Unmanned aerial vehicle harmful gas distribution monitoring system based on cloud platform

Abstract

The invention relates to the technical field of unmanned aerial vehicle autonomous control or cloud-side cooperative robot control, in particular to an unmanned aerial vehicle harmful gas distribution monitoring system based on a cloud platform, and the system comprises a communication quality analysis unit which generates a communication quality comprehensive index; the decision authorization migration unit generates a decision authorization factor; the cloud fusion updating unit generates a cloud planning waypoint; the airborne data processing unit generates airborne autonomous waypoints and data retention factors; the collaborative path planning unit is used for performing weighted fusion on the cloud planning waypoint and the airborne autonomous waypoint when the decision authorization factor is lower than an autonomous mode activation threshold value, and generating a final execution waypoint; when the decision authorization factor is greater than or equal to the autonomous mode activation threshold, setting the cloud planning waypoint as a final execution waypoint; according to the invention, the inherent contradiction between cloud centralized control and airborne autonomous detection is solved, and the overall toughness and efficient execution of a monitoring task are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle autonomous control or cloud edge collaborative robot control, in particular to an unmanned aerial vehicle harmful gas distribution monitoring system based on a cloud platform. BACKGROUND

[0002] With the wide application of unmanned aerial vehicles in the field of harmful gas monitoring, how to ensure that they can efficiently and reliably perform tasks in complex environments has become a key technical problem. Unmanned aerial vehicle monitoring systems generally rely on cloud platforms for data aggregation and global path planning to build a complete gas distribution model. In actual operations, the communication link between the unmanned aerial vehicle and the cloud platform is easily affected by obstacles, electromagnetic interference, and long distances, and presents a dynamic and unstable characteristic. This unstable communication environment leads to a fundamental technical contradiction: on the one hand, the system needs to ensure the immediacy of data upload to quickly update the global model on the cloud; on the other hand, it also needs to ensure the autonomy of the unmanned aerial vehicle in poor or interrupted communication to continue detecting local key areas. Traditional control strategies usually adopt a binary opposition mode, i.e., hard switching between two states of complete connection or complete disconnection. This mode has significant limitations: when the communication quality decreases but is not completely interrupted, the system lacks a flexible adjustment mechanism, or it misses the opportunity to detect due to adherence to outdated cloud instructions, or it deviates from the global task target due to premature switching to a completely autonomous mode. This rigid control logic leads to insufficient resilience of the system in a dynamic communication environment, low data collection efficiency, and delays in the convergence speed of the global situation model, making it difficult to meet the needs of high-standard monitoring tasks. SUMMARY

[0003] To solve the above technical problems, the present application provides an unmanned aerial vehicle harmful gas distribution monitoring system based on a cloud platform. Specifically, the technical solution of the present application is as follows: An unmanned aerial vehicle harmful gas distribution monitoring system based on a cloud platform, comprising: A communication quality analysis unit for real-time acquisition of communication link parameters between the unmanned aerial vehicle and the cloud platform, and generation of a communication quality comprehensive index; A decision authorization migration unit for receiving the communication quality comprehensive index and generating a decision authorization factor; A cloud fusion update unit for receiving upload data packets, updating the cloud global model, and generating cloud planning waypoints; An on-board data processing unit for generating on-board autonomous waypoints and data retention factors when the decision authorization factor is lower than a preset autonomous mode activation threshold; The cooperative path planning unit is configured to, when the decision authorization factor is lower than the autonomous mode activation threshold, generate a final execution waypoint by weighting and fusing the cloud planning waypoint and the onboard autonomous waypoint according to the decision authorization factor; and when the decision authorization factor is greater than or equal to the autonomous mode activation threshold, set the cloud planning waypoint as the final execution waypoint.

[0004] Preferably, the communication quality analysis unit generates the communication quality comprehensive index by the following process: The communication link parameters include real-time communication bandwidth, data transmission delay and data packet loss rate. The real-time communication bandwidth, data transmission delay and data packet loss rate are normalized. The communication quality comprehensive index is obtained by linearly weighting and summing the normalized parameters.

