Intelligent health monitoring and material distribution system based on unmanned aerial vehicle

Abstract

The application discloses an unmanned aerial vehicle (UAV) based intelligent health monitoring and material distribution system, and relates to the technical field of UAV application, comprising the following modules: a comprehensive information platform, a UAV platform, a data acquisition and transmission module, a communication module, an intelligent decision module, a data closed loop updating module, a ground base station and a user terminal; the system fuses medical MEWS scores and comprehensive material emergency degree scores through a dynamic priority management engine to realize intelligent allocation of bandwidth resources; a three-level decision mechanism of a hierarchical path planner combined with a fitness function in an improved genetic algorithm, obstacle avoidance reinforcement learning and A* emergency re-planning is adopted; full-dimensional monitoring of a transportation process is realized through YOLOv5 damage detection, torque deviation analysis and vibration frequency spectrum correlation; the real-time resource competition problem of medical first aid and material distribution is solved, the success rate of emergency material distribution is improved, and the transportation loss rate is effectively reduced.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of unmanned aerial vehicle application, and particularly relates to an intelligent health monitoring and material distribution system based on unmanned aerial vehicles. BACKGROUND

[0002] With the development of science and technology, unmanned aerial vehicles have shown great potential in many fields.

[0003] However, at present, unmanned aerial vehicles mostly operate independently in health monitoring and material distribution, and traditional unmanned aerial vehicle distribution systems cannot dynamically coordinate the communication resource competition between medical emergency and ordinary logistics, resulting in delay of high-priority tasks. In existing solutions, only path length optimization is considered, and multi-dimensional parameters such as moment balance and unmanned aerial vehicle battery health are not integrated. The conventional vibration monitoring method uses a fixed threshold alarm and cannot identify the correlation features of package damage and flight attitude, resulting in a high false alarm rate. At the same time, mainstream systems lack a federated learning framework, making it difficult to achieve closed-loop optimization of model iteration.

[0004] In some remote areas, disaster sites, etc., there is an urgent need for efficient and accurate health monitoring and timely material distribution, and the existing technology cannot meet these needs well. Therefore, an intelligent health monitoring and material distribution system based on unmanned aerial vehicles has emerged. SUMMARY

[0005] The present application aims to at least solve one of the technical problems existing in the prior art. To this end, the present application provides an intelligent health monitoring and material distribution system based on unmanned aerial vehicles, which is used to solve the following technical problems:

[0006] At present, unmanned aerial vehicles mostly operate independently in health monitoring and material distribution, and traditional unmanned aerial vehicle distribution systems cannot dynamically coordinate the communication resource competition between medical emergency and ordinary logistics, resulting in delay of high-priority tasks. In existing solutions, only path length optimization is considered, and multi-dimensional parameters such as moment balance and unmanned aerial vehicle battery health are not integrated. The conventional vibration monitoring method uses a fixed threshold alarm and cannot identify the correlation features of package damage and flight attitude, resulting in a high false alarm rate. At the same time, mainstream systems lack a federated learning framework, making it difficult to achieve closed-loop optimization of model iteration.

[0007] To solve the above problems, the present application provides an intelligent health monitoring and material distribution system based on unmanned aerial vehicles, which comprises a comprehensive information platform, an unmanned aerial vehicle platform, a data acquisition and transmission module, a communication module, an intelligent decision-making module, a data closed-loop updating module, a ground base station and a user terminal.

[0008] The comprehensive information platform constitutes a distributed data interaction hub of medical institutions, logistics centers and modular unmanned aerial vehicle platforms, and comprises a security encryption module and a dynamic priority management engine.

[0009] The data acquisition and transmission module collects the unmanned aerial vehicle flight state, material state and environmental information, outputs the packaging damage probability value, torque deviation coefficient and Pearson correlation coefficient, and transmits the preprocessed data to the comprehensive information platform.

[0010] The communication module establishes a two-way communication link between the unmanned aerial vehicle and the comprehensive information platform, ground base station and other related equipment.

[0011] The intelligent decision-making module constructs a hierarchical path planner, including a strategic layer, a tactical layer and an emergency layer, and constructs a suitability function with path length, dynamic level and battery health as parameters.

[0012] The data closed loop update module synchronously updates the MEWS score and time-sensitive emergency factor verification result through user terminal receipt confirmation after the unmanned aerial vehicle completes the delivery.

[0013] Preferably, the security encryption module and dynamic priority management engine include the following steps:

[0014] The security encryption module uses the national SM4 algorithm for end-to-end data transmission encryption.

[0015] The priority management engine dynamically adjusts data channel bandwidth allocation according to medical emergency level and logistics time-sensitive emergency level.

[0016] The medical emergency level is calculated by uploading the user's own medical data through the user terminal, and the MEWS score is automatically calculated as the medical emergency level, and the grading rules are set:

[0017] If 0 < MEWS score ≤ 4, the medical emergency level is medical level three.

