Medical cold chain logistics carrying robot and control system thereof

Abstract

This invention discloses a medical cold chain logistics handling robot and its control system, relating to the field of medical technology. The system includes: a state perception module, used to acquire raw microscopic state signals of medical supplies in real time through an optical sensor array; based on the raw microscopic state signals, determine whether there is a risk of low-temperature phase change or structural deformation, and generate corresponding risk identification signals; a linkage execution module, used to control temperature actuators and mechanical actuators according to the risk identification signals; a dynamic analysis module, used to establish an evaluation model of temperature deviation and mechanical vibration based on the type of medical supplies; input real-time values ​​into the evaluation model and output a comprehensive risk index; when the comprehensive risk index exceeds a warning threshold, generate a reverse instruction and output a dominant risk type identifier; and a reverse execution module, used to respond to the reverse instruction and control the temperature actuators and mechanical actuators according to the dominant risk type identifier.

Description

Technical Field

[0001] This application relates to the field of medical technology, and more specifically to the field of medical cold chain logistics. In particular, it relates to a medical cold chain logistics handling robot and its control system. Background Technology

[0002] Medical cold chain logistics handling robots, as key equipment connecting the storage and transportation of medical supplies, belong to the field of medical cold chain logistics. Through automated handling operations, they can reduce the risk of temperature fluctuations caused by human intervention, ensuring the quality stability of medical supplies such as vaccines and biological agents that are sensitive to the storage environment during the circulation process. This is of great significance for improving the efficiency of the medical supply chain and reducing losses.

[0003] In existing technologies, the control systems of medical cold chain logistics handling robots mostly focus on basic path planning and temperature maintenance functions, lacking the ability to accurately perceive subtle changes in the condition of medical supplies that may occur during handling. Furthermore, in terms of risk assessment, they often consider only single factors such as temperature or vibration, making it difficult to comprehensively assess the potential risks faced by the supplies, resulting in insufficient timeliness and specificity of response measures.

[0004] There is currently no effective solution to the above problems. Summary of the Invention

[0005] This application provides a medical cold chain logistics handling robot and its control system to solve the above-mentioned technical problems.

[0006] This application provides a medical cold chain logistics handling robot, comprising: a robot mobile platform; a transport compartment disposed on the robot mobile platform for carrying medical supplies; a temperature actuator integrated in the transport compartment for regulating the temperature environment inside the compartment; a mechanical actuator including a drive unit and a shock absorption unit, the mechanical actuator being installed at the bottom of the robot mobile platform; and an optical sensor array disposed on the inner wall of the transport compartment for real-time acquisition of raw signals of the microscopic state of the surface of the medical supplies.

[0007] This application provides a control system for a medical cold chain logistics handling robot, comprising: a state perception module, used to acquire raw microscopic state signals of medical supplies in real time through an optical sensor array; determine whether there is a risk of low-temperature phase change or structural deformation based on the raw microscopic state signals, and generate corresponding risk identification signals; a linkage execution module, used to control a temperature actuator and a mechanical actuator based on the risk identification signals; a dynamic analysis module, used to establish an evaluation model of temperature deviation and mechanical vibration based on the type of medical supplies; calculate the real-time values ​​of temperature deviation and mechanical vibration based on the raw microscopic state signals; input the real-time values ​​into the evaluation model and output a comprehensive risk index; when the comprehensive risk index exceeds a warning threshold, generate a reverse instruction and output a dominant risk type identifier; and a reverse execution module, used to respond to the reverse instruction and control the temperature actuator and the mechanical actuator based on the dominant risk type identifier.

[0008] Based on the embodiments provided in this application, the state perception module utilizes an optical sensor array to acquire the original microscopic state signals of medical supplies in real time, and judges the risk of low-temperature phase transition or structural deformation accordingly. This enables more accurate capture of subtle state changes of medical supplies during handling, providing a reliable basis for subsequent risk response and helping to ensure the quality stability of medical supplies. The dynamic analysis module establishes an assessment model for temperature deviation and mechanical vibration based on the type of medical supplies. By comprehensively calculating the real-time values ​​of both, a comprehensive risk index is obtained, overcoming the limitations of single-factor assessment and making risk judgment more comprehensive. At the same time, combined with the comprehensive risk index and the dominant risk type identifier, the reverse execution module can specifically control the temperature actuator and mechanical actuator, improving the accuracy and effectiveness of risk response and further adapting to the high requirements for material safety in medical cold chain logistics. Attached Figure Description

[0009] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0010] Figure 1 This is a structural diagram of a control system for an optional medical cold chain logistics handling robot according to an embodiment of this application;

[0011] Figure 2 This is a flowchart of an optional method for acquiring and analyzing raw signals of microscopic states according to an embodiment of this application;

[0012] Figure 3 This is a flowchart illustrating the optimization process performed by an optional feedback optimization module according to an embodiment of this application.

[0013] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0014] 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 a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0015] According to one aspect of the embodiments of this application, this application provides a medical cold chain logistics handling robot, comprising:

[0016] Robot mobile platform;

[0017] The transport compartment, set on the robot's mobile platform, is used to carry medical supplies.

[0018] A temperature actuator, integrated into the transport compartment, is used to regulate the temperature environment inside the compartment;

[0019] The mechanical actuator, including a drive unit and a shock absorption unit, is mounted on the bottom of the robot's mobile platform;

[0020] An optical sensor array is deployed on the inner wall of the transport compartment to collect raw signals of the microscopic state of the surface of medical supplies in real time.

[0021] According to another aspect of the embodiments of this application, a control system for a medical cold chain logistics handling robot is also provided. For example... Figure 1 As shown, the system includes:

[0022] The state perception module 101 is used to acquire the original microscopic state signals of medical supplies in real time through an optical sensor array; based on the original microscopic state signals, it determines whether there is a risk of low-temperature phase transition or structural deformation, and generates a corresponding risk identification signal.

[0023] Among them, the original signals of the microscopic state refer to the set of signals that directly reflect the microscopic physical state of medical supplies (such as vaccine vials and biological agent packaging) collected by optical sensor arrays (such as hyperspectral sensors and laser interferometers). Specifically, they can include spectral reflection signals from the surface of the material, minute displacement vibration signals, and temperature field distribution signals. For example, for freeze-dried vaccines, the spectral reflection changes caused by ice crystal formation on the surface, or the interference signals caused by minute deformations of the packaging bottle wall due to vibration, both belong to this type of original signal.

[0024] Low-temperature phase transition risk refers to the risk, in medical cold chain scenarios, that medical supplies may undergo phase transitions such as crystallization or solidification due to the ambient temperature being below their critical phase transition temperature (e.g., the freezing point of blood products or the glass transition temperature of protein preparations), thereby destroying their activity or structure. For example, if the active proteins in a vaccine form ice crystals at excessively low temperatures, these crystals can puncture the protein molecular structure, causing the vaccine to become ineffective; this is a typical example of low-temperature phase transition risk.

[0025] Structural deformation risk refers to the risk that medical supplies (including packaging) may experience minor structural deformations beyond safe limits due to mechanical vibration, temperature stress, etc. (such as micro-cracks in ampoules or wrinkles in aluminum-plastic packaging), which may lead to decreased sealing or damage to internal components due to compression. For example, micro-deformation of the tube wall caused by continuous vibration in biological sample tubes may lead to sample leakage or internal pressure imbalance.

[0026] The linkage execution module 102 is used to control the temperature actuator and the mechanical actuator according to the risk identification signal;

[0027] The dynamic analysis module 103 is used to establish an assessment model for temperature deviation and mechanical vibration based on the type of medical supplies; calculate the real-time values ​​of temperature deviation and mechanical vibration based on the original signals of the microscopic state; input the real-time values ​​into the assessment model and output a comprehensive risk index; when the comprehensive risk index exceeds the warning threshold, generate a reverse instruction and output the dominant risk type identifier.

[0028] Temperature deviation is used to quantify the degree of deviation between the ambient temperature of medical supplies and the preset safe temperature. It includes not only instantaneous temperature differences (such as the difference between the current temperature and the target temperature) but also the cumulative effect of deviation (such as the longer the duration of deviation, the higher the deviation). For example, if a vaccine needs to be maintained at 2-8℃, and the actual temperature is at 1℃ for 10 consecutive minutes, its temperature deviation should include both the "temperature difference of -1℃" and the "10-minute duration weight".

[0029] Mechanical vibration intensity is used to quantify the impact of mechanical vibration on medical supplies, taking into account factors such as vibration intensity (e.g., magnitude of acceleration), frequency (e.g., whether it falls within the sensitive frequency range of the supplies, such as biological agents being sensitive to vibrations in the 20-50Hz range), and duration. For example, the mechanical vibration intensity of high-frequency vibrations generated by a robot's sudden stop during transportation needs to be calculated by combining the effective value of vibration acceleration and the proportion of high-frequency components.

[0030] The assessment model is a pre-built mathematical model (such as a fuzzy logic model or a neural network model) based on the type of medical supplies (e.g., vaccines, blood, biological samples) used to integrate temperature deviation and mechanical vibration to calculate comprehensive risk. This model needs to include risk sensitivity coefficients for different supplies (e.g., blood is more sensitive to vibration, so mechanical vibration has a higher weight in the model).

[0031] The comprehensive risk index is a numerical value (e.g., a range of 0-100) that quantifies the overall risk currently faced by medical supplies, output by the evaluation model. A higher value indicates a higher risk. For example, the temperature deviation of a certain biological agent corresponds to a risk value of 30, and the mechanical vibration corresponds to a risk value of 40. After model fusion, the comprehensive risk index is 50 (considering the synergistic effect of the two).