[0005] Preferably, the decision authorization migration unit generates the decision authorization factor by the following process: The communication quality comprehensive index is substituted into a preset logistic function for nonlinear mapping processing to obtain the decision authorization factor.

[0006] Preferably, the onboard data processing unit generates the onboard autonomous waypoint by the following process: The gas concentration and the gas concentration gradient are obtained and weighted and summed to obtain a monitoring interest index. A limited three-dimensional spatial neighborhood centered on the current position of the unmanned aerial vehicle is generated. A candidate point in the limited three-dimensional spatial neighborhood that maximizes the monitoring interest index is set as the onboard autonomous waypoint.

[0007] Preferably, the onboard data processing unit is further configured to: The data retention factor is obtained by calculation based on the monitoring interest index through a bounded linear model.

[0008] Preferably, the cloud fusion update unit updates the cloud global model by the following process: The data delay and the data retention factor are obtained from the uploaded data packet and substituted into an exponential decay model to calculate an information decay factor. The information decay factor is used as a dynamic weight, and an exponential weighted moving average algorithm is adopted to perform weighted average processing on the global model before updating and the uploaded data packet to obtain an updated global model.

[0009] Preferably, the cooperative path planning unit performs weighting fusion by the following process: The decision authorization factor is used as the weight of the cloud planning waypoint. The difference between 1 and the decision authorization factor is used as the weight of the onboard autonomous waypoint. The position vectors of the cloud planning waypoints and the airborne autonomous waypoints are weighted and summed to obtain a final execution waypoint.

[0010] Compared with the prior art, the present application has the following beneficial effects: 1. The present application initiates a closed-loop adaptive control architecture, which solves the inherent contradiction between cloud centralized control and airborne autonomous detection. It evaluates multi-dimensional communication parameters comprehensively, and uses a logistic function to realize nonlinear smooth migration of decision-making power, thereby eliminating the problem of control mode oscillation caused by traditional threshold switching. This design greatly improves the control stability and behavior predictability of the unmanned aerial vehicle in a complex and dynamic network environment, ensuring the overall resilience and efficient execution of the monitoring task. 2. The present application establishes a cooperative path planning mechanism that takes into account global instructions and local findings. When the communication quality decreases, the system does not simply interrupt the cloud task, but dynamically weights and fuses the waypoints planned by the cloud and the airborne autonomous system according to the decision-making power factor. This design directly quantifies the abstract decision-making power into the physical path selection of the unmanned aerial vehicle, ensuring that its flight path always tends to be optimal under any communication condition, and exhibits unprecedented task continuity and environmental adaptability. 3. The present application endows the unmanned aerial vehicle with high autonomy and intelligence under limited communication, ensuring the lossless collection of key information. The airborne processing unit can autonomously identify and fly to the area with the highest information value by constructing a monitoring interest index, ensuring that the core task can still be efficiently performed without cloud instructions. At the same time, the system uses intelligent data condensation and optimal record mechanism, enabling the unmanned aerial vehicle to transform from a passive collection terminal to an active intelligent probe, significantly improving its autonomous operation ability and the value density of the collected data. 4. The present application establishes a robust asynchronous data fusion and update mechanism to ensure the long-term accuracy of the cloud global model. After receiving the data from the unmanned aerial vehicle, the cloud uses a unique information decay factor to scientifically evaluate the reliability of late or incomplete data, and uses an exponential weighted moving average algorithm for model updating. This mechanism can intelligently reduce the influence of outdated and incomplete data, effectively avoiding data pollution, and ensuring the stable convergence and long-term reliability of the global model even under harsh conditions of frequent disconnection. BRIEF DESCRIPTION OF DRAWINGS

[0011] The present application will be further explained in conjunction with the accompanying drawings and examples: Figure 1 is a structural block diagram of the system of the present application. DETAILED DESCRIPTION

[0012] To make the purpose, technical scheme and advantages of the present application clearer and more apparent, the present application will be further described in detail in conjunction with specific examples.