[0018] If 5 ≤ MEWS score ≤ 6, the medical emergency level is medical level two.

[0019] If MEWS score ≥ 7, the medical emergency level is medical level one.

[0020] Preferably, the logistics time-sensitive emergency level includes the following steps:

[0021] A mapping table of material type-time-sensitive attribute is established, wherein the value of the critical time corresponding to the transportation material type is determined according to the biological characteristics of the material, specifically:

[0022] Blood products: set the critical time ≤ 120 minutes, corresponding to the time-sensitive emergency coefficient 0.9; vaccines: set the critical time ≤ 180 minutes, corresponding to the time-sensitive emergency coefficient 0.8; regular medicines: set the critical time ≤240 minutes corresponds to an emergency coefficient of 0.5; the critical time value is determined based on the biological characteristics of the materials.

[0023] Real-time acquisition of remaining transportation time for supplies, and calculation of urgency factors, specifically:

[0024]

[0025] in, As a factor of timeliness and urgency, The remaining transportation time for the supplies. The critical time corresponding to the type of transported goods;

[0026] The corresponding timeliness urgency coefficient and timeliness urgency factor are weighted and added together to obtain a comprehensive material urgency score, specifically:

[0027]

[0028] in, To score the overall urgency of the supplies, As the urgency factor, As a factor of timeliness and urgency, and These are the corresponding weighting coefficients;

[0029] Based on the comprehensive urgency level score of supplies, a tiered system is established:

[0030] If 0 < If the value is less than 0.4, the urgency level of the supplies is Level 3.

[0031] If 0.4≤ If the value is less than 0.7, the urgency level of the supplies is classified as Level 2.

[0032] like If the value is ≥0.7, the urgency level of the supplies is Level 1;

[0033] The combined priority level is calculated using a weighted scoring method.

[0034] Preferably, the joint priority level includes the following steps:

[0035] Define a joint priority matrix based on the urgency of medical care and supplies, and dynamically allocate data channel bandwidth.

[0036] The emergency levels of medical care and supplies are updated in real time based on the MEWS score and the comprehensive emergency level score.

[0037] The bandwidth allocation algorithm uses a weighted scoring method to calculate the joint priority level. The urgency of medical and material needs is converted into corresponding numerical values ​​based on the level: level three corresponds to a numerical value of 1, level two to 2, and level one to 3. Specifically:

[0038]

[0039] in, As a joint priority level, For medical emergencies, Depending on the urgency of the supplies, and For the corresponding weighting coefficients, and The values ​​are all 1;

[0040] The bandwidth allocation ratio is mapped to the joint priority level as follows:

[0041] like If ≥5, the bandwidth allocation ratio is 80%; if 3 < If <5, then the bandwidth allocation ratio is 60%; if If the value is ≤3, then the bandwidth allocation ratio is 40%.

[0042] Preferably, the data acquisition and transmission module includes the following steps:

[0043] The UAV receives mission instructions from the integrated information platform and takes off from the ground base station according to the preset takeoff procedure. During the flight, the data acquisition and transmission module begins to collect flight status data, material status data and environmental information of the UAV in real time.

[0044] The images of the packaging materials captured by the camera are input into the pre-trained YOLOv5 model, which outputs the probability value of packaging damage. If the probability value of packaging damage is ≥0.7, it is judged as a risk of damage, triggering the emergency retransmission mechanism, and the data is uploaded to the data integration information platform first through the SM4 encrypted channel.

[0045] Meanwhile, a battery health predictor is built based on extended Kalman filtering, taking historical data of voltage, current and temperature as input, and outputting the drone battery health status; MEMS gyroscope and visual SLAM data are fused to output six-degree-of-freedom pose; and the torque deviation coefficient of the dynamic load balancer is calculated based on the weight distribution of the cargo.

[0046] Establish a correlation model between the vibration spectrum of materials and the attitude angle of the UAV, and preset a correlation coefficient threshold. If the correlation coefficient is lower than the preset correlation coefficient threshold, the vibration is determined to be abnormal and the vibration optimization command is triggered.

[0047] By using a federated learning framework, local monitoring data is compared with failure modes in a comprehensive information platform to generate predictive maintenance recommendations.

[0048] Preferably, the probability value of packaging damage includes the following steps:

[0049] The training data for the YOLOv5 model includes images of damaged packaging materials in multiple scenarios, and an attention mechanism is introduced.

[0050] The probability value of packaging damage is calculated using a confidence-weighted average, specifically as follows:

[0051]

[0052] in, This represents the probability value of packaging damage. For the first The confidence weight of each detection box. This represents the total number of detection frames. The first output of the YOLOv5 model The probability of damage corresponding to each detection box.

