the formula: Construct the matching function L;
[0119] in, The value range is {0,1}, where a is the number of the cold chain goods, b is the number of the intermodal transport unit, and A and B are the total number of cold chain goods and the total number of intermodal transport units, respectively.
[0120] , , , These are, respectively, the cumulative temperature deviation of cargo a during transportation in transport capacity b, the temperature sensitivity coefficient of cargo a, the estimated transportation time of cargo a through intermodal transport capacity unit b, and the temperature compensation rate of intermodal transport capacity unit b;
[0121] The matching function value is used to match the corresponding intermodal transport capacity unit for cold chain goods.
[0122] It should be explained that the theoretical temperature compensation capacity provided by the manufacturer is obtained based on the model of the refrigeration system of the intermodal transport capacity unit;
[0123] When matching intermodal transport capacity based on the matching function value, the matching score of each cargo and the transport capacity is calculated; the optimal allocation scheme is generated by solving the problem through reinforcement learning algorithm in artificial intelligence (AI) with the goal of minimizing the cumulative temperature deviation, while satisfying the constraints of cold storage time and safety window.
[0124] Step 4: Based on the determined intermodal transport capacity unit, perform temperature deviation correction analysis on the intermodal transport capacity unit to determine whether the temperature deviation of the intermodal transport capacity unit meets the requirements. If it does not meet the requirements, adjust the capacity adjustment strategy of the transport unit and calibrate the internal loss dynamic model in combination with the quality loss after arrival.
[0125] The method for performing temperature deviation correction analysis on intermodal transport capacity units is as follows:
[0126] Obtain the cargo sampling temperature of the intermodal transport capacity unit, as well as the preset target temperature;
[0127] The absolute deviation between the sampled temperature of the goods and the preset target temperature is calculated to obtain the absolute value of the temperature deviation. ;
[0128] Through formula one: Get the remaining amount of the safety window ;
[0129] in, For the time points of intermodal transport, Time has been used for intermodal transport;
[0130] Through formula two: Obtain the corrected urgency coefficient Ur;
[0131] Among them, Wp max This is the upper limit of the temperature deviation safety window;
[0132] It is understandable that the urgency coefficient is used to quantify the urgency of temperature deviation risks in cold chain cargo transportation;
[0133] The correction strategy is graded based on the correction urgency coefficient to determine whether the temperature deviation of the intermodal transport capacity unit meets the requirements. The closer the correction urgency coefficient is to 1, the closer the current cumulative temperature deviation is to the upper limit of the safety window, and the higher the urgency of correcting the temperature deviation. This coefficient is used to drive the graded temperature control strategy. When the value is below 0.3, the cooling state is maintained. When it is between 0.3 and 0.7, the first-level correction (increasing the cooling power) is initiated. When it is not below 0.7, the second-level correction (activating the backup cold storage material and executing closed-loop control) is triggered, so as to realize the dynamic quantification and response of temperature deviation risk.
[0134] If it does not meet the requirements, the capacity adjustment strategy of the capacity unit will be adjusted through the proportional integral algorithm.
[0135] Understandably, the correction urgency coefficient is used to classify the levels: when Ur < 0.3, the temperature deviation is considered to meet the requirements, and the current cooling state is maintained; when 0.3 ≤ Ur < 0.7, the first-level correction is initiated, increasing the power of the vehicle cooling system by 20% to 50%; when Ur ≥ 0.7, the temperature deviation is considered to exceed the standard, triggering the second-level correction, simultaneously activating the backup cold storage material and executing PID closed-loop control through edge computing to dynamically adjust the cooling intensity until the temperature deviates back to within the safe window.
[0136] When controlling the thermal strategy using the proportional-integral (PI) algorithm, the real-time temperature deviation error is first calculated. The proportional stage generates an instantaneous control signal based on the current error to quickly respond to changes in temperature deviation. The integral stage accumulates historical errors to reduce the long-term cumulative effect of temperature deviation (e.g., when the error is consistently 0.5℃, the cooling power is gradually increased until the error returns to zero). The two signals are then superimposed and output to the vehicle's cooling system to dynamically adjust the compressor power or fan speed, thereby achieving temperature deviation convergence within a safe window.