[0032] The alert threshold is a pre-set critical value of a comprehensive risk index based on the risk tolerance of medical supplies (e.g., a lower threshold for highly sensitive live vaccines and a higher threshold for highly stable freeze-dried preparations) and the transportation stage (e.g., the threshold is stricter at the outbound stage than at the storage stage). When the comprehensive risk index exceeds this value, reverse intervention must be initiated. For example, if the alert threshold for live vaccines is set at 40, a reverse instruction is triggered when the comprehensive risk index reaches 45.

[0033] Reverse commands differ from conventional linkage control commands. They refer to commands used to proactively adjust the robot's operating state to reduce risk when the overall risk index exceeds a warning threshold (such as changing the transport path or adjusting the temperature control strategy). They prioritize risk mitigation over efficiency. For example, when the overall risk index is too high, a reverse command might require the robot to suspend regular transport and return to the insulated compartment to adjust the environment.

[0034] The dominant risk type identifier is a label used to identify the risk type (temperature risk or vibration risk) that plays a major role in the overall risk (e.g., "T" indicates temperature-dominant, "V" indicates vibration-dominant), providing control direction for the reverse execution module. For example, if 60% of the overall risk index originates from temperature deviation and 40% from vibration, it is identified as "T," guiding the reverse execution module to prioritize adjusting the temperature-controlled actuator.

[0035] The reverse execution module 104 is used to respond to reverse instructions and identify the control temperature actuator and mechanical actuator according to the dominant risk type.

[0036] refer to Figure 1 The medical cold chain logistics handling robot of this application belongs to the field of medical cold chain logistics equipment.

[0037] Furthermore, the risk identification signals include low-temperature phase change risk signals and / or structural deformation risk signals; the temperature actuator includes a semiconductor cooling chip array and a pulsed refrigerant injection device; the mechanical actuator includes an electromagnetic levitation device, a hydraulic shock absorption device, and a wheel drive module;

[0038] The linkage execution module controls the temperature actuator and the mechanical actuator based on the risk identification signal, and is configured as follows:

[0039] When a low-temperature phase change risk signal is detected, the pulse-type refrigerant discrete supply of the temperature actuator is activated, and the electromagnetic levitation transport mode of the mechanical actuator is switched.

[0040] Pulsed refrigerant discrete replenishment refers to the intermittent, metered injection of refrigerant (such as liquid nitrogen mist) by a temperature actuator (e.g., a pulsed refrigerant injection device), rather than continuous injection. This aims to avoid exacerbating phase changes in materials due to sudden localized temperature drops. For example, when signs of ice crystal formation are detected in a vaccine, a small dose of refrigerant is injected every 2 seconds. Each dose is calculated based on the current temperature difference, inhibiting ice crystal growth without causing excessive cooling.

[0041] Electromagnetic levitation transport mode refers to a transport mode in which the mechanical actuator generates electromagnetic force through an electromagnetic levitation device, so that the medical supplies (or their carrying trays) are suspended at a preset height (e.g., 5mm) without physical contact with the robot's contact surface, thereby eliminating mechanical friction and vibration. This mode is suitable for vibration-sensitive materials (e.g., cell culture dishes).

[0042] When a structural deformation risk signal is detected, the temperature actuator's constant temperature locking mechanism is activated, and the hydraulic damping device and speed limiter of the mechanical actuator are started.

[0043] The isothermal lock-in mechanism refers to a temperature actuator (such as a semiconductor cooling array) using closed-loop control to stabilize the temperature of the area where the material is located at a fixed value (such as 5°C), and limit temperature fluctuations to a very small range (such as ±0.5°C), in order to avoid thermal stress caused by temperature changes from exacerbating structural deformation. For example, when microcracks are detected in glass-bottled biological agents, the temperature is locked at 4°C to prevent temperature fluctuations from causing changes in the pressure inside the bottle.

[0044] The speed limiter is a control component integrated into the wheel drive module. It is used to limit the robot's movement speed (e.g., from 1m / s to 0.3m / s) to reduce vibrations caused by rapid acceleration and deceleration due to excessive speed. It works with the hydraulic damping device to reduce the overall vibration intensity.

[0045] Based on the embodiments provided in this application, for the risk of low-temperature phase change, pulsed refrigerant discrete replenishment can avoid localized overcooling caused by continuous refrigeration. Simultaneously, switching to electromagnetic levitation transport mode can reduce additional vibration caused by mechanical contact, preventing vibration from exacerbating ice crystal growth. For the risk of structural deformation, activating the constant temperature locking mechanism can stabilize the temperature environment of the materials, avoiding the superposition of thermal stress caused by temperature fluctuations. Furthermore, the coordinated activation of the hydraulic shock absorber and speed limiter can directly reduce the impact of mechanical vibration on the material structure. This allows the control system to mobilize the most suitable hardware resources when facing different risks, improving the stability of medical supplies during handling.

[0046] Furthermore, the reverse execution module identifies the temperature actuator and the mechanical actuator based on the dominant risk type and is configured as follows:

[0047] If the dominant risk type is identified as temperature risk, plan the insulated transport route and control the temperature actuator to adjust the cabin temperature according to the stepped cooling curve.

[0048] In this context, insulated aisle transport routes refer to the routes that robots prioritize in cold chain warehouses or transport scenarios, choosing those with stable ambient temperatures and good insulation (such as passing through temperature-controlled corridors and avoiding temperature fluctuation areas near cold storage doors) to minimize external environmental interference with the internal temperature. For example, if the standard deviation of temperature fluctuation for the "Area A - Transfer Station - Cold Storage" route in a warehouse is ±1℃, while that for the "Area B - Elevator - Cold Storage" route is ±3℃, then the former will be the preferred insulated aisle transport route.

[0049] In a stepped cooling curve, the target cooling process is divided into multiple continuous temperature stages (e.g., from 10℃→8℃→6℃→4℃), with each stage maintained for a preset time (e.g., 3 minutes per stage). This ensures a uniform decrease in the internal temperature of the medical supplies, avoiding the stress caused by the temperature difference between the inside and outside of the device due to rapid cooling. For example, when cooling plasma, the temperature is first reduced from 15℃ to 10℃ and maintained for 2 minutes, then reduced to 5℃ and maintained for 3 minutes, finally stabilizing at 2℃.

[0050] If the dominant risk type is identified as vibration risk, switch the mechanical actuator to step-by-step movement mode and trigger the refrigerant metering compensation device.

[0051] Among them, the stepping movement mode refers to the mechanical actuator operating in a cyclical manner of "short distance movement + brief pause" (such as moving 5cm and pausing for 1 second). By discretizing the movement, the accumulation of vibration energy is reduced, which is suitable for materials that are sensitive to continuous vibration (such as unsealed liquid reagents).

[0052] The refrigerant metering compensation device is a component linked to the temperature actuator. It can calculate and replenish the refrigerant dosage just enough to offset the temperature fluctuation based on the temperature changes inside the cabin during the pause (such as a slight temperature rise caused by the pause in movement). For example, it can replenish 5 ml of refrigerant when the temperature rises by 0.5°C, ensuring that the temperature remains stable even when vibration risk is dominant.

[0053] Based on the embodiments provided in this application, differentiated control is implemented according to the dominant risk type, making reverse intervention more aligned with the nature of the risk. When temperature risk is dominant, planning the insulated passageway transportation route can reduce the interference of the external environment on the cabin temperature, while the stepped cooling curve can adapt to the heat capacity and phase change characteristics of medical supplies, avoiding internal stress caused by rapid cooling. When vibration risk is dominant, the step-by-step movement mode reduces the cumulative effect of vibration through discrete movement, while triggering the refrigerant metering compensation device can offset temperature fluctuations during movement. This accurate control based on the dominant risk type ensures that reverse operation effectively suppresses the core risk while also taking into account the coordinated control of secondary risks, improving the safety of supplies under high-risk conditions.

[0054] Furthermore, the control system of the medical cold chain logistics handling robot also includes:

[0055] The feedback optimization module is used to collect a new round of original micro-state signals after the reverse execution operation is completed; update the comprehensive risk index based on the new round of original micro-state signals; compare the updated comprehensive risk index with the comprehensive risk index when the reverse instruction was generated; and update the model parameters of the evaluation model based on the comparison results.

[0056] The new round of original microscopic state signals refers to the microscopic state signals of medical supplies re-collected by the state perception module after the reverse operation (such as adjusting temperature or changing movement mode) is completed. These signals are used to evaluate the actual effect of the reverse operation. For example, after reverse-execution of step-cooling, the re-collected surface spectral signal of the vaccine (to determine whether ice crystals have decreased) and vibration interference signal (to determine whether deformation has been alleviated).

[0057] The model parameters of the evaluation model are adjustable parameters that constitute the evaluation model. These include weighting coefficients for temperature deviation (e.g., 0.6 for temperature and 0.4 for vibration in vaccines), fusion coefficients for risk index calculation (e.g., the synergistic effect coefficient of temperature and vibration), and trigger thresholds in the rule base (e.g., how much temperature deviation triggers a high-risk rule). For example, if feedback reveals that a certain type of formulation is more sensitive to vibration, the temperature weight can be reduced and the vibration weight increased.

[0058] The specific workflow of the feedback optimization module is to compare the comprehensive risk index before and after the reverse operation (e.g., the index is 60 before the operation and 30 after the operation, the risk is reduced by 50%), determine the deviation of the current evaluation model (e.g., the original model overestimated the impact of temperature), and then adjust the model parameters accordingly (e.g., reduce the weight of temperature deviation) so that the subsequent evaluation is more in line with the actual risk.