[0053] Preferably, the establishment of the correlation model between the vibration spectrum of the material and the attitude angle of the UAV includes the following steps:

[0054] The correlation model uses wavelet packet decomposition to extract the vibration spectrum features of the data and performs Pearson correlation analysis with the UAV attitude angle, specifically:

[0055]

[0056] in, The Pearson correlation coefficient is used. The first characteristic of the vibration spectrum of materials One observation value, The first of the attitude angular acceleration of the UAV One observation value, This represents the mean value of the vibration spectrum characteristics of the material. This represents the mean of the angular acceleration of the UAV. This represents the total number of observations.

[0057] Preferably, the torque deviation coefficient includes the following steps:

[0058] The torque deviation coefficient is specifically:

[0059]

[0060] in, This is the torque deviation coefficient. For the first The weight of each item For the first The distance from the center of gravity of the supplies to the center of the drone. For the first The angle between the supplies and the drone's longitudinal axis. This represents the total quantity of supplies.

[0061] Preferably, the intelligent decision-making module includes:

[0062] The integrated information platform pre-marks high-risk areas and assigns them preset sensitivity levels. The drone dynamically adjusts the area level based on the current torque deviation coefficient.

[0063] The dynamic level is obtained by multiplying the torque deviation coefficient by the preset sensitivity level;

[0064] The fitness function is specifically as follows:

[0065]

[0066] in, For the fitness function, For path length, For dynamic levels, For battery health, , and These are the maximum path length, maximum dynamic level, and maximum battery health, respectively. , and These are the corresponding weight coefficients;

[0067] The tactical layer adjustment includes updating the obstacle map through visual SLAM and MEMS data, and triggering a shock absorption command when the Pearson correlation coefficient is less than a preset correlation coefficient threshold.

[0068] The triggering conditions for the emergency response include: MEWS score ≥ 7, packaging damage probability value ≥ preset damage threshold, battery health < 20%, and torque deviation coefficient ≥ preset deviation threshold.

[0069] Preferably, the data closed-loop update module includes:

[0070] Once the drone arrives at the target location, the user terminal confirms receipt, the MEWS score is updated synchronously, and a QR code is scanned to verify whether the timeliness and urgency factor has been met.

[0071] At the same time, the data is sent back to the integrated information platform to update the material type-timeliness mapping table and retrain the YOLOv5 model.

[0072] The beneficial effects of this invention are:

[0073] This invention constructs a joint priority matrix by using MEWS medical scores and material timeliness and urgency factors, dynamically adjusts communication bandwidth allocation, improves bandwidth allocation response speed, and avoids low-priority tasks from occupying critical resources.

[0074] This invention achieves multi-objective optimization by introducing torque deviation coefficient, dynamic risk level and UAV battery health into the fitness function of the genetic algorithm. Furthermore, the tactical layer adopts reinforcement learning dynamic obstacle avoidance, and the emergency layer adopts A* algorithm to force replanning, adapting to sudden obstacles or extreme weather and reducing the time of ineffective flight.

[0075] This invention uses YOLOv5 combined with an attention mechanism to detect images of material packaging, sets up a corresponding emergency retransmission mechanism, constructs a Pearson correlation coefficient vibration model to analyze the correlation between vibration spectrum and flight attitude in real time, and predicts battery health through extended Kalman filtering to provide early warning of insufficient drone power, effectively improving the identification rate of damaged material packaging and reducing transportation loss rate. Attached Figure Description

[0076] Figure 1 This is a schematic diagram of the module flow of the present invention;

[0077] Figure 2 This is a schematic diagram of the hierarchical path planner of the present invention. Detailed Implementation

[0078] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0079] Please see Figure 1 As shown, this invention is an intelligent health monitoring and material delivery system based on unmanned aerial vehicles (UAVs), comprising an integrated information platform, a UAV platform, a data acquisition and transmission module, a communication module, an intelligent decision-making module, a data closed-loop update module, a ground base station, and a user terminal.

[0080] The integrated information platform constitutes a distributed data interaction hub for medical institutions, logistics centers, and modular drone platforms, and includes: a security encryption module and a dynamic priority management engine;

[0081] The data acquisition and transmission module collects information on the drone's flight status, material status, and environment, outputs the probability value of packaging damage, torque deviation coefficient, and Pearson correlation coefficient, and transmits the pre-processed data to the integrated information platform.

[0082] The communication module establishes a two-way communication link between the UAV and the integrated information platform, ground base station and other related equipment.

[0083] The intelligent decision-making module constructs a hierarchical path planner, including a strategic layer, a tactical layer, and an emergency layer, and constructs a suitability function with path length, dynamic level, and battery health as parameters.

[0084] The data closed-loop update module updates the MEWS score and the timeliness urgency factor verification results synchronously through the user terminal's signature confirmation after the drone completes delivery.

[0085] In one embodiment of the present invention, the security encryption module and the dynamic priority management engine include the following steps:

[0086] The security encryption module uses the national standard SM4 algorithm for end-to-end data transmission encryption;

[0087] The priority management engine dynamically adjusts the data channel bandwidth allocation based on the urgency of medical needs and the urgency of logistics timeliness.