[0137] The method for calibrating the internal loss kinetic model, taking into account the quality loss after delivery, is as follows:
[0138] Obtain the actual quality screening quantity and the attenuation quantity predicted by the internal loss kinetic model, calculate the deviation ratio between the actual quality screening quantity and the attenuation quantity predicted by the internal loss kinetic model, and obtain the attenuation deviation ratio.
[0139] A calibration strategy for the internal loss kinetic model is developed based on the attenuation deviation ratio, and the calibration of the internal loss kinetic model is adjusted.
[0140] The technical solution of this embodiment is as follows: Intermodal transport risk analysis is performed based on the temperature deviation safety window to obtain intermodal transport risk values. A matching algorithm is established based on these risk values, with the goal of minimizing the cumulative temperature deviation, to match intermodal transport capacity units. Based on the determined intermodal transport capacity units, temperature deviation correction analysis is performed to determine whether the temperature deviation of the unit meets the requirements. If not, the capacity adjustment strategy of the unit is adjusted. The internal loss kinetic model is calibrated by combining the quality loss amount after arrival. The parameters of the internal loss kinetic model are calibrated by calculating the attenuation deviation ratio, combining the bioactivity detection data after arrival, thus forming a self-evolving closed loop of monitoring, control, feedback, and optimization, achieving dynamic control of the entire cold chain transport cargo damage risk.
[0141] Example 3: Figure 3 As shown, the AI-based multimodal transport collaborative optimization system includes the following modules:
[0142] Trend identification module: In cold chain multimodal transport, the module obtains the temperature of cold chain goods corresponding to the non-temperature-controlled area of flight delays and constructs a temperature sampling sequence. It establishes a temperature adaptive baseline and identifies the temperature peak factor of the temperature sampling sequence, and performs peak deviation analysis to identify the peak change trend.
[0143] Window determination module: Based on the identified trend, the cumulative effect analysis of temperature deviation with the risk increase trend is carried out, the internal loss dynamic model is constructed to obtain the preset maximum quality loss, and then the safety window boundary function and curve are constructed to determine the temperature deviation safety window.
[0144] Intermodal matching module: Based on the temperature deviation safety window, intermodal risk analysis is performed to obtain intermodal risk values. Based on the intermodal risk values, a matching algorithm is established with the goal of minimizing the cumulative temperature deviation to match intermodal transport capacity units.
[0145] Model optimization module: Based on the determined intermodal transport capacity unit, temperature deviation correction analysis is performed on the intermodal transport capacity unit to determine whether the temperature deviation of the intermodal transport capacity unit meets the requirements. If it does not meet the requirements, the capacity adjustment strategy of the transport unit is adjusted. Combined with the quality loss after arrival, the internal loss dynamic model is calibrated.
[0146] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.
Claims
1. A multimodal transport collaborative optimization method and system based on AI, characterized in that, Includes the following steps: The temperature of cold chain goods in non-temperature-controlled areas of flight delays is obtained and a temperature sampling sequence is constructed. A temperature adaptive baseline is established and the temperature peak factor of the temperature sampling sequence is identified. Peak deviation analysis is performed to identify the peak change trend. The method for establishing the temperature adaptive baseline is as follows: Calculate the moving average of the temperature sampling sequence within a preset time window to obtain the temperature adaptive baseline; The method for identifying the temperature peak factor in the temperature sampling sequence is as follows: The temperature adaptive baseline and temperature sampling sequence are input into the spike identification model based on the flow cytometry anomaly detection algorithm to obtain the temperature spike factor; The method for identifying peak variation trends through peak deviation analysis is as follows: The peak deviation is obtained by calculating the percentage difference between the temperature peak factor and the preset peak threshold. The peak deviation of the current and previous monitoring periods is used to form a time series. The trend slope is calculated by linear regression. The peak change trend is determined to be an upward risk trend based on the negative change magnitude of the trend slope. If the risk is determined to be rising, the cumulative temperature deviation is calculated based on the real-time temperature value, the temperature adaptive baseline value, and the temperature sampling time interval within the current monitoring period. An inertial cumulative effect analysis is performed on the cumulative amount of temperature deviation with an upward risk trend. An internal loss dynamic model is constructed to obtain the preset maximum quality loss amount. Then, a safety window boundary function and curve are constructed to determine the safety window of temperature deviation. The method for constructing the internal loss dynamics model is as follows: An inertial accumulation effect analysis is performed on the cumulative temperature deviation, and