[0059] Based on the embodiments provided in this application, by collecting the microscopic state signals after reverse execution and comparing the comprehensive risk index before and after the update, the actual effect of the reverse operation can be intuitively evaluated. Updating the evaluation model parameters based on this effect allows the model to gradually align with the risk response characteristics of specific medical supplies. For example, considering the differences in phase transition sensitivity among different vaccines, the model can optimize and adjust the weighting of temperature and vibration through feedback. This dynamic optimization mechanism avoids the static nature of the evaluation model, enabling it to continuously improve the accuracy of risk assessment and enhance the adaptability of the control system in long-term applications.

[0060] Furthermore, such as Figure 2 As shown, the state-aware module acquires and analyzes the original signals of the microscopic state, and is configured as follows:

[0061] S201, synchronously triggers the hyperspectral imaging unit and the laser interferometry measurement unit to acquire hyperspectral image sequences of surface reflections of medical supplies and surface micro-displacement interference fringe patterns formed by laser interference, respectively.

[0062] S202, for hyperspectral image sequences, an ice crystal feature extraction algorithm is used for processing; the ice crystal feature extraction algorithm focuses on specific ice crystal indicator bands, and calculates the rate of change of branch length of ice crystal contours in adjacent frames of images in this band as the ice crystal growth rate to capture early signs of phase change; when the ice crystal growth rate continuously exceeds the morphological evolution threshold preset based on the phase change characteristics of materials, it is determined that there is a risk of low-temperature phase change;

[0063] The ice crystal feature extraction algorithm includes processing logic for identifying and quantifying features related to ice crystal formation in hyperspectral image sequences. Its core is to capture early phase transition signals by focusing on the specific spectral response of ice crystals. Specific processing steps include:

[0064] First, the hyperspectral image sequence is preprocessed (e.g., noise reduction, band selection) to retain effective bands related to ice crystal formation. Then, image segmentation algorithms (e.g., thresholding, edge detection) are used to extract the contours of suspected ice crystal regions from images in specific bands. Morphological analysis (e.g., number of branches, length, growth direction) is performed on the extracted contours to calculate the dynamic changes of the contours between adjacent frames, which are finally quantified as the ice crystal growth rate.

[0065] In some embodiments, a specific ice crystal indicator band refers to a band that has a specific spectral response to ice crystal formation, where the spectral reflectance changes significantly with increasing ice crystal content (e.g., increased reflectance or the appearance of a specific absorption peak). In medical cold chain scenarios, since the main component of medical supplies (such as vaccines and blood) contains water, and the difference in spectral characteristics between water and ice crystals is concentrated in the near-infrared band, the specific ice crystal indicator band is typically selected at 1.5-1.8 μm (near the absorption peak of water) or 2.0-2.5 μm (the scattering enhancement band of ice crystals).

[0066] For example, for a water-containing vaccine formulation, when the temperature approaches the freezing point, ice crystals begin to form, and their reflectivity in the 1.65 μm band increases from 0.3 (no ice crystals) to 0.5 (a small amount of ice crystals). This band serves as the specific ice crystal indicator band for the vaccine.

[0067] The rate of change of branch length of ice crystal profiles in this band between adjacent frames refers to the dynamic index of ice crystal growth calculated through the following steps for a specific ice crystal indicator band in a continuously acquired hyperspectral image sequence:

[0068] In the nth frame image, extract all branches of the ice crystal contour, calculate the length of each branch and sum them to obtain the total branch length L. n ;

[0069] In the (n+1)th frame of the image, the total branch length L is obtained using the same method. n+1 ;

[0070] Branch length change rate = (L n+1 -L n ) / Δt (Δt is the acquisition interval between two frames of images, such as 0.5 seconds), used to characterize the growth rate of ice crystals.

[0071] For example, if the total branch length of ice crystals in a certain vaccine is 0.8 mm in the 10th frame and 1.0 mm in the 11th frame (0.5 seconds interval), then the branch length change rate is (1.0-0.8) / 0.5 = 0.4 mm / s, which intuitively reflects the growth trend of ice crystals.

[0072] In one implementation, the expression for the ice crystal growth rate is:

[0073]

[0074] Among them, V vice ΔL represents the ice crystal growth rate, expressed in mm / s, characterizing the average growth length of the ice crystal profile on the surface of medical supplies per unit time, directly reflecting the dynamic changes in the risk of low-temperature phase transition; m represents the number of specific ice crystal indicator bands used in the calculation, such as selecting three bands (1.5μm, 1.65μm, and 2.0μm) (m=3), reducing single-signal errors through multi-band fusion; ΔL i ω represents the difference in ice crystal profile branch lengths (in mm) between two adjacent hyperspectral images in the i-th band, extracted from the hyperspectral image using an image segmentation algorithm (such as edge detection); Δt is the acquisition time interval (in seconds) between two adjacent images, determined by the sampling frequency of the optical sensor array (e.g., 20 frames / second corresponds to Δt = 0.05s), ensuring accuracy in the time dimension; iLet ω be the sensitivity weighting coefficient for the i-th band (value 1-2). Based on experimental calibration, the 1.65μm band has the highest accuracy in identifying ice crystals (ω). i =2), the sensitivity is lower in the 1.5μm and 2.0μm bands (ω). i =1), highlighting the contribution of the effective band.

[0075] It should be noted that early phase transition signs refer to microscopic changes in medical materials before an irreversible phase transition (such as the formation of numerous ice crystals or protein coagulation). At this stage, the phase transition has not yet caused substantial damage to the activity or structure of the material, but detectable signal characteristics have already appeared. In this scenario, early phase transition signs specifically manifest as: slight fluctuations in the spectral reflectance of a specific ice crystal indicator band, a slow increase in the branch length of the ice crystal profile (but without forming a continuous ice layer), or the appearance of sporadic ice crystal nuclei in localized areas. For example, when the temperature of a freeze-dried vaccine drops to -2°C, no visible ice crystals have yet formed, but hyperspectral imaging can detect fluctuations in the reflectance of the 1.65μm band starting from a stable value, and the rate of change in the branch length of the ice crystal profile between adjacent frames increases from 0 to 0.1 mm / s, which is an early phase transition sign.

[0076] The morphological evolution threshold refers to a pre-defined critical value for the ice crystal growth rate based on the phase transition sensitivity characteristics of medical supplies (such as protein stability, water content, and glass transition temperature). Exceeding this value indicates that ice crystal growth has entered an accelerated phase, which may trigger an irreversible phase transition. Its value is strongly correlated with the type of medical supply.

[0077] Example 1: For freeze-dried vaccines (containing a small amount of free water, with moderate ice crystal sensitivity), the morphological evolution threshold can be set to 0.2 mm / s (i.e., an early warning is issued when the branch length change rate exceeds 0.2 mm / s);

[0078] Example 2: For blood products (containing a large amount of free water, ice crystals can easily puncture red blood cells), the morphological evolution threshold should be set to a lower 0.1 mm / s.

[0079] The duration of "continuity" refers to the time during which the ice crystal growth rate continuously exceeds the morphological evolution threshold. A reasonable duration needs to be set to eliminate occasional noise interference. In medical scenarios, considering the sampling frequency of optical sensors (e.g., 10 frames / second), "continuity" is usually defined as three consecutive sampling cycles (i.e., 0.3 seconds) to avoid misjudgment based on a single fluctuation and to ensure that early risks are not delayed.

[0080] S203, for the surface micro-displacement interference fringe pattern, the phase unwrapping algorithm is used to process it in order to reconstruct the three-dimensional micro-deformation field of the material surface;

[0081] The phase unwrapping algorithm refers to an algorithm used to process laser interference fringe patterns. Its core is to eliminate the phase ambiguity of the interference fringes (because the periodicity of the interference fringes causes the phase values ​​to be "wrapped" within the 0-2π range), reconstructing the actual phase distribution on the material surface, and then converting it into a three-dimensional micro-displacement. Specific steps include:

[0082] Denoising the interference fringe pattern (e.g., smoothing filtering) and enhancing the fringe edges;

[0083] The phase difference between adjacent pixels is calculated step by step from the reference point using a path tracing method (such as the branch cutting method) to eliminate phase jumps.

[0084] The unwrapped phase value is converted into the actual micro-displacement of the material surface through the wavelength-phase relationship (phase difference = 4πΔd / λ, where Δd is displacement and λ is laser wavelength), and finally the three-dimensional micro-deformation field is reconstructed.

[0085] The algorithm is based on classical interferometry theory and can be implemented by those skilled in the art using existing numerical calculation tools (such as the phase unwrapping function in MATLAB).

[0086] S204 compares the reconstructed three-dimensional micro-deformation field with the known elastic modulus database of the material to identify local micro-strain concentration areas and calculate their proximity to the material's yield limit; when the proximity exceeds the preset safety margin threshold, it is determined that there is a risk of structural deformation.

[0087] The database of known elastic moduli for medical supplies refers to a pre-built database that stores the elastic moduli of common medical supplies and packaging materials (such as glass ampoules, plastic vials, aluminum-plastic composite films, and rubber stoppers). Elastic modulus is a parameter (unit: GPa) characterizing a material's resistance to deformation; the higher the value, the less easily the material deforms. For example, the database includes: glass (70-80 GPa), medical-grade polypropylene (1.5-2.0 GPa), and butyl rubber (0.001-0.01 GPa). During testing, by accessing this database, the elastic modulus of the transported material (such as the glass material of vaccine vials) can be quickly obtained, providing a basis for determining whether micro-deformation exceeds the safe range.