[0013] Example 1: Referring to Figure 1 The application discloses a cloud platform-based unmanned aerial vehicle harmful gas distribution monitoring system, which comprises the following units in cooperation: A communication quality analysis unit is configured to collect communication link parameters between the unmanned aerial vehicle and the cloud platform in real time and generate a communication quality comprehensive index; A decision authorization migration unit is configured to receive the communication quality comprehensive index and generate a decision authorization factor; A cloud fusion update unit is configured to receive an uploaded data packet, update a cloud global model, and generate a cloud planning flight point; An on-board data processing unit is configured to generate an on-board autonomous flight point and a data retention factor when the decision authorization factor is lower than a preset autonomous mode activation threshold; A cooperative path planning unit is configured to, when the decision authorization factor is lower than the autonomous mode activation threshold, weight and fuse the cloud planning flight point and the on-board autonomous flight point according to the decision authorization factor to generate a final execution flight point; and when the decision authorization factor is greater than or equal to the autonomous mode activation threshold, set the cloud planning flight point as the final execution flight point The application discloses a cloud platform-based unmanned aerial vehicle harmful gas distribution monitoring system, which has the core purpose of solving the fundamental conflict between data uploading immediacy and task execution autonomy of the unmanned aerial vehicle caused by dynamic changes in a communication environment; the application is innovative in that a cloud-edge cooperative control architecture is constructed, which can perform smooth and dynamic decision right migration between centralized control in the cloud and on-board autonomous intelligence according to real-time communication quality, thereby ensuring rapid convergence of a global data model and realizing lossless and efficient collection of local key information; In the embodiment, the system comprises the following units in cooperation: The communication quality analysis unit is configured to perform real-time and accurate quantitative evaluation on an unreliable communication link between the unmanned aerial vehicle and the cloud platform; the unit continuously collects key performance parameters of the communication link and integrates the parameters into a single, standardized comprehensive index, thereby providing an objective and reliable basis for subsequent decision right migration; The decision authorization migration unit is configured to convert the quantitative communication quality in the previous step into a decision authorization in control logic; in the embodiment, the unit receives the communication quality comprehensive index generated by the communication quality analysis unit and generates a core control variable, i.e., a decision authorization factor, through a nonlinear mapping mechanism; the value range of the factor is designed to be between 0 and 1, which intuitively represents the allocation proportion of the decision right from the unmanned aerial vehicle to the cloud platform; The purpose of the cloud-based fusion update unit is to maintain an authoritative and global map of the distribution of hazardous gases, i.e., a cloud-based global model, and to perform macroscopic path planning based on this model. After receiving data packets intermittently uploaded by the UAV, this unit evaluates the timeliness and completeness of the data and integrates the new data into the existing global model through a weighted algorithm to achieve asynchronous model updates. After the update is completed, it generates one or more cloud-based waypoints to guide the UAV in macroscopic exploration based on the latest global situation. The airborne data processing unit aims to endow the UAV with autonomous intelligence when communication is restricted or interrupted. The activation of this unit is strictly controlled by a decision-granting factor, and activation only occurs when this factor falls below a pre-set autonomous mode activation threshold. The unit only starts working when the threshold value is determined by analyzing historical data. This involves analyzing the task completion rate under different communication qualities and selecting the critical value of the decision authorization factor that achieves the best balance between autonomous detection and global planning as the activation threshold. After activation, it will autonomously analyze the local environmental information detected by the airborne sensors, generate airborne autonomous waypoints to guide the UAV in refined detection, and calculate the data retention factor for data condensation based on the value of the detected information. The collaborative path planning unit aims to act as a decision arbitrator, ensuring a smooth transition and optimal execution of UAV waypoint selection. Its behavior is entirely determined by the decision authorization factor. When communication quality is excellent (i.e., the decision authorization factor is greater than or equal to the autonomous mode activation threshold), the system exhibits a typical cloud-based centralized control mode, where the unit directly sets the cloud-planned waypoint as the final execution waypoint the UAV needs to fly to. Conversely, when communication quality is poor (i.e., the decision authorization factor is below the threshold), the system switches to a cloud-edge collaborative mode. This unit dynamically weights and merges the cloud-planned waypoint and the airborne autonomous waypoint based on the magnitude of the decision authorization factor, generating an execution waypoint that balances global commands and local discovery. Through the organic combination of the above units, this invention realizes a complete closed-loop adaptive control system. It is no longer a binary opposition of connection or disconnection in the traditional sense, but can make stepless intelligent adjustments according to the quality of the communication link. This allows the UAV to act as an efficient data acquisition terminal in the cloud when communication is good, quickly building a global situation. When communication deteriorates, it can seamlessly switch to an autonomous intelligent agent with local problem-solving capabilities, ensuring that continuous detection of high-value areas is not interrupted. In complex communication environments, this invention greatly improves the system resilience, data acquisition efficiency, and model convergence speed of hazardous gas monitoring tasks.

