A cooperative monitoring system for unmanned aerial vehicle automatic inspection energy supplement process

Abstract

The application discloses a kind of unmanned vehicle automation inspection energy supplementing process's cooperative monitoring system, it is related to unmanned vehicle automation energy supplementing control technical field, including state space multidimensional perception mapping module, semantic deviation analysis decision module, cooperative monitoring control module, shadow state machine redundancy monitoring module and environment adaptive sensitivity adjustment module.System is through the construction execution module high-dimensional state vector, real-time analysis the deviation between actual running state and ideal model, and when deviation is over limit, control bus topological reconfiguration is implemented, execution timing cooperative adjustment and sliding mode compensation control;While combining shadow state machine carries out semantic consistency check, and according to environmental noise dynamically adjusts monitoring sensitivity.Through multi-module cooperative monitoring and closed-loop compensation control, realize the abnormal perception of unmanned vehicle energy supplementing process, deviation suppression and safety protection, solve the problem that existing automation energy supplementing process is insufficient, environmental interference adaptability is poor and the problem of low running reliability of execution mechanism cooperation.

Description

Technical Field

[0001] This invention relates to the field of automated refueling control technology for unmanned aerial vehicles (UAVs), and more particularly to a collaborative monitoring system for the automated inspection and refueling process of UAVs. Background Technology

[0002] Currently, sequential control logic based on fixed time steps and absolute displacement thresholds is commonly used to control the coordinated movement of the internal lifting platform, rotating parking apron, and battery swapping robotic arm of the automated battery swapping station, thereby ensuring that the energy replenishment operation after long-distance automated inspection is closed in sequence.

[0003] When the control device is deployed in a complex outdoor environment with frequent sandstorms or large temperature differences, the drive mechanism is subjected to alternating aerodynamic loads and asymmetric thermal deformation for a long time. As the components move in the space track, random displacement deviations and nonlinear response delays occur. The sequential control mechanism lacks the ability to monitor the global linkage of high-dimensional multi-channel response states and cannot identify the implicit position loss that accumulates with the advancement of the action sequence. This causes a dynamic mismatch between the timing command stream of the control bus and the actual docking state of heterogeneous components in the physical space. Furthermore, the forced advancement of subsequent components can trigger cumulative mechanical damage risks such as battery interface engagement failure and rigid mechanical impact. Common improvement solutions in the industry focus on changing the stiffness of the physical linkage of the robotic arm, or adding mechanical buffer springs to the transmission base, or even improving the single-point calibration accuracy by adding multiple physical measurement sensors. However, hardware superposition solutions increase the mechanical and electrical topology complexity of the control system and are prone to sensor cascade failures in harsh environments. On the other hand, they cannot eliminate the inaccuracy caused by the discrete characteristics of control data through command calculation.

[0004] From the perspective of software-based collision avoidance solutions such as control algorithms, for example, Chinese invention patent CN114803384B discloses a motion-following collision avoidance method, device, and battery swapping station. By establishing a coordinate system and calculating the relative distance between two moving parts, and combining a preset linear deceleration function to determine the following speed to avoid collision, this solution implicitly relies on the underlying premise that the mechanism's motion trajectory is singular and in an ideal, undisturbed, rigid state. Under harsh variable load conditions such as UAV field inspection and refueling, due to the mechanism's long-term exposure to alternating aerodynamic loads and extreme temperature differences leading to asymmetric thermal deformation, the multi-axis heterogeneous drive units generate highly nonlinear transient response delays and implicit displacement losses in the overlapping space. In this scenario, due to the lack of global linkage perception of multi-dimensional state flows such as drive current surges and spatial phase deviations, relying solely on a single coordinate system distance calculation and static linear deceleration mechanism, it is impossible to capture the step-coupling state between heterogeneous mechanisms across signals. It is easy for the accumulation of front-end deviations to trigger subsequent mechanisms to forcibly advance when the coaxial pose boundary is not met, resulting in rigid impact or engagement failure. Summary of the Invention

[0005] The present invention aims to at least partially solve one of the technical problems in the related art.

[0006] The main objective of this invention is to provide a collaborative monitoring system for the automated inspection and refueling process of unmanned aerial vehicles (UAVs).

[0007] To achieve the above objectives, embodiments of the present invention propose a collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs), comprising: The state space multidimensional perception mapping module is used to collect current signals and position signals from heterogeneous execution modules, and construct a high-dimensional state vector based on the current signals and position signals. The semantic deviation analysis adjudication module is used to calculate the Euclidean distance between the high-dimensional state vector and the ideal model state vector, so as to output the deviation gradient; A collaborative monitoring and control module is used to reconstruct the control bus topology and regulate the startup timing of subsequent execution modules when the deviation gradient exceeds a preset evolution boundary. The collaborative monitoring and control module is further used to: generate logic damping timing parameters based on the travel feedback residual of the non-converged execution module, and write the logic damping timing parameters into the state machine of the subsequent execution module to generate discrete logic yield; activate the sliding mode control operator during the yielding period, solve the sliding mode control equation online, and output nonlinear dynamic deviation compensation parameters limited by the maximum threshold of the safe contact force; write the nonlinear dynamic deviation compensation parameters into the control port register of the current execution module; and release the timing constraints of the state machine of the subsequent execution module in the next control cycle when the travel feedback residual converges to the 0.05mm boundary. The shadow state machine redundancy monitoring module is used to run the virtual ideal process in parallel and verify the actual running state flow with the semantic fingerprint. When there is a semantic mismatch, the system enters a safe reset state. An environment-adaptive sensitivity adjustment module is used to dynamically adjust the sensitivity window of the monitoring operator based on the noise level fed back by the environment sensing module.