[0088] Localized micro-strain concentration areas refer to micro-strain (unit: με, 1με=10) that appears on the surface of medical supplies due to vibration, temperature stress, etc. -6 The area with a significantly higher risk of structural deformation is a localized location that is significantly higher than the surrounding area (such as the neck of the ampoule or the corner of the packaging).

[0089] The proximity of a localized microstrain concentration region to the material's yield strength refers to the ratio of the actual microstrain value of the localized microstrain concentration region to the material's yield strength (the critical microstrain at which the material undergoes plastic deformation). It is used to quantify the urgency of the deformation risk. The calculation method is: Proximity = Actual microstrain value / Material yield strength (the ratio ranges from 0 to 1; the closer to 1, the closer to plastic deformation).

[0090] For example, if the yield strength of a glass ampoule is 1000 με, and the actual micro-strain in the micro-strain concentration area of ​​its neck is detected to be 800 με, then the closeness = 800 / 1000 = 0.8, indicating that the area is close to the critical state of plastic deformation.

[0091] The safety margin threshold refers to a preset proximity threshold. When the proximity of a localized area of ​​concentrated micro-strain exceeds this value, a structural deformation risk is identified. Its value is related to the application scenario of the material (e.g., materials with high sealing requirements require a stricter threshold). For example, for glass ampoules (where deformation could lead to breakage or leakage), the safety margin threshold can be set to 0.7 (i.e., proximity ≤ 0.7 is safe, exceeding it triggers a warning); for aluminum-plastic packaged tablets (which have good toughness, and slight deformation does not affect sealing), the safety margin threshold can be set to 0.8.

[0092] S205 generates the corresponding low-temperature phase transition risk signal or structural deformation risk signal based on the determination results of the low-temperature phase transition risk or structural deformation risk.

[0093] Based on the embodiments provided in this application, the accurate capture of the microscopic state of medical supplies is achieved through a combination of hyperspectral imaging and laser interferometry. Hyperspectral imaging focuses on specific ice crystal indicator bands, and combined with the calculation of the change rate of ice crystal profile branch length, it can capture early signs of phase transition before ice crystals have formed significantly, overcoming the limitation of traditional temperature sensors that can only detect ambient temperature. Laser interferometry, combined with a phase unwrapping algorithm, can reconstruct the three-dimensional micro-deformation field on the surface of the material. By comparing it with an elastic modulus database, the proximity of local micro-strain to the material's yield limit can be quantified, enabling microscopic prediction of structural deformation risks. This early, microscopic detection capability based on multi-physics field signals provides more sufficient response time for risk intervention, preventing risks from evolving from subtle states into substantial damage.

[0094] Furthermore, upon receiving a low-temperature phase transition risk signal, the linkage execution module performs the following coordinated operations:

[0095] Based on the real-time calculation of the ice crystal growth rate by the state perception module, the control parameters for pulsed refrigerant discrete replenishment are dynamically calculated and output. The pulse frequency is positively correlated with the ice crystal growth rate, and the single injection dose is compensated according to the instantaneous temperature difference of the target area. For example, the frequency is appropriately reduced when the descent rate is fast, and the frequency is increased when the rate is slow or when there is a risk of rebound.

[0096] In addition to pulse frequency and single-spray dose, the control parameters also include: Spray direction: Based on the concentrated ice crystal area located by hyperspectral image (such as the top of the vaccine vial), adjust the angle of the spray device (such as 30°, 45°) to ensure that the refrigerant acts accurately on the target area and avoid overcooling of unnecessary areas; Spray pressure: Set according to the type of material packaging (such as 0.2MPa low-pressure spray for glass bottle materials to avoid impact damage; 0.3MPa medium-pressure spray for plastic packaging to enhance refrigerant diffusion); Interval period: In addition to pulse frequency, set a minimum interval time (such as ≥1 second) to prevent local temperature drop caused by multiple sprays in a short period of time and adapt to the sensitivity of biological agents to temperature fluctuations.

[0097] The target area refers to the key areas in medical supplies where early ice crystal formation has been detected (such as the space above the liquid surface of vaccine vials, or near the wall of biological sample tubes), rather than the entire robot cabin, to ensure that the refrigerant acts on the source of risk.

[0098] Instantaneous temperature difference refers to the real-time difference between the current actual temperature of the target area and the preset "safe phase transition temperature" (such as the 0℃ critical temperature of a vaccine) of that area (which can be positive or negative). For example, if the current temperature of the target area is -1℃ and the safe phase transition temperature is 2℃, then the instantaneous temperature difference is -3℃ (below the safe temperature).

[0099] The compensation calculation for a single spray dose aims to "offset the instantaneous temperature difference and inhibit ice crystal growth." Specifically, it involves: setting a base dose (e.g., the standard compensation dose D0 for a 1°C temperature difference); calculating the dose coefficient based on the instantaneous temperature difference ΔT (taking the absolute value): if ΔT = 1°C, the coefficient is 1.0; if ΔT = 2°C, the coefficient is 1.5 (because the larger the temperature difference, the stronger the driving force for ice crystal growth, requiring over-compensation); single spray dose = base dose D0 × dose coefficient. For example, if the safe phase transition temperature of a certain vaccine is 2°C, the instantaneous temperature difference in the target area is -2°C (current temperature 0°C), and the base dose D0 = 5ml, then the single spray dose = 5ml × 1.5 = 7.5ml, thus inhibiting the expansion of ice crystals from 0°C to the -2°C region through precise compensation.

[0100] In electromagnetic levitation transport mode, by reading data from multiple levitation gap sensors in real time, an adaptive sliding mode control algorithm is used to dynamically adjust the excitation current of each independent electromagnet, so as to maintain the levitation height at the set reference value under the change of transport load and suppress gap fluctuation within the preset allowable range.

[0101] Among them, the adaptive sliding mode control algorithm is a robust control strategy for electromagnetic levitation systems under load changes (such as medical supplies of different weights) and external disturbances (such as vibrations caused by uneven ground). Its core logic is as follows:

[0102] Using the deviation between the suspension height and the set reference value as the core (deviation = actual height - reference value), the control law makes the deviation converge to 0 quickly;

[0103] The system collects height data from multiple suspension gap sensors in real time (e.g., one sensor at each of the four corners of the tray). If the actual height is lower than the reference value due to increased load (e.g., adding a box of vaccines), the algorithm automatically increases the electromagnet excitation current (e.g., from 2A to 2.5A), and vice versa, to ensure stability under different loads.

[0104] By introducing "switching gain", the height fluctuations caused by sudden vibrations (such as the impact when the robot crosses a threshold) can be quickly offset, thus avoiding the collapse of the suspension state.

[0105] The set reference value refers to the target levitation height that the electromagnetic levitation device needs to maintain, which is set according to the vibration sensitivity of the medical supplies. For example, for cell culture dishes that are extremely sensitive to vibration (slight vibration may cause cell rupture), the set reference value is 8mm (to increase the levitation gap and reduce the transmission of ground vibration); for example, for freeze-dried vaccines with better stability (more robust packaging), the set reference value is 5mm (to balance stability and the utilization of the internal space).

[0106] The preset allowable range refers to the maximum tolerance for deviation between the actual suspension height and the reference value. It needs to be small enough to ensure smooth transportation, yet large enough to avoid frequent adjustments to the control system. For example, the allowable range for cell culture dishes is ±0.3mm (current adjustment is triggered when the deviation exceeds 0.3mm) to ensure near-stable suspension; the allowable range for freeze-dried vaccines is ±0.5mm (moderately relaxed to reduce electromagnet energy consumption).

[0107] When a structural deformation risk signal is received, the linkage execution module performs the following coordinated operations:

[0108] Activate the constant temperature locking mechanism; the constant temperature locking mechanism includes: calling the thermal expansion coefficient database of the material, combining it with the currently detected maximum deformation position, calculating the optimal locking temperature point (this temperature point includes a small thermal expansion compensation offset), and driving the semiconductor cooling array to stabilize the target area to the optimal locking temperature point at a preset maximum safe temperature change rate;

[0109] It should be explained that the database of thermal expansion coefficients of medical supplies refers to the database that stores the linear expansion coefficients (unit: °C) of common medical supplies and packaging materials (including containers and contents). -1 This database contains a coefficient that characterizes the degree of expansion and contraction of a material with changes in temperature (positive values ​​indicate thermal expansion, and negative values ​​indicate thermal contraction). The core content of the database includes:

[0110] Packaging materials: such as glass (3×10) -6 ℃ -1), medical polypropylene (10×10 -6 ℃ -1 ), aluminum-plastic composite film (20×10) -6 ℃ -1 ); contents: such as blood (4.5×10 ); -4 ℃ -1 Because it contains a large amount of water), freeze-dried protein preparations (1.2×10 -5 ℃ -1 (Dense structure).

[0111] The optimal temperature point is determined by adjusting the temperature and utilizing the thermal expansion and contraction properties of the material to offset the target temperature of the current maximum deformation. The calculation logic combines the "material properties at the location of maximum deformation" and the "allowable residual deformation." Specific steps include (taking a glass ampoule as an example): Locating the location of maximum deformation: Determining the area of ​​most significant deformation using laser interferometry (e.g., the neck of the ampoule, where the current micro-deformation is ΔL = 0.02 mm, in the direction of contraction, which may cause the bottle neck seal to loosen); Accessing the database: For example, the thermal expansion coefficient of the glass is α = 3 × 10⁻⁶. -6 ℃ -1 The original neck length L = 50 mm; calculate the temperature difference compensation: the goal is to elongate the neck by ΔL' = 0.01 mm through heating (compensating for 50% deformation while retaining a safety margin); determine the optimal locking temperature point: for example, if the current temperature is 4℃, the optimal locking temperature point = 4℃ + 6.7℃ ≈ 10.7℃ (slight expansion of the neck through heating to offset contraction deformation). The thermal expansion compensation deviation needs to be set according to the material's sealing requirements (e.g., vaccine vials need to be compensated to a residual deformation ≤ 0.005 mm, while ordinary reagent vials can be relaxed to ≤ 0.02 mm).