[0014] Example 2: The process by which the communication quality analysis unit generates the comprehensive communication quality index is as follows: obtaining communication link parameters, the communication link parameters including real-time communication bandwidth, data transmission delay and data packet loss rate; normalizing the real-time communication bandwidth, the data transmission delay and the data packet loss rate; obtaining a communication quality comprehensive index by linearly weighting and summing the normalized parameters.

[0015] The embodiment is a specific and optimized implementation of the communication quality analysis unit, and the core is how to scientifically integrate multiple communication parameters with different physical dimensions into a single communication quality comprehensive index Q. Specifically, the communication link parameters obtained by the communication quality analysis unit are explicitly three core indicators: Real-time communication bandwidth, unit: Mbps, which is collected in real time by the performance monitoring interface of the unmanned aerial vehicle communication module. This parameter directly reflects the rate capability of data upload. Data transmission delay, unit: ms, which is calculated by ping-pong test or using a time-stamped response mechanism. This parameter reflects the immediacy of data transmission. Data packet loss rate, unit: %, which is obtained by calculating the ratio of the number of successfully received data packets to the total number of data packets in a certain time window. This parameter reflects the reliability of data transmission. In order to eliminate the differences in physical dimensions of different parameters and make them additive, the unit normalizes the above three parameters respectively. In this embodiment, the ratio normalization method is used, that is, the real-time value is divided by a preset reference value. Based on the normalized parameters, the communication quality comprehensive index is obtained by linearly weighting and summing The internal logic is to integrate multi-dimensional performance evaluation into a single-dimensional evaluation value, and the calculation formula is as follows:

[0016] Wherein, Communication quality comprehensive index, dimensionless, calculated by the formula; Weight coefficient, dimensionless, determined as follows: construct a calibration data set containing N scenes, where each scene i corresponds to a set of communication parameters And a system performance score evaluated by experts or quantified based on task success rate ; Through regression analysis methods such as least squares, solve The weight coefficient that minimizes the cost function The sum of the three values is 1. : reference value, with the same dimension as the corresponding parameter, which is an empirical value set according to the basic communication capability required by the protection task, for example It can be set to a video transmission code rate that meets the minimum requirement. Compared with the prior art which only relies on a single indicator or simple logic, the embodiment constructs a quantitative indicator that can comprehensively, dynamically and finely reflect the comprehensive state of the communication link through linear weighted summation The indicator not only has a clear physical meaning, but also makes the system evaluation model flexible to adapt to specific needs in different scenarios through adjustable weight coefficients, thereby providing a solid and reliable input basis for the accurate migration of subsequent decision authorization, and significantly improving the perception accuracy and adaptability of the system to changes in the network environment.

[0017] Embodiment 3: The process of generating the decision authorization factor by the decision authorization migration unit is as follows: The communication quality comprehensive index is substituted into a preset logistic function for nonlinear mapping processing to obtain the decision authorization factor.