[0088] The medical urgency level is determined by automatically calculating a MEWS score based on the user's uploaded medical data, and a grading system is set accordingly.

[0089] If 0 < MEWS score ≤ 4, it indicates that the medical emergency level is Level 3.

[0090] If 5 ≤ MEWS score ≤ 6, the medical emergency level is classified as Level 2.

[0091] If the MEWS score is ≥7, it indicates that the medical urgency level is Medical Level 1.

[0092] Specifically, a 128-bit encryption key is generated using the national standard SM4 algorithm and distributed to each node in the system through a secure channel. A unique 128-bit initialization vector is generated for each encryption, ensuring that the same plaintext will result in different encryption results. The data to be transmitted is divided into 128-bit blocks, with padding for any insufficient blocks. Each data block is encrypted using the ECB mode of the SM4 algorithm. For higher security, the CBC mode is used, with the ciphertext of the previous block serving as the IV of the next block. The encrypted data is transmitted to the receiving end through a secure channel. The receiving end verifies the key consistency, ensures decryption authorization, and decrypts the ciphertext using the same SM4 algorithm and IV to recover the original data.

[0093] Medical data is collected from user terminals and uploaded to a comprehensive information platform. The MEWS score is automatically calculated according to preset scoring rules. The MEWS score is obtained by adding the scores of various parameters, including heart rate, respiratory rate, systolic blood pressure, body temperature, and level of consciousness. Based on the range of the MEWS score, the medical emergency level is divided into three levels: Level 3, Level 2, and Level 1, with Level 1 being the most urgent and decreasing sequentially, and Level 3 being the least urgent.

[0094] In one embodiment of the present invention, the urgency of logistics timeliness includes the following steps:

[0095] Establish a mapping table between material type and timeliness attribute, where the critical time value corresponding to the transportation material type is determined based on the material's biological characteristics, specifically:

[0096] Blood products: Setting a critical time ≤120 minutes corresponds to an urgency coefficient of 0.9; Vaccines: set a critical time. ≤180 minutes corresponds to an urgency coefficient of 0.8; for routine medicines: set a critical time. ≤240 minutes corresponds to an emergency coefficient of 0.5; the critical time value is determined based on the biological characteristics of the materials.

[0097] Real-time acquisition of remaining transportation time for supplies, and calculation of urgency factors, specifically:

[0098]

[0099] in, As a factor of timeliness and urgency, The remaining transportation time for the supplies. The critical time corresponding to the type of transported goods;

[0100] The corresponding timeliness urgency coefficient and timeliness urgency factor are weighted and added together to obtain a comprehensive material urgency score, specifically:

[0101]

[0102] in, To score the overall urgency of the supplies, As the urgency factor, As a factor of timeliness and urgency, and These are the corresponding weighting coefficients;

[0103] Based on the comprehensive urgency level score of supplies, a tiered system is established:

[0104] If 0 < If the value is less than 0.4, the urgency level of the supplies is Level 3.

[0105] If 0.4≤ If the value is less than 0.7, the urgency level of the supplies is classified as Level 2.

[0106] like If the value is ≥0.7, the urgency level of the supplies is Level 1.

[0107] The combined priority level is calculated using a weighted scoring method.

[0108] Specifically, the remaining transportation time of supplies is obtained in real time, and a timeliness urgency factor is calculated. This calculated urgency factor is then weighted and added to a timeliness urgency coefficient to obtain a comprehensive supply urgency score. Based on this comprehensive score, supplies are classified into three urgency levels: Level 3, Level 2, and Level 1, with Level 1 being the highest urgency, decreasing sequentially to Level 3 being the lowest. and These are the corresponding weighting coefficients, with values ​​of 0.6 and 0.4 respectively.

[0109] In one embodiment of the present invention, the joint priority level includes the following steps:

[0110] Define a joint priority matrix based on the urgency of medical care and supplies, and dynamically allocate data channel bandwidth;

[0111] The emergency levels of medical care and supplies are updated in real time based on the MEWS score and the comprehensive emergency level score.

[0112] The bandwidth allocation algorithm uses a weighted scoring method to calculate the joint priority level. The urgency of medical and material needs is converted into corresponding numerical values ​​based on the level: level three corresponds to a numerical value of 1, level two to 2, and level one to 3. Specifically:

[0113]

[0114] in, As a joint priority level, For medical emergencies, Depending on the urgency of the supplies, and For the corresponding weighting coefficients, and The values ​​are all 1;

[0115] The bandwidth allocation ratio is mapped to the joint priority level as follows:

[0116] like If ≥5, the bandwidth allocation ratio is 80%; if 3 < If <5, then the bandwidth allocation ratio is 60%; if If the value is ≤3, then the bandwidth allocation ratio is 40%.