[0112] The maximum safe temperature change rate refers to the maximum allowable rate of heating / cooling (unit: °C / min) when the temperature actuator adjusts the temperature of the target area. It needs to be adapted to the tolerance of medical supplies to temperature changes. For example, live virus vaccines (such as influenza vaccines) are sensitive to rapid cooling, and the maximum safe temperature change rate is set at 0.5 °C / min (to avoid denaturation of viral capsid proteins); chemical preparations (such as contrast agents) have higher stability, and the maximum safe temperature change rate can be set at 2 °C / min.

[0113] Stabilizing the target area to the optimal locking temperature point at the preset maximum safe temperature change rate means that the temperature regulation process consists of two steps:

[0114] Rapid approach: Heat / cool at the maximum safe temperature change rate (e.g., from 4℃ to 10℃ at 0.5℃ / min, it takes 12 minutes) to avoid excessive deformation caused by slow temperature adjustment;

[0115] Stable locking: When approaching the target temperature (e.g., within ±0.5℃ range), the rate is reduced to 0.1℃ / min, eventually stabilizing at the optimal locking temperature point (e.g., 10.7℃) to prevent overshoot from causing new temperature stress.

[0116] The hydraulic damping device is activated; the hydraulic damping device identifies the current dominant vibration frequency by analyzing the spectral characteristics of the vibration acceleration sensor signal in real time, and adjusts the opening of each hydraulic damping valve according to the rule-based feedforward and feedback composite controller to suppress the vibration at the current dominant vibration frequency.

[0117] In addition, based on the severity level of deformation risk, the pre-stored speed-deformation relationship model is queried, and the maximum allowable movement speed command is calculated and output to the drive system in real time.

[0118] Among them, spectrum feature analysis refers to converting the time-domain signal (such as the vibration acceleration value that changes with time) collected by the vibration acceleration sensor into a frequency-domain signal through Fourier transform to obtain a "frequency-energy" distribution map (spectrum). Each frequency in the spectrum corresponds to the magnitude of vibration energy (the higher the energy, the more significant the vibration at that frequency).

[0119] Dominant vibration frequency identification refers to selecting the 1-2 frequencies with the highest energy (i.e., the frequencies that have the greatest impact on the material structure) from the frequency spectrum. For example, when a robot is moving on a smooth surface, the energy of the 10Hz frequency in the vibration spectrum accounts for 60% (due to the rotation of the wheels), so 10Hz is the current dominant vibration frequency; when turning, the energy of the 20Hz frequency jumps to the highest, and the dominant frequency switches to 20Hz.

[0120] Among them, the rule-based feedforward and feedback composite controller is a combined control strategy adapted to vibration suppression in the medical cold chain. Specifically: Feedforward control: Based on the identified dominant vibration frequency, the basic opening of the hydraulic damping valve is preset (e.g., rule base: 10Hz dominant → opening 30%, 20Hz dominant → opening 40%) to suppress vibrations of known frequencies in advance; Feedback control: The vibration signal after adjustment is collected in real time. If the actual vibration energy is still higher than the target value (e.g., after adjusting the opening of 10Hz vibration by 30%, the energy only decreases by 20%), the opening is dynamically corrected (e.g., increased to 35%) to compensate for the deviation of the feedforward control.

[0121] The hydraulic damping valve opening adjustment logic includes: the damping valve opening (0-100%) directly determines the flow resistance of the hydraulic oil (the larger the opening, the greater the resistance, and the stronger the damping). The adjustment rule is positively correlated with the dominant vibration frequency: low frequency dominant (e.g., 5-10Hz, caused by uneven ground): opening 30%-40% (medium damping, allowing small buffering); high frequency dominant (e.g., 20-30Hz, caused by mechanical resonance): opening 50%-60% (strong damping, quickly absorbing high frequency energy).

[0122] For example, if the dominant vibration frequency is detected to be 15Hz (high frequency), the feedforward setting is 50%, and the feedback detection still shows that the vibration energy exceeds the standard by 10%, then the composite controller will adjust the opening to 55% to quickly suppress the vibration at this frequency by enhancing damping, thus avoiding tube wall fatigue caused by high frequency vibration in the biological sample tube.

[0123] Based on the embodiments provided in this application, when dealing with the risk of low-temperature phase change, the frequency of the pulsed refrigerant is positively correlated with the ice crystal growth rate, and the cooling intensity can be dynamically adjusted according to the phase change progress. The adaptive sliding mode control algorithm can adjust the excitation current of the electromagnetic levitation in real time to ensure that it maintains a stable levitation state when the load changes, reducing the interference of vibration on the phase change. When dealing with the risk of structural deformation, the constant temperature locking mechanism calculates the optimal temperature point in combination with the thermal expansion coefficient of the material, which can reduce the superposition of thermal stress by stabilizing the temperature. The hydraulic damping device adjusts the opening of the damping valve according to the dominant vibration frequency, which can accurately suppress the vibration component that has the greatest impact on the material structure.

[0124] Furthermore, the dynamic analysis module performs the following steps to conduct a comprehensive risk assessment:

[0125] Based on the identified type of medical supplies, a fuzzy logic evaluation model structure customized for that type of supplies is loaded from a pre-built knowledge base. The fuzzy logic evaluation model structure includes specific input variable fuzzy partitioning, rule base, and weight configuration.

[0126] It should be explained that the pre-built knowledge base is a structured database that stores risk assessment-related data for different types of medical supplies (such as vaccines, blood products, biological samples, chemical reagents, etc.). The core content includes: a classification table of supply types (such as "highly sensitive", "medium sensitive" and "low sensitive" according to sensitivity); a fuzzy logic assessment model template for each type of supply (including input variable division, rule base and weight configuration); phase change characteristic parameters of the supply (such as critical temperature and coefficient of thermal expansion) and vibration tolerance parameters (such as sensitive frequency band and maximum allowable acceleration).

[0127] For example, under the "Highly Sensitive Class - Live Vaccine" entry in the knowledge base, the temperature weight (0.6) and vibration weight (0.4) of its fuzzy logic model are stored. The rule base contains rules such as "Temperature deviation ≥2℃ and lasting for 5 minutes → high risk", which supports the dynamic analysis module to quickly call the adapted model.

[0128] In some embodiments, the input variables are fuzzy partitioned: the two input variables, temperature deviation and mechanical vibration, are divided into fuzzy subsets (such as "low", "medium", and "high"), each subset corresponding to a specific numerical range. For example, for temperature deviation, "low" corresponds to 0-2℃, "medium" to 2-5℃, and "high" to >5℃; for mechanical vibration, "low" corresponds to 0-0.3g, "medium" to 0.3-0.6g, and "high" to >0.6g (g is the acceleration due to gravity).

[0129] Rule base: Based on the interaction between temperature and vibration in medical cold chain scenarios, preset "IF-THEN" rules are used. For example: "IF temperature deviation is high AND mechanical vibration is medium, THEN overall risk is high" and "IF temperature deviation is medium AND mechanical vibration is high, THEN overall risk is high".

[0130] Weighting: Assign weights to input variables based on the differences in the sensitivity of materials to temperature and vibration. For example, blood products are more sensitive to temperature, so the weight for temperature deviation is 0.7 and for mechanical vibration is 0.3; glass-bottled reagents are more sensitive to vibration, so the weights are set to temperature 0.4 and vibration 0.6.

[0131] Based on the original signals of the microscopic state, the temperature deviation index and the mechanical vibration index are calculated.

[0132] Among them, the temperature deviation index includes not only the absolute deviation of the temperature sensor reading from the set value, but also the weighted integral of the duration of the deviation to characterize the cumulative effect of the temperature difference; the mechanical vibration index integrates the effective value of vibration acceleration, high-frequency component energy (characterizing impact) and the proportion of vibration energy in a specific sensitive frequency band.

[0133] It should be noted that the calculation of the temperature deviation index includes not only the "absolute difference between the current temperature and the target temperature," but also reflects the cumulative effect through the "weighted integral of the deviation duration." The formula can be simplified as follows:

[0134] Temperature deviation = instantaneous temperature difference × k1 + ∫(instantaneous temperature difference) dt × k2

[0135] Where k1 and k2 are weighting coefficients, t is time, and k2 increases with the duration. For example, k2 = 0.1 for 1 minute and k2 = 0.3 for 5 minutes. For example, if a vaccine has a target temperature of 5℃, the temperature in the first stage (0-2 min) is 6℃ (temperature difference 1℃), and the temperature in the second stage (2-5 min) is 7℃ (temperature difference 2℃), then the temperature deviation = (1×k1 + 2×k1) + (1×2×k2 + 2×3×k2). The cumulative calculation reflects the harm of long-term deviation.

[0136] Mechanical vibration indexes include high-frequency components and sensitive frequency bands.

[0137] Among them, the high-frequency component energy refers to the energy contained in the component with a frequency higher than 20Hz in the vibration signal (because medical supplies packaging is mostly made of brittle materials, high-frequency vibration is prone to cause resonance). The quantification method is the proportion of high-frequency (20-100Hz) energy to the total vibration energy (the higher the proportion, the greater the vibration intensity).