[0018] The embodiment is a specific and optimized implementation of the decision authorization migration unit, and the core lies in how to convert the linear communication quality comprehensive index into a nonlinear decision authorization factor that is more suitable for the smooth transition needs of the control logic. In the embodiment, the received communication quality comprehensive index is substituted into a preset logistic function for nonlinear mapping processing to obtain the decision authorization factor . The technical motivation of using this function is that the migration of decision authority is not a linear process, but a qualitative change process near a critical point. The S-shaped curve characteristics of the logistic function can simulate the smooth transition from one stable state to another, and its calculation formula is as follows:

[0019] wherein, : decision authorization factor, dimensionless, value range (0, 1), calculated by the formula, as the core control variable of all subsequent decisions of the system; : communication quality comprehensive index, dimensionless, calculated by the communication quality analysis unit; : migration slope coefficient, dimensionless, determined together with by calibration. The greater the value, the more aggressive the state transition; otherwise, the more gentle it is; : transition threshold, dimensionless, its calibration method is as follows: collect a series of historical communication quality data and their corresponding optimal control modes, for example, for each set of historical communication data , give a most ideal decision-making authority factor evaluation value at that time by experienced operators or through simulation evaluation , which can reflect the degree of dependence on cloud instructions at that time, find the best fitting data pair and parameters through a nonlinear fitting algorithm; it defines the dividing line between acceptable and unacceptable communication quality.

[0020] This embodiment realizes a flexible decision-making intelligent transition mechanism by introducing a logistic function; it cleverly converts linear communication quality changes into nonlinear control distribution, perfectly solving the problem that traditional threshold methods may cause frequent and drastic switching of system control modes due to slight fluctuations near the critical point; this smooth transition feature greatly enhances the control stability and behavior predictability of the entire system in a dynamic communication environment, ensuring the continuity and efficiency of task execution; To further suppress the control mode jitter that may be caused by the fluctuations of near , this unit can also introduce a hysteresis comparison mechanism; specifically, the system can set two independent transition thresholds and , where , the control logic is optimized as follows: the system only transitions to the cloud dominant mode when rises from an interval lower than and crosses ; and the system only transitions to the onboard autonomous mode when falls from an interval higher than and crosses ; within the hysteresis interval , , the system will keep the control state of the previous moment unchanged, effectively ensuring the smoothness of flight path decision-making.

[0021] Embodiment 4: The process of generating onboard autonomous waypoints by the onboard data processing unit is as follows: obtain the gas concentration and gas concentration gradient, and perform weighted summation to obtain the monitoring interest index; generate a finite three-dimensional spatial neighborhood centered on the current position of the UAV; set the candidate point that maximizes the monitoring interest index within the finite three-dimensional spatial neighborhood as the onboard autonomous waypoint.

[0022] The airborne data processing unit is further configured to: based on the monitoring interest index, calculate a data retention factor by a bounded linear model.

[0023] The embodiment is a specific implementation of the airborne data processing unit; when the decision-making authorization factor λ is lower than a preset autonomous mode activation threshold, the unit is activated, so that the UAV changes from a passive data collector to an active intelligent probe; To guide the UAV to fly autonomously to the most valuable area in the out-of-contact or weakly connected state, the airborne data processing unit will perform the process of generating an airborne autonomous waypoint ; The unit obtains the gas concentration and the gas concentration gradient at the current position, and obtains a monitoring interest index by weighted summation ; the purpose of the index is to unify different physical measurements into a priority score, and to introduce an exploration factor to balance exploitation and exploration, so as to avoid the UAV falling into a local concentration extreme point, and the calculation formula is as follows:

[0024] wherein, is the monitoring interest index; is the corresponding weight; the exploration factor encourages the UAV to explore areas with old information or areas that have not been visited, and the specific calculation formula is as follows: ; in the formula, is the current timestamp, is the last time stamp when the candidate point was visited, is a preset time reference constant for controlling the growth rate of the exploration weight forgetting over time. This method ensures that areas that have not been visited for a long time will have a higher exploration priority; is the exploration weight, and the mechanism encourages the UAV to explore areas with old information or areas that have not been visited while tracking high-concentration areas in a timely manner; : gas concentration and concentration gradient, which is estimated by the airborne sensor by performing a preset micro-probing action, such as a short four-point hovering around the current position and rapid sampling, and performing real-time difference calculation on the collected multiple adjacent concentration values, thereby using finite difference method to estimate the concentration gradient at the current position; : reference benchmark value, which is derived from the environmental background value or the industry safety specification value; The unit takes the current position of the UAV as the center, virtually generates a limited three-dimensional neighborhood S in the memory, for example, a cubic space with the UAV as the center and a side length of 50 meters; the neighborhood defines a set of candidate waypoints that the UAV can reach in the next step; By searching in the neighborhood S, the monitoring interest index of the candidate point that reaches the maximum value is set as the onboard autonomous waypoint The search process can be implemented using a discretized grid search method: first, divide the neighborhood S into a limited number of grid points, then quickly evaluate the value of each grid point, and finally select the optimal one, whose decision logic can be represented as:

[0025] Among them, : onboard autonomous waypoint, position vector, target point of the local optimal path; At the same time, to optimize the limited onboard storage space and prepare for efficient uploading after communication recovery, the onboard data processing unit also calculates the data retention factor based on the monitoring interest index calculated above through a bounded linear model, the calculation formula is:

[0026] Among them, : data retention factor, dimensionless, value range , used to determine the compression rate or discard rate of the current collected data; : monitoring interest index, dimensionless, derived from the previous step calculation; : basic retention rate, dimensionless, derived from the minimum requirement of the task for the integrity of the background data, ensuring that even in low-interest areas, basic information is retained; : interest sensitivity coefficient, dimensionless, an adjustable parameter, derived from the estimation of onboard storage capacity and expected time of disconnection, to balance between information fidelity and storage resource consumption; This embodiment endows the UAV with the ability to make autonomous decisions without cloud-based commands. By constructing a quantified interest model, it enables the UAV to autonomously fly to the area with the highest information value, ensuring that the monitoring task can still be carried out efficiently and purposefully during communication interruptions. An intelligent data condensation mechanism determines the degree of information retention based on information value, realizing the transformation from full recording to selective recording. The synergy of these two mechanisms enables the UAV to not only intelligently plan its path in autonomous mode but also efficiently manage data, greatly improving its autonomous operation capability and data value density in harsh environments. Specifically, when a drone collects a set of data frames, it will use the formula... Calculate the data retention factor This factor will be used as a probability threshold for Bernoulli sampling, and the system sets a random number. Only when Only then is the data frame fully stored in the onboard buffer, awaiting upload. This probability-based sampling method ensures that data in regions of high interest is retained with a high probability, while data in regions of low interest is appropriately sparsified.

[0027] Example 5: The process of updating the cloud-based global model by the cloud-based fusion update unit is as follows: Data latency and data retention factor are obtained from the uploaded data packets and substituted into the exponential decay model to calculate the information decay factor; The information decay factor is used as a dynamic weight, and an exponentially weighted moving average algorithm is used to perform a weighted average processing on the global model before the update and the uploaded data packet to obtain the updated global model.