[0008] In one embodiment of the present invention, the heterogeneous execution module includes a lifting control module, a rotation control module, and a battery swapping control module; the state space multidimensional perception mapping module is used to synchronously acquire the real-time current data and position signals of each heterogeneous execution module, and construct the high-dimensional state vector through differential processing.

[0009] In one embodiment of the present invention, the semantic bias analysis adjudication module is further configured to be used in the bias gradient When E exceeds the preset evolution boundary, the deviation gradient will be... The evolutionary characteristics of E are compared with a preset interference model library to identify the type of interference source in the current external environment.

[0010] In one embodiment of the present invention, the logic damping timing parameters It has a monotonically positive correlation with the travel feedback residual of the unconverged execution module, and is used to dynamically configure the suspension delay time of the state machine of the subsequent execution module.

[0011] In one embodiment of the present invention, the sliding mode control operator uses the drive loop current differential data, position feedback differential data, and the first-order time derivative of the position feedback differential data as multidimensional system state variables to solve the sliding mode control equations online.

[0012] In one embodiment of the present invention, the control saturation limit corresponding to the maximum threshold of the safe contact force is preset by the background controller; when the nonlinear dynamic deviation compensation parameter exceeds the control saturation limit, the collaborative monitoring and control module will limit the output control parameter to the control saturation limit.

[0013] In one embodiment of the present invention, the method by which the collaborative monitoring and control module writes the nonlinear dynamic deviation compensation parameter into the control port register of the current execution module includes: shortening the control sampling window of the current execution module and increasing the output drive frequency according to the nonlinear dynamic deviation compensation parameter.

[0014] In one embodiment of the present invention, the method by which the collaborative monitoring and control module releases timing constraints includes: when the residual travel feedback reaches the 0.05mm boundary, triggering a microsecond-level hardware timer interrupt, and clearing the suspended delay state in the state machine of the subsequent execution module in the next control cycle.

[0015] In one embodiment of the present invention, the shadow state machine redundancy monitoring module is used to maintain a virtual ideal control state synchronized with the actual power replenishment process and convert the control bus data stream into an operational semantic fingerprint; when the semantic fingerprint of the actual operational state stream does not match the semantic fingerprint corresponding to the virtual ideal control state, bus jitter noise elimination or a safety reset command is issued.

[0016] In one embodiment of the present invention, the environment adaptive sensitivity adjustment module is used to expand the tolerance range of the monitoring operator for non-critical action state transitions and shorten the discrete detection cycle of the core limit state node when the noise level fed back by the environment sensing module exceeds a preset threshold.

[0017] The embodiments of the present invention have the following beneficial effects: 1. In the collaborative monitoring of the automated inspection and recharging process of UAVs, by synchronously collecting the current differential characteristics of multiple heterogeneous execution units and the pulse difference signal of the position encoder, the system transforms multidimensional physical parameters into high-dimensional state vectors and calculates the distance between the vector and the preset ideal semantic model online to obtain the deviation gradient reflecting the spatiotemporal offset of the execution sequence. This cross-signal dynamic perception mechanism can accurately identify the asymmetric lag in the action response of heterogeneous components caused by environmental interference, thereby transforming the traditional single-point physical position verification into a holistic audit of the system's operating pace, effectively avoiding the potential risk of subsequent rigid collisions triggered by the accumulation of front-end position deviations under sequential control logic.

[0018] 2. The collaborative monitoring and control module in the system reconstructs the internal data flow topology in real time based on the acquired deviation gradient. It converts the residual physical travel of the preceding components that are not yet in place into logic damping timing parameters in the background controller and forcibly embeds these parameters into the migration axis of the state machine of the subsequent execution unit. This causes the timing of the control commands issued by the subsequent components to generate discrete logic yield that is monotonically positively correlated with the current physical deviation, thereby creating a spatial safety gap. At the same time, the built-in discrete-time sliding mode surface control law is activated during the yielding period. The sliding mode control equation is solved online using real-time state variables, and the nonlinear dynamic deviation compensation parameters are output under the rigid truncation constraint of the maximum threshold of the safety contact force. When the residual physical travel converges to the critical alignment boundary of 0.05mm, the timing constraint is released in the next 10μs control cycle, and the corrected control command is written. This achieves spatiotemporal synchronization and logic self-healing of heterogeneous actuators under complex interference conditions.