[0117] Specifically, The corresponding value is 0.6. The corresponding value is 0.4; every fixed time interval, such as 5 seconds, the joint priority level is recalculated and the bandwidth allocation is updated; it also includes a conflict handling mechanism, including the medical priority principle and time-slice scheduling. The medical priority principle is: when there is a conflict between the urgency of medical care and supplies, medical data transmission is given priority; the time-slice scheduling is: for data with the same priority, such as level 2 medical care + level 2 supplies, time-slice round-robin transmission is used to avoid resource contention.

[0118] In one embodiment of the present invention, the data acquisition and transmission module includes:

[0119] The UAV receives mission instructions from the integrated information platform and takes off from the ground base station according to the preset takeoff procedure. During the flight, the data acquisition and transmission module begins to collect flight status data, material status data and environmental information of the UAV in real time.

[0120] The images of the packaging materials captured by the camera are input into the pre-trained YOLOv5 model, which outputs the probability value of packaging damage. If the probability value of packaging damage is greater than or equal to the preset damage threshold, it is judged as a risk of damage and the emergency retransmission mechanism is triggered. The data is uploaded to the data integration information platform through the SM4 encrypted channel first.

[0121] Meanwhile, a battery health predictor is built based on extended Kalman filtering, taking historical data of voltage, current and temperature as input, and outputting the drone battery health status; MEMS gyroscope and visual SLAM data are fused to output six-degree-of-freedom pose; and the torque deviation coefficient of the dynamic load balancer is calculated based on the weight distribution of the cargo.

[0122] Establish a correlation model between the vibration spectrum of materials and the attitude angle of the UAV, and preset a correlation coefficient threshold. If the correlation coefficient is lower than the preset correlation coefficient threshold, the vibration is determined to be abnormal and the vibration optimization command is triggered.

[0123] By using a federated learning framework, local monitoring data is compared with failure modes in a comprehensive information platform to generate predictive maintenance recommendations.

[0124] Specifically, the UAV receives task indicators from the integrated information platform via its communication module. The UAV takes off from the ground base station according to a preset program and enters a predetermined flight path. It collects its six-degree-of-freedom pose using sensors such as MEMS gyroscopes, accelerometers, and GPS; captures images of the packaged goods using a camera; collects the vibration spectrum of the goods using a vibration sensor; and collects environmental data such as temperature, humidity, and air pressure using a meteorological sensor. Simultaneously, it records the UAV's battery voltage, current, and temperature parameters in real time. This UAV monitoring data, including battery health, vibration spectrum, and pose, is stored as a local dataset. A pre-trained federated learning model is loaded and distributed through the integrated information platform. The local dataset is input into the model, which outputs predictive maintenance suggestions, such as battery replacement and motor repair. The local model update parameters are uploaded to the integrated information platform via an SM4 encrypted channel. The integrated information platform aggregates data from multiple UAVs, updates the global model, and distributes it.

[0125] In one embodiment of the present invention, the packaging damage probability value includes the following steps:

[0126] The training data for the YOLOv5 model includes images of damaged packaging materials in multiple scenarios, and an attention mechanism is introduced.

[0127] The probability value of packaging damage is calculated using a confidence-weighted average, specifically as follows:

[0128]

[0129] in, This represents the probability value of packaging damage. For the first The confidence weight of each detection box. This represents the total number of detection frames. The first output of the YOLOv5 model The probability of damage corresponding to each detection box.

[0130] Specifically, images of material packaging from multiple scenarios are collected, including intact and damaged packaging. Damaged areas are labeled using a labeling tool to generate YOLOv5 format annotation files. An attention mechanism is integrated into the YOLOv5 model to enhance its sensitivity to damaged areas. The YOLOv5 model is then used for image recognition of material packaging. The labeled data is used as a training set to train the model. The trained model outputs the probability value of packaging damage. The trained model is converted to ONNX or TensorRT format and deployed to the UAV's onboard processor. If the probability value of packaging damage is greater than or equal to a preset damage threshold, it is considered a damage risk, triggering an emergency retransmission mechanism. Data is uploaded to the data processing module via the SM4 encrypted channel first. If the preset damage threshold is met, a test set of 10,000 images of material packaging from multiple scenarios is constructed, covering intact packaging and various typical damage types. The model performance is evaluated under different probability thresholds, and precision and recall are recorded. The threshold point that maximizes the F1-Score is selected.

[0131] In one embodiment of the present invention, establishing the correlation model between the vibration spectrum of the material and the attitude angle of the UAV includes the following steps:

[0132] The correlation model uses wavelet packet decomposition to extract the vibration spectrum features of the data and performs Pearson correlation analysis with the UAV attitude angle, specifically:

[0133]

[0134] in, The Pearson correlation coefficient is used. The first characteristic of the vibration spectrum of materials One observation value, The first of the attitude angular acceleration of the UAV One observation value, This represents the mean value of the vibration spectrum characteristics of the material. This represents the mean of the angular acceleration of the UAV. This represents the total number of observations.