[0138] Specific sensitive frequency bands refer to the frequency range in which different medical supplies are most sensitive to vibration due to their structural characteristics (such as packaging rigidity and the flowability of contents). For example, liquid reagents (such as unsealed cell culture medium) are sensitive to vibrations of 10-15Hz (which can easily cause liquid surface fluctuations and spillage); freeze-dried vaccines (solid particles) are sensitive to vibrations of 30-50Hz (which can easily cause structural breakage due to particle collisions).

[0139] The mechanical vibration index achieves a comprehensive quantification of vibration risk by integrating the effective value of vibration acceleration (reflecting intensity), the proportion of high-frequency component energy (reflecting high-frequency hazards), and the proportion of sensitive frequency band energy (reflecting specific hazards).

[0140] Input the calculated real-time values ​​of temperature deviation and mechanical vibration into the fuzzy logic evaluation model;

[0141] After the fuzzification process is performed in the fuzzy logic evaluation model, a preset rule triggering mechanism is adopted. The preset rule triggering mechanism is based on the mutual influence of temperature and vibration risks in the medical cold chain scenario and is configured as a two-level structure that combines temperature risk priority triggering and vibration risk cumulative triggering.

[0142] The specific logic of the two-level structure is as follows:

[0143] Level 1: Temperature Risk Priority Trigger: If the temperature deviation reaches the "high" fuzzy subset (e.g., temperature difference > 5℃), then regardless of the vibration level, the high-risk rule will be triggered directly (e.g., the rule "high temperature → high risk" in the rule base will be executed first), because sudden temperature changes usually damage medical supplies (e.g., protein preparations) more rapidly.

[0144] Level 2: Vibration risk accumulation trigger: If the temperature deviation is "medium" or "low", the cumulative vibration duration (e.g., vibration "medium" lasting more than 3 minutes) will trigger a vibration-related high-risk rule (e.g., "vibration is medium and accumulates for 3 minutes → high risk") when the cumulative value reaches the threshold, because vibration hazards are mostly cumulative effects.

[0145] For example, if the temperature deviation of a biological sample is "medium" (temperature difference 3℃), but the vibration level is "medium" for 4 minutes (exceeding the 3-minute threshold), then the second-level rule is triggered, and the overall risk is determined to be increased.

[0146] By using rule-based reasoning and weighted averaging to defuzzify, a comprehensive risk index ranging from 0 to 100 is output.

[0147] Rule-based reasoning involves matching real-time values ​​of temperature deviation and mechanical vibration with a rule base to trigger multiple relevant rules and generate fuzzy outputs (such as "high risk" and "medium risk"). For example, a "medium" temperature deviation and a "high" vibration might trigger Rule 1 (outputting "high risk") and Rule 2 (outputting "high risk"). Weighted average defuzzification assigns weights to the fuzzy outputs of the triggering rules (e.g., 0.6 for Rule 1 and 0.4 for Rule 2), and converts the fuzzy outputs into specific values ​​of 0-100 (100 for extremely high risk and 0 for no risk) using a formula (e.g., (100 × 0.6 + 70 × 0.4) = 88), i.e., a comprehensive risk index.

[0148] When the comprehensive risk index exceeds the warning threshold dynamically calculated based on the sensitivity level of the materials and the current transportation stage, a reverse instruction is generated.

[0149] The alert thresholds are dynamically adjusted based on the following factors: Material sensitivity level: High sensitivity (e.g., live vaccines) → Low threshold; Low sensitivity (e.g., solid tablets) → High threshold. Transportation stage: Outbound / inbound stage (frequent operations, increased risk) → Low threshold; Stable transportation stage → High threshold.

[0150] For example, for highly sensitive live vaccines during the outbound stage: the alert threshold is set to 30 (a reverse instruction is triggered if the comprehensive risk index is >30); for low-sensitivity tablets during the stable transportation stage: the alert threshold is set to 60 (allowing for higher risk fluctuations).

[0151] Analyze the core rule clusters that are triggered most strongly and contribute the most during the rule reasoning process, determine whether the core rule clusters are mainly associated with temperature risk factors or vibration risk factors, and output the dominant risk type identifier accordingly.

[0152] It should be explained that high-intensity triggering refers to a rule in the rule base being frequently or strongly triggered during the reasoning process, which needs to meet the following conditions: the number of triggers accounts for ≥50% of the total number of triggering rules (e.g., the rule is triggered 5 times or more out of 10 triggering rules); or the triggering intensity value of the rule (calculated based on the matching degree between the rule and real-time data, ranging from 0 to 1) is ≥0.8 (indicating a high degree of matching).

[0153] The core rule cluster refers to multiple associated rules that are triggered by high intensity. The dominant risk type is determined by analyzing the proportion of "temperature-related conditions" and "vibration-related conditions" in the rules.

[0154] For example, in a comprehensive risk assessment, the rule clusters triggered by high intensity include: Rule 1: High temperature deviation and medium vibration → High risk; Rule 2: High temperature deviation and low vibration → High risk; Rule 3: Medium temperature deviation and low vibration → Medium risk.

[0155] Analysis shows that all three rules take "temperature deviation" as the main condition (the first two explicitly include "high temperature deviation"). Therefore, it is determined that the core rule cluster is associated with temperature risk factors, and the "temperature risk-dominated" label is output.

[0156] For example, in a comprehensive risk assessment, if the rule "high temperature deviation → increased risk" is triggered 6 times (out of a total of 10 triggering rules), and the intensity of each trigger is 0.9, it is determined to be a "high intensity trigger".

[0157] Based on the embodiments provided in this application, the fuzzy logic assessment model has a customized structure for different types of materials, adaptable to the risk-sensitive characteristics of different materials such as vaccines and biological agents; temperature deviation is incorporated into the weighted integral of the deviation duration, reflecting the cumulative effect of temperature difference (such as the slow denaturation effect of long-term slight overheating on proteins); mechanical vibration intensity integrates multiple parameters to comprehensively characterize the combined effect of vibration on materials; the two-level rule triggering mechanism takes into account both the immediacy of temperature risk and the cumulative nature of vibration risk, conforming to the mutual influence between the two in medical scenarios; the comprehensive risk index of 0-100 and the dynamic warning threshold (combining the sensitivity level of materials and the transportation stage) make the risk assessment results more intuitive and adaptable to different application scenarios. These designs make the comprehensive risk assessment both comprehensive and accurate, providing a reliable basis for subsequent intervention.

[0158] Furthermore, when the dominant risk type is identified as temperature risk, the reverse execution module executes the following strategy:

[0159] Based on real-time temperature data and historical temperature stability analysis reports for each area within the facility, the standard deviation of temperature fluctuation and the maximum temperature gradient for each candidate path area are calculated.

[0160] In some embodiments, real-time temperature data and historical temperature stability analysis reports for each area within the facility are obtained by requesting them from the central environmental monitoring system. The facility refers to the physical space and supporting systems involved in medical cold chain logistics, including but not limited to: cold chain warehouses (including refrigerated areas and transit areas), automated transport channels (such as AGV tracks and temperature-controlled corridors), loading and unloading platforms, temporary storage cabinets, etc. The temperature stability of these facilities directly affects cabin temperature control. This report is a long-term statistical analysis of temperature data for each area of ​​the facility, including: temperature monitoring data for the past 3 months (recorded every 5 minutes); standard deviation of temperature fluctuations (reflecting stability), maximum temperature difference, abnormal temperature events (such as temperature rises caused by equipment failure) and their duration for each area; and differences in temperature characteristics at different times (such as day / night, weekday / weekend).

[0161] In some embodiments, candidate path regions are derived from a pre-defined path library (such as fixed transport routes within a warehouse) and real-time planned possible paths (based on the robot's current and target positions). The partitioning method includes dividing the space into continuous small segments (e.g., every 5 meters), with each segment serving as an independent unit for evaluating temperature stability. For example, a pre-defined path from "Cold Storage A" to "Inspection Area B" is divided into candidate path regions such as "Cold Storage Exit Area (0-5m)," "Main Channel Zone 1 (5-10m)," and "Transfer Platform Area (10-15m)."

[0162] The maximum temperature gradient refers to the ratio of the maximum temperature difference at different locations within a region to the distance (unit: ℃ / m), reflecting the uniformity of temperature spatial distribution. For example, if the temperatures at both ends of the "transfer station area" are 6℃ and 9℃ respectively, and the distance is 3m, then the maximum temperature gradient = (9-6) / 3 = 1℃ / m. The larger the gradient, the more significant the temperature difference within the region.

[0163] The A* path search algorithm is adopted, which takes temperature stability as a cost function of equal importance to path length, and introduces a temperature decay coefficient to simulate the potential impact rate of temperature on materials in different regions, so as to plan a transportation route with the minimum total temperature risk cost.

[0164] Among them, the A* algorithm incorporates a temperature stability index into traditional path planning (at the cost of path length), and its core logic is as follows:

[0165] Define a node: Each candidate path region is a node;

[0166] The cost function is f(n) = g(n) + h(n), where:

[0167] g(n): The actual cost from the starting point to the current node (including path length cost and temperature stability cost);

[0168] h(n): Estimated cost from the current node to the target node (same logic as g(n)).

[0169] Temperature stability is transformed into a quantifiable cost (the larger the σ, the higher the cost) by "temperature fluctuation standard deviation × weight", thus achieving a balance between "shortest path" and "most stable temperature".

[0170] Cost function: Temperature stability and path length are given equal weight (50% each), i.e.:

[0171] Total cost = Path length × 0.5 + Temperature fluctuation standard deviation × 0.5 × k (k is a conversion factor to make the two units consistent).