[0028] This embodiment is a specific implementation of the process of updating the cloud global model M by the cloud fusion update unit; this process is completed after the UAV and the cloud platform resume communication and upload measurement data, timestamps, and data retention factors. data packets It is subsequently triggered; its core is to solve how to reliably integrate these potentially delayed and incomplete late data into the global model. From the uploaded data packet Data acquisition latency With data retention factor ;in, The information decay factor is calculated from the difference between the timestamp within the data packet and the current server time on the cloud platform. These two parameters are then substituted into an exponential decay model to calculate the information decay factor. The value of information diminishes over time and with the incompleteness of the information itself, as calculated by the following formula:

[0029] wherein, : information decay factor, dimensionless, value range (0, 1), quantifies the credibility or weight of the uploaded data packet; : time decay constant, unit: second, which is set based on prior knowledge of the physical diffusion or chemical half-life of the monitoring target, such as a specific harmful gas, representing the characteristic time of the environmental information remaining valid; : data retention factor corresponding to the uploaded data, dimensionless, which is directly read from the uploaded data packet ; The information decay factor just calculated is taken as a dynamic weight, and an exponentially weighted moving average algorithm is used to perform weighted average processing on the global model before updating and the newly uploaded data packet to obtain the updated global model ; the logic of this algorithm is that the generation of the new model should be a weighted average of the old model state and the new input information, and the weight reflects the degree of trust in the new information, and the calculation formula is as follows:

[0030] wherein, : updated / previous global model, physical dimension is concentration field, such as kg / m 3 ; D u : uploaded data packet, physical dimension is the same as M, which is uploaded by the unmanned aerial vehicle; : basic update rate, dimensionless, value range (0, 1), a hyperparameter, which is set according to the degree of dynamic change of the environment, and the value is larger when the environment changes faster; . This embodiment constructs a highly robust asynchronous data fusion mechanism through two-step calculation; its gain effect is that it does not simply overwrite old data with new data, but scientifically evaluates the value of each uploaded data through the δ factor, and dynamically adjusts the proportion of each uploaded data in model updating; this enables the cloud global model to intelligently reduce the weight of old and incomplete data, while increasing the weight of fresh and complete data; even in the case of frequent disconnection of unmanned aerial vehicles and chaotic data upload timing, the long-term stability and accuracy of the cloud global model can still be ensured, effectively avoiding data pollution.

[0031] Embodiment 6: The process of the collaborative path planning unit for weighted fusion is as follows: the decision-making authorization factor is taken as the weight of the cloud planning waypoint; the difference between 1 and the decision-making authorization factor is taken as the weight of the airborne autonomous waypoint; the position vectors of the cloud planning waypoint and the airborne autonomous waypoint are weighted and summed to obtain the final execution waypoint.

[0032] This embodiment is a specific implementation of the process of the collaborative path planning unit for weighted fusion in the cloud-edge collaborative mode; this process is the control core of the entire system, which accurately couples the macro strategy of the cloud and the local tactics of the airborne in mathematics; The process of the collaborative path planning unit for weighted fusion is based on a vector linear combination; the specific implementation is as follows: The decision-making authorization factor generated by the decision-making authorization migration unit is directly taken as the weight of the cloud planning waypoint ; at the same time, is taken as the weight of the airborne autonomous waypoint generated by the airborne data processing unit; the position vectors of the cloud planning waypoint and the airborne autonomous waypoint are weighted and summed to obtain the final execution waypoint , and the calculation formula is as follows

[0033] Among them, : position vector, unit: meter; the formula is consistent in physical dimension; : decision-making authorization factor, dimensionless, which is calculated by the decision-making authorization migration unit; : cloud planning waypoint, position vector, which is calculated by the cloud fusion update unit; : airborne autonomous waypoint, position vector, which is calculated by the airborne data processing unit; This formula closely couples the three core links of the system to form a complete closed-loop control logic; the internal logic of the operation mechanism is that: When the communication quality is excellent ( → 1), the unmanned aerial vehicle strictly executes the instructions issued by the cloud based on the global optimal model; When the communication quality is poor ( → 0), the unmanned aerial vehicle completely makes decisions based on the detection of the surrounding environment and executes the local optimal path; When the communication quality is in an intermediate state ( (0, 1), the behavior of the unmanned aerial vehicle is an organic combination of cloud macroscopic guidance and local discovery, and the waypoint is a balance point between the cloud instruction and the autonomous intention; The embodiment realizes a delicate control right smooth handover mechanism through linear weighted fusion, and has the following remarkable technical effects: the abstract decision right concept is directly mapped to actual and quantified influence on the physical path of the unmanned aerial vehicle, which ensures that the flight path of the unmanned aerial vehicle is optimal under any communication condition, neither misses local detection opportunities due to sticking to outdated cloud instructions, nor deviates from the global task target due to complete autonomy, and the cooperative path planning mechanism makes the whole monitoring system exhibit unprecedented toughness, adaptability and high efficiency in a dynamically uncertain environment.