the trend of the cumulative temperature deviation over time is analyzed. If the growth rate conforms to the exponential growth trend, it is determined that there is an inertial accumulation effect. A nonlinear correction function is constructed by introducing a preset activation energy and the deviation between the current temperature and the reference temperature. The rate constant at the reference temperature is obtained and combined with the nonlinear correction function to construct an internal loss dynamic model. The method for determining the safe window for temperature deviation is as follows: The internal loss dynamics model is validated by first-order kinetic response to obtain the preset maximum quality loss. Based on the preset maximum quality loss, a safety window boundary function is constructed to establish the relationship between temperature deviation and time constraint, and then the temperature deviation safety window is determined. Intermodal transport risk analysis is performed based on the temperature deviation safety window to obtain intermodal transport risk values. Based on the intermodal transport risk values, a matching algorithm is established with the goal of minimizing the cumulative temperature deviation to match intermodal transport capacity units. The method for obtaining the intermodal transport risk value is as follows: The frequency deviation of the temperature peak factor between the current monitoring period and the previous monitoring period is calculated as the peak rise factor; the ratio of the cumulative temperature deviation in the current monitoring period to the temperature deviation safety window is calculated as the window deviation ratio; the window deviation ratio and the peak rise factor are summed to obtain the intermodal transport risk value. The method for matching intermodal transport capacity units is as follows: If the intermodal transport risk value is higher than or equal to the preset intermodal transport risk threshold, a matching algorithm is established that includes the temperature sensitivity coefficient of cold chain goods and aims to minimize the cumulative temperature deviation, and the corresponding intermodal transport capacity unit is matched based on the matching function value. Based on the determined intermodal transport capacity unit, a temperature deviation correction analysis is performed on the intermodal transport capacity unit to determine whether the temperature deviation of the intermodal transport capacity unit meets the requirements. If it does not meet the requirements, the capacity adjustment strategy of the transport capacity unit is adjusted. Combined with the quality loss after arrival, the internal loss dynamics model is calibrated. The method for determining whether the temperature deviation of intermodal transport capacity units meets the requirements is as follows: The absolute deviation between the cargo sampling temperature of the intermodal transport capacity unit and the preset target temperature is calculated. The correction urgency coefficient is obtained by combining the remaining safety window and the cumulative temperature deviation. The correction strategy is graded and determined based on the correction urgency coefficient. If the conditions are not met, the method for adjusting the capacity of the transport unit is as follows: If the temperature deviation is determined to be unacceptable, the power of the vehicle's onboard cooling system is dynamically adjusted based on the proportional-integral algorithm and the graded determination results.
2. The AI-based multimodal transport collaborative optimization method and system according to claim 1, characterized in that, The method for extracting the peak rise factor is as follows: Calculate the frequency deviation of the temperature peak factor between the current monitoring cycle and the previous monitoring cycle. If the frequency deviation is positive, the frequency deviation will be used as the peak rise factor.
3. An AI-based multimodal transport collaborative optimization system, used to implement the AI-based multimodal transport collaborative optimization method according to any one of claims 1-2, characterized in that, Includes the following modules: Trend identification module: In cold chain multimodal transport, the module obtains the temperature of cold chain goods corresponding to the non-temperature-controlled area of flight delays and constructs a temperature sampling sequence. It establishes a temperature adaptive baseline and identifies the temperature peak factor of the temperature sampling sequence, and performs peak deviation analysis to identify the peak change trend. Window determination module: Based on the identified trend, the cumulative effect analysis of temperature deviation with the risk increase trend is carried out, the internal loss dynamic model is constructed to obtain the preset maximum quality loss, and then the safety window boundary function and curve are constructed to determine the temperature deviation safety window. Intermodal matching module: Based on the temperature deviation safety window, intermodal risk analysis is performed to obtain intermodal risk values. Based on the intermodal risk values, a matching algorithm is established with the goal of minimizing the cumulative temperature deviation to match intermodal transport capacity units. Model optimization module: Based on the determined intermodal transport capacity unit, perform temperature deviation correction analysis on the intermodal transport capacity unit to determine whether the temperature deviation of the intermodal transport capacity unit meets the requirements. If it does not meet the requirements, adjust the capacity adjustment strategy of the transport unit. Combined with the quality loss after arrival, calibrate the internal loss dynamic model.