[0135] In one embodiment of the present invention, the torque deviation coefficient includes the following steps:

[0136] The torque deviation coefficient is specifically:

[0137]

[0138] in, This is the torque deviation coefficient. For the first The weight of each item For the first The distance from the center of gravity of the supplies to the center of the drone. For the first The angle between the supplies and the drone's longitudinal axis. This represents the total quantity of supplies.

[0139] Specifically, by introducing and Orientation projection weights are essentially weighting factors in the decomposition of torque directions. They are non-independent vectors that distinguish the torque influence of materials in the forward / backward and left / right directions of the drone. For example, taking the drone's forward direction (vertical axis) as 0° and clockwise rotation as the positive direction, materials behind the drone have a greater impact on the pitch torque. This is measured in real-time by weight sensors inside the cargo hold. The results were obtained through a visual ranging system combined with cargo hold coordinate system calibration. Combined with the included angle The torque deviation coefficient is calculated.

[0140] In one embodiment of the present invention, the intelligent decision-making module includes:

[0141] The integrated information platform pre-marks high-risk areas and assigns them preset sensitivity levels. The drone dynamically adjusts the area level based on the current torque deviation coefficient.

[0142] The dynamic level is obtained by multiplying the torque deviation coefficient by the preset sensitivity level;

[0143] The fitness function is specifically as follows:

[0144]

[0145] in, For the fitness function, For path length, For dynamic levels, For battery health, , and These are the maximum path length, maximum dynamic level, and maximum battery health, respectively. , and These are the corresponding weight coefficients;

[0146] The tactical layer adjustment includes updating the obstacle map through visual SLAM and MEMS data, and triggering a shock absorption command when the Pearson correlation coefficient is less than a preset correlation coefficient threshold.

[0147] The triggering conditions for the emergency response layer include: MEWS score ≥ 7, packaging damage probability value ≥ preset damage threshold, battery health < 20%, and torque deviation coefficient ≥ preset deviation threshold. Specifically, in the GIS map of the integrated information platform, the following high-risk areas are marked: terrain-sensitive areas (mountains with slope > 15°) and valleys (crosswinds ≥ 8), meteorological-sensitive areas (areas with frequent severe convective weather), and man-made-sensitive areas (areas with dense high-voltage power lines and no-fly zones). Preset sensitivity levels are assigned as follows: flat cities, corresponding to low-risk areas, are assigned a preset sensitivity level of 1.0; hilly areas and areas with moderate winds, corresponding to medium-risk areas, are assigned a preset sensitivity level of 2.0; high mountains and areas with strong winds, corresponding to high-risk areas, are assigned a preset sensitivity level of 3.0. The current torque deviation coefficient is obtained in real time to calculate the dynamic risk level; the weights are dynamically adjusted through federated learning to obtain the result. , and These correspond to values ​​of 0.4, 0.4, and 0.2 respectively. The strategic-layer path generation takes GIS map nodes, a suitability function, and the current dynamic level as input, and outputs the Top 3 Pareto optimal paths generated by an improved genetic algorithm. A Pareto optimal path refers to a path in a multi-objective optimization problem where no single path can further optimize a particular objective (e.g., path length, dynamic level, or battery health) without degrading the performance of other objectives. "Top 3" indicates that the algorithm outputs three paths that perform best in the Pareto front, allowing the system to select and execute paths based on real-time situations (e.g., weather changes, sudden obstacles). These paths achieve an optimal balance between path length, flight safety, and energy consumption, enhancing the system's performance in complex environments. The system aims to improve the adaptability and robustness of the system. The preset correlation coefficient threshold is a critical value determined through correlation analysis between the vibration spectrum and flight attitude. It is used to determine whether abnormal vibrations exist during material transportation. By collecting 1000 sets of vibration-attitude data pairs, including the vibration spectrum of the material and the attitude angle of the UAV, the Pearson correlation coefficient is calculated for each set of data. Real vibration anomalies are labeled, and the ROC curve of the Pearson correlation coefficient and the anomaly events is plotted. Based on the ROC curve, the optimal equilibrium point (maximizing the Youden exponent) is determined, and the preset correlation coefficient threshold is established. The preset deviation threshold is determined by applying a progressive off-center load torque to the UAV in a wind tunnel laboratory, recording the runaway critical point, and taking 80% of the critical value as the threshold.

[0148] In one embodiment of the present invention, the data closed-loop update module includes:

[0149] Once the drone arrives at the target location, the user terminal confirms receipt, the MEWS score is updated synchronously, and a QR code is scanned to verify whether the timeliness and urgency factor has been met.

[0150] At the same time, the data is sent back to the integrated information platform to update the material type-timeliness mapping table and retrain the YOLOv5 model.