[0172] Temperature decay coefficient: The rate at which the simulated area temperature affects the cabin temperature (the farther the distance and the better the insulation of the area, the weaker the impact), with a value of 0-1 (e.g., 0.3 for a constant temperature corridor and 0.8 for a normal passage). For example: If the temperature of a certain area fluctuates greatly but is far from the robot's current position, the temperature decay coefficient reduces its impact on the total cost, avoiding excessive avoidance.

[0173] In another implementation, this approach is a further extension of the above solution. For example, the expression for the cost function can be calculated in one of the following ways, or it can be in other forms. The form of the cost function does not affect the implementation of the preceding technical solution:

[0174] T cost =λ×L′ path +(1-λ)×(σ_T'×k_1+G_T'×k_2)×β

[0175] Among them, T cost The total cost of the transportation route (range 0-10) comprehensively weighs the trade-off between route length and temperature risk; a smaller value indicates a better route. λ is the route length weighting coefficient (value 0.5), because temperature stability is as important as route length, thus balancing with the temperature risk weighting. L' path The normalized path length (range 0-5) is obtained by... Calculation (L) actual L represents the actual path length in meters. max The longest path in the scene (e.g., 100m); σ_T' is the normalized standard deviation of temperature fluctuation (dimensionless, range 0-5). Calculate (σ_T is the actual temperature standard deviation, in °C; σ max , where k_1 is the maximum temperature standard deviation in the scenario (e.g., 2℃), reflecting temperature stability; k_1 is the temperature fluctuation weighting coefficient (1.2 in this embodiment), because frequent temperature fluctuations have a more significant cumulative impact on materials, its weight is higher than that of the temperature gradient; G_T' is the normalized maximum temperature gradient (dimensionless, range 0-5), obtained through... Calculate (G_T is the actual temperature gradient, unit: °C / m; G max 1 represents the maximum temperature gradient in the scenario (e.g., 1℃ / m), reflecting the uniformity of temperature distribution within the area; k_2 is the temperature gradient weighting coefficient (value 0.8), which has a smaller impact than temperature fluctuations and adapts to the influence of gradients on materials in the scenario; β is the temperature decay coefficient (range 0.3-0.8), with 0.3 for the constant temperature corridor (weak temperature influence) and 0.8 for the ordinary channel (strong temperature influence), simulating the degree of interference of different areas on the cabin temperature.

[0176] Based on the difference between the current cabin temperature and the target safe cabin temperature, the heat capacity characteristics of the materials, and the latent heat requirement of phase change, determine the number of steps and the target temperature of each step in order to control the temperature actuator to perform step-by-step cooling curve adjustment.

[0177] The target safety chamber temperature is a temperature range set according to medical supply storage standards. For example, for blood products: target safety chamber temperature 2-6℃; for freeze-dried vaccines: target safety chamber temperature 2-8℃.

[0178] Heat capacity refers to the amount of heat required to raise the temperature of a unit mass of medical supplies by 1°C (unit: J / (kg·°C)). The higher the heat capacity, the slower the temperature change, and the longer the holding time is required. Latent heat of phase change refers to the heat absorbed / released when a material undergoes a phase change (such as solidification or crystallization). If the cooling process may cross the phase change point, additional cooling capacity and a longer holding time are required.

[0179] The logic for setting up stepped cooling includes: based on the difference between the current chamber temperature and the target safe chamber temperature (e.g., current temperature 15℃, target temperature 5℃, difference 10℃); combined with heat capacity characteristics (e.g., the heat capacity of a certain vaccine is 3000J / (kg·℃)) and latent heat of phase change (e.g., no phase change, latent heat 0), dividing the temperature into steps (e.g., 15→10→5℃, 2 steps); the holding time for each step is set according to the thermal relaxation time (the time required for the internal temperature of the material to become uniform, the larger the heat capacity, the longer the time) (e.g., holding for 2 minutes for a 10℃ step and holding for 3 minutes for a 5℃ step).

[0180] Each step has a defined heat preservation time, which is determined based on the thermal relaxation time of the material in that temperature zone, and the rate of temperature change between adjacent steps is limited to a preset safe temperature change rate.

[0181] The maximum safe temperature change rate refers to the maximum cooling rate between adjacent steps, which avoids stress caused by rapid temperature difference in materials. For example, blood plasma (containing a large amount of water, which is prone to freezing when cooled rapidly): maximum safe temperature change rate 0.5℃ / min; solid tablets (structurally stable): maximum safe temperature change rate 2℃ / min.

[0182] When the dominant risk type is identified as vibration risk, the reverse execution module executes the following strategy:

[0183] Switch to step movement mode to break down continuous movement into discrete movement cycles and pause cycles;

[0184] Among them, the moving step size of each cycle is dynamically reduced according to the current comprehensive risk index. The higher the risk, the shorter the step size. The pause time is set according to the minimum stabilization time required for the shock absorption device to absorb residual vibration.

[0185] During each movement and pause, the refrigerant metering compensation device is triggered. Based on the temperature change trend of the cabin and the feedback of the cargo temperature during the pause, the refrigerant metering compensation device predicts the minimum amount of refrigerant required to maintain the target temperature in the next movement phase and replenishes it.

[0186] For example, the logic for predicting the minimum cooling dose required for the next movement phase includes:

[0187] During the pause, the temperature change rate of the chamber was collected (e.g., a temperature rise of 0.3℃ within 1 minute);

[0188] Estimate the cabin temperature change in the next movement phase (e.g., 2 minutes) (0.3℃ / min × 2min = 0.6℃);

[0189] Based on the cabin's heat capacity and refrigerant efficiency (e.g., 1ml of refrigerant can reduce the temperature by 0.2℃), the required refrigerant dosage is calculated as 0.6℃ / 0.2℃ × 1ml = 3ml, ensuring that the cabin temperature remains stable within the target range after compensation.

[0190] Based on the embodiments provided in this application, when temperature risk is dominant, path planning incorporates the standard deviation of temperature fluctuation and gradient as cost functions. Combined with the A* algorithm, it can prioritize temperature-stable regions. The holding time in step-cooling is set based on thermal relaxation time, ensuring uniform temperature change within the material and avoiding excessive local temperature differences. When vibration risk is dominant, the step size of the step-by-step movement dynamically decreases with the comprehensive risk index, and the pause time is set according to vibration reduction requirements, reducing the accumulation of vibration energy. Furthermore, the refrigerant quantitative compensation, based on temperature feedback during pauses, can offset temperature fluctuations during movement. This refined control based on material characteristics and risk status allows reverse operations to suppress dominant risks while minimizing the additional impact of intervention on the material.

[0191] Furthermore, such as Figure 3 As shown, the feedback optimization module performs the following optimization process after the reverse execution operation is completed:

[0192] S301, collects a new round of raw signals of the microscopic state;

[0193] S302, based on the new round of original micro-state signals, recalculate the updated comprehensive risk index using the previous evaluation model;

[0194] S303, calculate the difference between the updated comprehensive risk index and the comprehensive risk index when the reverse instruction is triggered, and convert the difference into a risk reduction percentage;

[0195] S304. Based on the preset effect evaluation interval to which the risk reduction percentage belongs, select and execute the corresponding parameter optimization strategy; wherein, the effect evaluation interval includes the significant reduction interval, the moderate reduction interval, and the insufficient reduction interval;

[0196] In some embodiments, if the risk is significantly reduced, the weight of the dominant risk factor in the assessment model is appropriately reduced, or the trigger threshold of its related rules is lowered, so that the system's response to the controlled risk is slightly smoother.

[0197] If the risk reduction is insufficient, then conduct in-depth analysis of the residual or newly emerging risk characteristic patterns in the new round of signals, add rules for such characteristic patterns to the fuzzy rule base, or strengthen the weight of existing relevant rules.

[0198] S305, calculates the rate of decrease of the comprehensive risk index when the reverse instruction is triggered during the execution of the reverse operation; based on this rate value, dynamically sets the signal acquisition frequency for the next monitoring cycle;

[0199] For example, when the rate of descent is fast, the frequency can be appropriately reduced, while when the rate is slow or the risk rebounds, the frequency can be increased.

[0200] S306 applies all optimized and adjusted model parameters, rule base content, and signal acquisition frequency settings to subsequent risk assessment iterations for the same type of medical supplies.

[0201] Based on the embodiments provided in this application, the actual effect of reverse operation can be quantified by calculating the risk reduction percentage and dividing the evaluation interval. Selecting parameter optimization strategies accordingly ensures that model adjustments match actual needs. Dynamically setting the signal acquisition frequency based on the rate of decrease of the comprehensive risk index allows for increased detection density when risk changes rapidly and decreased frequency when risk is stable, reducing resource consumption while maintaining monitoring accuracy. Applying the optimized parameters to subsequent evaluations of similar materials enables the control system to gradually accumulate experience in controlling specific materials, such as forming a more accurate evaluation logic for the temperature and vibration interaction risk of a certain type of biological agent. This refined iteration based on actual operating data allows the performance of the control system to continuously improve as application scenarios expand, enhancing its reliability and adaptability in long-term applications.