[0034] The above is only a preferred embodiment of the present application, and is not used to limit the protection scope of the present application; any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0035] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the protection scope of the present application, and although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A cloud-based unmanned aerial vehicle (UAV) hazardous gas distribution monitoring system, characterized in that, include: The communication quality analysis unit is used to collect communication link parameters between the UAV and the cloud platform in real time and generate a comprehensive communication quality index. The decision authorization migration unit is used to receive the comprehensive communication quality index and generate the decision authorization factor; The cloud-based fusion update unit is used to receive uploaded data packets, update the global cloud model, and generate cloud-based planned waypoints. The airborne data processing unit is used to generate airborne autonomous waypoints and data retention factors when the decision authorization factor is lower than the preset autonomous mode activation threshold. The collaborative path planning unit is used to generate the final execution waypoint by weighted fusion of cloud-planned waypoints and airborne autonomous waypoints based on the decision authorization factor when the decision authorization factor is lower than the autonomous mode activation threshold. When the decision authorization factor is greater than or equal to the autonomous mode activation threshold, the cloud-planned waypoint will be set as the final execution waypoint.

2. The cloud-based UAV hazardous gas distribution monitoring system according to claim 1, characterized in that, The process by which the communication quality analysis unit generates the comprehensive communication quality index is as follows: Obtain communication link parameters, including real-time communication bandwidth, data transmission latency, and packet loss rate; Normalize the real-time communication bandwidth, data transmission latency, and packet loss rate; The overall communication quality index is obtained by linearly weighting and summing the normalized parameters.

3. The cloud-based UAV hazardous gas distribution monitoring system according to claim 1, characterized in that, The process by which the decision authorization transfer unit generates decision authorization factors is as follows: The communication quality comprehensive index is substituted into a preset logistic function for nonlinear mapping to obtain the decision authorization factor.

4. The cloud-based UAV hazardous gas distribution monitoring system according to claim 1, characterized in that, The process by which the airborne data processing unit generates airborne autonomous waypoints is as follows: The gas concentration and gas concentration gradient are obtained and then weighted and summed to obtain the monitoring interest index; Generate a finite three-dimensional spatial neighborhood centered on the current location of the drone; Candidate points that maximize the monitoring interest index within a limited three-dimensional spatial neighborhood are designated as airborne autonomous waypoints.

5. The cloud-based UAV hazardous gas distribution monitoring system according to claim 4, characterized in that, The airborne data processing unit is also used for: Based on the monitoring interest index, the data retention factor is calculated using a bounded linear model.

6. The cloud-based UAV hazardous gas distribution monitoring system according to claim 1, characterized in that, The process by which the cloud-based fusion update unit updates the cloud-based global model is as follows: Data latency and data retention factor are obtained from the uploaded data packets and substituted into the exponential decay model to calculate the information decay factor; The information decay factor is used as a dynamic weight, and an exponentially weighted moving average algorithm is used to perform a weighted average processing on the global model before the update and the uploaded data packet to obtain the updated global model.

7. The cloud-based UAV hazardous gas distribution monitoring system according to claim 1, characterized in that, The weighted fusion process of the collaborative path planning unit is as follows: Use decision-making authorization factors as weights for waypoint planning in the cloud; The difference between 1 and the decision authorization factor is used as the weight of the airborne autonomous waypoint; The final executed waypoint is obtained by weighted summation of the position vectors of the cloud-planned waypoints and the airborne autonomous waypoints.

Images