[0151] Specifically, after the drone arrives at the target location, the system performs the following closed-loop signing operation: confirming the signatory's authority through the user's terminal's biometric identification or SMS verification code; if the transported supplies are emergency medicines, the user needs to input the patient's latest vital signs; the system automatically calculates and updates the MEWS score and marks the timeliness verification status, i.e., meeting the standard or exceeding the time limit; scanning the encrypted QR code on the supplies packaging, decoding it to obtain the following information, including the type of supplies, the critical time, and the actual transportation time; the signing result is encrypted using the national cryptographic SM4 algorithm and transmitted back to the integrated information platform via the 4G / 5G network; the integrated information platform receives the following data transmitted back by each drone, including packaging damage detection results, transportation environment parameters, and the deviation between the timeliness verification results and the actual transportation time; each edge node (drone) uploads the local model gradient (not the raw data); the central server aggregates and generates a global model, focusing on optimizing the YOLOv5 damage detection confidence threshold and the weight coefficient for calculating the timeliness urgency factor; and adjusting the critical time of the supplies type-timeliness mapping table based on historical transportation data.

[0152] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. An intelligent health monitoring and material delivery system based on unmanned aerial vehicles (UAVs), characterized in that, It consists of an integrated information platform, a drone platform, a data acquisition and transmission module, a communication module, an intelligent decision-making module, a data closed-loop update module, a ground base station, and a user terminal; The integrated information platform constitutes a distributed data interaction hub for medical institutions, logistics centers, and modular drone platforms, and includes: a security encryption module and a dynamic priority management engine; The data acquisition and transmission module collects information on the drone's flight status, material status, and environment, outputs the probability value of packaging damage, torque deviation coefficient, and Pearson correlation coefficient, and transmits the pre-processed data to the integrated information platform. The communication module establishes a two-way communication link between the UAV and the integrated information platform, ground base station and other related equipment. The intelligent decision-making module constructs a hierarchical path planner, including a strategic layer, a tactical layer, and an emergency layer, and constructs a suitability function with path length, dynamic level, and battery health as parameters. The data closed-loop update module updates the MEWS score and the timeliness urgency factor verification results synchronously through the user terminal's signature confirmation after the drone completes delivery.

2. The intelligent health monitoring and material delivery system based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The security encryption module and dynamic priority management engine include the following steps: The security encryption module uses the national standard SM4 algorithm for end-to-end data transmission encryption; The priority management engine dynamically adjusts the data channel bandwidth allocation based on the urgency of medical needs and the urgency of logistics timeliness. The medical urgency level is determined by automatically calculating a MEWS score based on the user's uploaded medical data, and a grading system is set accordingly. If 0 < MEWS score ≤ 4, it indicates that the medical emergency level is Level 3. If 5 ≤ MEWS score ≤ 6, the medical emergency level is classified as Level 2. If the MEWS score is ≥7, it indicates that the medical urgency level is Medical Level 1.

3. The intelligent health monitoring and material delivery system based on unmanned aerial vehicles (UAVs) according to claim 2, characterized in that, The urgency level of the logistics delivery includes the following steps: Establish a mapping table between material type and timeliness attribute, where the critical time value corresponding to the transportation material type is determined based on the material's biological characteristics, specifically: Blood products: Setting a critical time ≤120 minutes corresponds to an urgency coefficient of 0.9; Vaccines: set a critical time. ≤180 minutes corresponds to an urgency coefficient of 0.8; for routine medicines: set a critical time. ≤240 minutes corresponds to an emergency coefficient of 0.5; the critical time value is determined based on the biological characteristics of the materials. Real-time acquisition of remaining transportation time for supplies, and calculation of urgency factors, specifically: in, As a factor of timeliness and urgency, The remaining transportation time for the supplies. The critical time corresponding to the type of transported goods; The corresponding timeliness urgency coefficient and timeliness urgency factor are weighted and added together to obtain a comprehensive material urgency score, specifically: in, To score the overall urgency of the supplies, As the urgency factor, As a factor of timeliness and urgency, and These are the corresponding weighting coefficients; Based on the comprehensive urgency level score of supplies, a tiered system is established: If 0 < If the value is less than 0.4, the urgency level of the supplies is Level 3. If 0.4≤ If the value is less than 0.7, the urgency level of the supplies is classified as Level 2. like If the value is ≥0.7, the urgency level of the supplies is Level 1; The combined priority level is calculated using a weighted scoring method.