[0202] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A control system for a medical cold chain logistics handling robot, comprising the medical cold chain logistics handling robot, characterized in that, include: The state perception module is used to acquire raw signals of the microscopic state of medical supplies in real time through an optical sensor array; Based on the original microscopic state signal, determine whether there is a risk of low-temperature phase transition or structural deformation, and generate a corresponding risk identification signal; The linkage execution module is used to control the temperature actuator and the mechanical actuator according to the risk identification signal; The dynamic analysis module is used to establish an assessment model for temperature deviation and mechanical vibration based on the type of medical supplies; based on the original microscopic state signal, it calculates the real-time values ​​of temperature deviation and mechanical vibration; the real-time values ​​are input into the assessment model, and a comprehensive risk index is output. When the comprehensive risk index exceeds the warning threshold, a reverse instruction is generated and the dominant risk type identifier is output. A reverse execution module is used to respond to the reverse command and control the temperature actuator and the mechanical actuator according to the dominant risk type identifier; The medical cold chain logistics handling robot includes: a robot mobile platform; a transport compartment set on the robot mobile platform for carrying medical supplies; a temperature actuator integrated in the transport compartment for regulating the temperature environment inside the compartment; a mechanical actuator including a drive unit and a shock absorption unit, the mechanical actuator being installed at the bottom of the robot mobile platform; and an optical sensor array deployed on the inner wall of the transport compartment for real-time acquisition of raw signals of the microscopic state of the surface of the medical supplies. The risk identification signals include low-temperature phase change risk signals and / or structural deformation risk signals; the temperature actuator includes a semiconductor cooling chip array and a pulsed refrigerant injection device; the mechanical actuator includes an electromagnetic levitation device, a hydraulic shock absorption device, and a wheel drive module; The linkage execution module controls the temperature actuator and the mechanical actuator according to the risk identification signal, and is configured as follows: When the low-temperature phase change risk signal is detected, the pulse-type refrigerant discrete supply of the temperature actuator is activated, and the electromagnetic levitation transport mode of the mechanical actuator is switched. When the structural deformation risk signal is detected, the constant temperature locking mechanism of the temperature actuator is activated, and the hydraulic damping device and speed limiter of the mechanical actuator are started.

2. The control system of the medical cold chain logistics handling robot according to claim 1, characterized in that, The reverse execution module controls the temperature actuator and the mechanical actuator according to the dominant risk type identifier, and is configured as follows: If the dominant risk type is identified as temperature risk, plan the insulated passage transportation route and control the temperature actuator to adjust the cabin temperature according to the stepped cooling curve. If the dominant risk type is identified as vibration risk, switch the stepping movement mode of the mechanical actuator and trigger the refrigerant metering compensation device.

3. The control system of the medical cold chain logistics handling robot according to claim 1, characterized in that, The control system of the medical cold chain logistics handling robot also includes: The feedback optimization module is used to collect a new round of original micro-state signals after the reverse execution operation is completed; update the comprehensive risk index based on the new round of original micro-state signals; compare the updated comprehensive risk index with the comprehensive risk index when the reverse instruction was generated; and update the model parameters of the evaluation model based on the comparison results.

4. The control system of the medical cold chain logistics handling robot according to claim 1, characterized in that, The state-aware module acquires and analyzes the original microscopic state signal, and is configured as follows: The hyperspectral imaging unit and the laser interferometry unit are simultaneously triggered to acquire hyperspectral image sequences of surface reflections from medical supplies and surface micro-displacement interference fringe patterns formed by laser interferometry, respectively. For the hyperspectral image sequence, an ice crystal feature extraction algorithm is used for processing; wherein, the ice crystal feature extraction algorithm focuses on a specific ice crystal indicator band, and calculates the rate of change of the branch length of the ice crystal outline in the image of the band between adjacent frames as the ice crystal growth rate to capture early signs of phase change; when the ice crystal growth rate continuously exceeds the morphological evolution threshold preset based on the phase change characteristics of the material, it is determined that there is a risk of low temperature phase change; The micro-displacement interference fringe pattern on the surface is processed using a phase unwrapping algorithm to reconstruct the three-dimensional micro-deformation field of the material surface. The reconstructed three-dimensional micro-deformation field is compared with a database of known elastic moduli of materials to identify local micro-strain concentration areas and calculate their proximity to the material's yield limit. When the proximity exceeds a preset safety margin threshold, it is determined that there is a risk of structural deformation. Based on the determination results of low-temperature phase transition risk or structural deformation risk, the corresponding low-temperature phase transition risk signal or structural deformation risk signal is generated.

5. The control system of the medical cold chain logistics handling robot according to claim 4, characterized in that, When the low-temperature phase transition risk signal is received, the linkage execution module performs the following coordinated operation: Based on the ice crystal growth rate calculated in real time by the state perception module, the control parameters for pulsed refrigerant discrete replenishment are dynamically calculated and output. The pulse frequency is positively correlated with the ice crystal growth rate, and the single injection dose is compensated based on the instantaneous temperature difference of the target area. In the electromagnetic levitation transport mode, by reading data from multiple levitation gap sensors in real time, an adaptive sliding mode control algorithm is used to dynamically adjust the excitation current of each independent electromagnet, so as to maintain the levitation height at a set reference value under changes in transport load and suppress gap fluctuations within a preset allowable range. When the structural deformation risk signal is received, the linkage execution module performs the following coordinated operation: Activate the constant temperature locking mechanism; wherein, the constant temperature locking mechanism includes: calling the thermal expansion coefficient database of the material, combining it with the currently detected maximum deformation position, calculating the optimal locking temperature point, and driving the semiconductor cooling array to stabilize the target area to the optimal locking temperature point at a preset maximum safe temperature change rate; The hydraulic damping device is activated; wherein, the hydraulic damping device identifies the current dominant vibration frequency by analyzing the spectral characteristics of the vibration acceleration sensor signal in real time, and adjusts the opening of each hydraulic damping valve according to the rule-based feedforward and feedback composite controller to suppress the vibration of the current dominant vibration frequency.

6. The control system of the medical cold chain logistics handling robot according to claim 1, characterized in that, The dynamic analysis module performs the following steps to conduct a comprehensive risk assessment: Based on the identified type of medical supplies, a fuzzy logic evaluation model structure customized for that type of supplies is loaded from a pre-built knowledge base. The fuzzy logic evaluation model structure includes specific input variable fuzzy partitioning, rule base, and weight configuration. Based on the original signal of the microstate, the temperature deviation index and the mechanical vibration index are calculated. The temperature deviation index includes not only the absolute deviation of the temperature sensor reading from the set value, but also the weighted integral of the duration of the deviation to characterize the cumulative effect of the temperature difference; the mechanical vibration index integrates the effective value of vibration acceleration, high-frequency component energy, and the proportion of vibration energy in a specific sensitive frequency band. Input the calculated real-time values ​​of temperature deviation and mechanical vibration into the fuzzy logic evaluation model; After the fuzzification process is performed in the fuzzy logic evaluation model, a preset rule triggering mechanism is adopted; wherein, the preset rule triggering mechanism is based on the mutual influence of temperature and vibration risks in the medical cold chain scenario, and is configured as a two-level structure combining temperature risk priority triggering and vibration risk cumulative triggering. By using rule-based reasoning and weighted averaging to defuzzify, a comprehensive risk index ranging from 0 to 100 is output. When the comprehensive risk index exceeds the warning threshold dynamically calculated based on the sensitivity level of the materials and the current transportation stage, a reverse instruction is generated. Analyze the core rule clusters that are triggered most strongly and contribute the most during the rule reasoning process, determine whether the core rule clusters are mainly associated with temperature risk factors or vibration risk factors, and output the dominant risk type identifier accordingly.

7. The control system of the medical cold chain logistics handling robot according to claim 2, characterized in that, When the dominant risk type is identified as temperature risk dominant, the reverse execution module executes the following strategy: Based on real-time temperature data and historical temperature stability analysis reports for each area within the facility, the standard deviation of temperature fluctuation and the maximum temperature gradient for each candidate path area are calculated. Using A The path search algorithm treats temperature stability as a cost function of equal importance to path length, and introduces a temperature decay coefficient to simulate the potential impact rate of temperature on materials in different regions, so as to plan a transportation path with the minimum total temperature risk cost. Based on the difference between the current cabin temperature and the target safe cabin temperature, the heat capacity characteristics of the materials, and the latent heat requirement of phase change, determine the number of steps and the target temperature of each step in order to control the temperature actuator to perform step-by-step cooling curve adjustment. Each step has a clearly defined heat preservation time, which is determined based on the thermal relaxation time of the material within the corresponding temperature range of each step. The rate of temperature change between adjacent steps is limited to a preset safe rate of temperature change. When the dominant risk type is identified as vibration risk, the reverse execution module executes the following strategy: Switch to the step-by-step movement mode to decompose continuous movement into discrete movement cycles and pause cycles; Among them, the moving step size of each cycle is dynamically reduced according to the current comprehensive risk index, and the pause time is set according to the minimum stabilization time required for the shock absorption device to absorb residual vibration. During each movement and pause, the refrigerant metering compensation device is triggered; wherein, the refrigerant metering compensation device predicts the minimum amount of refrigerant required to maintain the target temperature in the next movement phase based on the change trend of the cabin temperature and the feedback of the material temperature during the pause, and replenishes it.

8. The control system of the medical cold chain logistics handling robot according to claim 3, characterized in that, The feedback optimization module performs the following optimization process after the reverse execution operation is completed: Collect a new round of raw signals of the microscopic state; Based on the new round of original micro-state signals, the updated comprehensive risk index is recalculated using the previous assessment model; Calculate the difference between the updated composite risk index and the composite risk index when the reverse instruction is triggered, and convert the difference into a risk reduction percentage; Based on the preset effect evaluation range to which the risk reduction percentage belongs, select and execute the corresponding parameter optimization strategy; Calculate the rate of decrease of the comprehensive risk index when the reverse instruction is triggered during the execution of the reverse operation; based on this rate value, dynamically set the signal acquisition frequency for the next monitoring cycle; All optimized and adjusted model parameters, rule base content, and signal acquisition frequency settings will be applied to subsequent risk assessment iterations for the same type of medical supplies.

Images