4. The intelligent health monitoring and material delivery system based on unmanned aerial vehicles (UAVs) according to claim 3, characterized in that, The joint priority level includes the following steps: Define a joint priority matrix based on the urgency of medical care and supplies, and dynamically allocate data channel bandwidth. The emergency levels of medical care and supplies are updated in real time based on the MEWS score and the comprehensive emergency level score. The bandwidth allocation algorithm uses a weighted scoring method to calculate the joint priority level. The urgency of medical and material needs is converted into corresponding numerical values ​​based on the level: level three corresponds to a numerical value of 1, level two to 2, and level one to 3. Specifically: in, As a joint priority level, For medical emergencies, Depending on the urgency of the supplies, and For the corresponding weighting coefficients, and The values ​​are all 1; The bandwidth allocation ratio is mapped to the joint priority level as follows: like If ≥5, the bandwidth allocation ratio is 80%; if 3 < If <5, then the bandwidth allocation ratio is 60%; if If the value is ≤3, then the bandwidth allocation ratio is 40%.

5. The intelligent health monitoring and material delivery system based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The data acquisition and transmission module includes: The UAV receives mission instructions from the integrated information platform and takes off from the ground base station according to the preset takeoff procedure. During the flight, the data acquisition and transmission module begins to collect flight status data, material status data and environmental information of the UAV in real time. The images of the packaging materials captured by the camera are input into the pre-trained YOLOv5 model, which outputs the probability value of packaging damage. If the probability value of packaging damage is greater than or equal to the preset damage threshold, it is judged as a risk of damage and the emergency retransmission mechanism is triggered. The data is uploaded to the data integration information platform through the SM4 encrypted channel first. Meanwhile, a battery health predictor is built based on extended Kalman filtering, taking historical data of voltage, current and temperature as input, and outputting the drone battery health status; MEMS gyroscope and visual SLAM data are fused to output six-degree-of-freedom pose; and the torque deviation coefficient of the dynamic load balancer is calculated based on the weight distribution of the cargo. Establish a correlation model between the vibration spectrum of materials and the attitude angle of the UAV, and preset a correlation coefficient threshold. If the correlation coefficient is lower than the preset correlation coefficient threshold, the vibration is determined to be abnormal and the vibration optimization command is triggered. By using a federated learning framework, local monitoring data is compared with failure modes in a comprehensive information platform to generate predictive maintenance recommendations.

6. The intelligent health monitoring and material delivery system based on unmanned aerial vehicles (UAVs) according to claim 5, characterized in that, The probability value of packaging damage includes the following steps: The training data for the YOLOv5 model includes images of damaged packaging materials in multiple scenarios, and an attention mechanism is introduced. The probability value of packaging damage is calculated using a confidence-weighted average, specifically as follows: in, This represents the probability value of packaging damage. For the first The confidence weight of each detection box. This represents the total number of detection frames. The first output of the YOLOv5 model The probability of damage corresponding to each detection box.

7. The intelligent health monitoring and material delivery system based on unmanned aerial vehicles (UAVs) according to claim 5, characterized in that, The process of establishing the correlation model between the vibration spectrum of materials and the attitude angle of the UAV includes the following steps: The correlation model uses wavelet packet decomposition to extract the vibration spectrum features of the data and performs Pearson correlation analysis with the UAV attitude angle, specifically: in, The Pearson correlation coefficient is used. The first characteristic of the vibration spectrum of materials One observation value, The first of the attitude angular acceleration of the UAV One observation value, This represents the mean value of the vibration spectrum characteristics of the material. This represents the mean of the angular acceleration of the UAV. This represents the total number of observations.

8. The intelligent health monitoring and material delivery system based on unmanned aerial vehicles according to claim 5, characterized in that, The torque deviation coefficient includes the following steps: The torque deviation coefficient is specifically: in, This is the torque deviation coefficient. For the first The weight of each item For the first The distance from the center of gravity of the supplies to the center of the drone. For the first The angle between the supplies and the drone's longitudinal axis. This represents the total quantity of supplies.

9. The intelligent health monitoring and material delivery system based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The intelligent decision-making module includes: The integrated information platform pre-marks high-risk areas and assigns them preset sensitivity levels. The drone dynamically adjusts the area level based on the current torque deviation coefficient. The dynamic level is obtained by multiplying the torque deviation coefficient by the preset sensitivity level; The fitness function is specifically as follows: in, For the fitness function, For path length, For dynamic levels, For battery health, , and These are the maximum path length, maximum dynamic level, and maximum battery health, respectively. , and These are the corresponding weight coefficients; The tactical layer adjustment includes updating the obstacle map through visual SLAM and MEMS data, and triggering a shock absorption command when the Pearson correlation coefficient is less than a preset correlation coefficient threshold. The triggering conditions for the emergency response include: MEWS score ≥ 7, packaging damage probability value ≥ preset damage threshold, battery health < 20%, and torque deviation coefficient ≥ preset deviation threshold.

10. The intelligent health monitoring and material delivery system based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The data closed-loop update module includes: Once the drone arrives at the target location, the user terminal confirms receipt, the MEWS score is updated synchronously, and the user scans a QR code to verify whether the timeliness and urgency factor has been met. At the same time, the data is sent back to the integrated information platform to update the material type-timeliness mapping table and retrain the YOLOv5 model.

Images