[0019] 3. By running a virtual ideal energy replenishment control process in parallel in the background, the shadow state machine redundant monitoring module verifies the actual physical state flow with the semantic fingerprint output from the background in real time. Through this dual-track spatiotemporal consistency verification mechanism, the control bus data is cross-audited. When a semantic mismatch occurs at a critical node, the system accurately identifies and eliminates interference noise caused by communication bus jitter or sensor signal jumps, or directly forces the control port to enter a preset safe reset state. This constructs a redundant monitoring closed loop independent of the front-end dynamic perception chain, improving the anti-interference performance of the entire control device under harsh operating conditions. Attached Figure Description

[0020] The above-described and additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, in which: Figure 1 This is an information flow architecture diagram of the collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs) proposed in this embodiment of the invention. Figure 2 This is a three-dimensional schematic diagram illustrating the evolution of contact stress with the energy replenishment operation cycle as proposed in an embodiment of the present invention; Figure 3This is a comparison curve of peak contact stress under different equivalent wind speeds proposed in the embodiments of the present invention; Figure 4 This is a convergence curve of the actuator speed characteristics with the sampling time sequence proposed in the embodiments of the present invention; Figure 5 This is a comparison diagram of state-space multidimensional perception mapping tracking features proposed in this embodiment of the invention; Figure 6 This is a comparison chart of the driving current effect of the collaborative monitoring and control module proposed in the embodiments of the present invention. Detailed Implementation

[0021] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0022] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0023] The following describes a collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs) according to an embodiment of the present invention, with reference to the accompanying drawings.

[0024] Example 1 This embodiment provides a collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs). For example... Figure 1 As shown, the system includes: The state space multidimensional perception mapping module is used to collect current signals and position signals from heterogeneous execution modules, and construct a high-dimensional state vector based on the current signals and position signals. The semantic deviation analysis adjudication module is used to calculate the Euclidean distance between the high-dimensional state vector and the ideal model state vector, so as to output the deviation gradient; A collaborative monitoring and control module is used to reconstruct the control bus topology and regulate the startup timing of subsequent execution modules when the deviation gradient exceeds a preset evolution boundary. The collaborative monitoring and control module is further used to: generate logic damping timing parameters based on the travel feedback residual of the non-converged execution module, and write the logic damping timing parameters into the state machine of the subsequent execution module to generate discrete logic yield; activate the sliding mode control operator during the yielding period, solve the sliding mode control equation online, and output nonlinear dynamic deviation compensation parameters limited by the maximum threshold of the safe contact force; write the nonlinear dynamic deviation compensation parameters into the control port register of the current execution module; and release the timing constraints of the state machine of the subsequent execution module in the next control cycle when the travel feedback residual converges to the 0.05mm boundary. The shadow state machine redundancy monitoring module is used to run the virtual ideal process in parallel and verify the actual running state flow with the semantic fingerprint. When there is a semantic mismatch, the system enters a safe reset state. An environment-adaptive sensitivity adjustment module is used to dynamically adjust the sensitivity window of the monitoring operator based on the noise level fed back by the environment sensing module.

[0025] In one embodiment of the present invention, the heterogeneous execution module includes a lifting control module, a rotation control module, and a battery swapping control module; the state space multidimensional perception mapping module is used to synchronously acquire the real-time current data and position signals of each heterogeneous execution module, and construct the high-dimensional state vector through differential processing.

[0026] In one embodiment of the present invention, the semantic bias analysis adjudication module is further configured to be used in the bias gradient When E exceeds the preset evolution boundary, the deviation gradient will be... The evolutionary characteristics of E are compared with a preset interference model library to identify the type of interference source in the current external environment.

[0027] In one embodiment of the present invention, the logic damping timing parameters It has a monotonically positive correlation with the travel feedback residual of the unconverged execution module, and is used to dynamically configure the suspension delay time of the state machine of the subsequent execution module.

[0028] In one embodiment of the present invention, the sliding mode control operator uses the drive loop current differential data, position feedback differential data, and the first-order time derivative of the position feedback differential data as multidimensional system state variables to solve the sliding mode control equations online.

[0029] In one embodiment of the present invention, the control saturation limit corresponding to the maximum threshold of the safe contact force is preset by the background controller; when the nonlinear dynamic deviation compensation parameter exceeds the control saturation limit, the collaborative monitoring and control module will limit the output control parameter to the control saturation limit.

[0030] In one embodiment of the present invention, the method by which the collaborative monitoring and control module writes the nonlinear dynamic deviation compensation parameter into the control port register of the current execution module includes: shortening the control sampling window of the current execution module and increasing the output drive frequency according to the nonlinear dynamic deviation compensation parameter.

[0031] In one embodiment of the present invention, the method by which the collaborative monitoring and control module releases timing constraints includes: when the residual travel feedback reaches the 0.05mm boundary, triggering a microsecond-level hardware timer interrupt, and clearing the suspended delay state in the state machine of the subsequent execution module in the next control cycle.

[0032] In one embodiment of the present invention, the shadow state machine redundancy monitoring module is used to maintain a virtual ideal control state synchronized with the actual power replenishment process and convert the control bus data stream into an operational semantic fingerprint; when the semantic fingerprint of the actual operational state stream does not match the semantic fingerprint corresponding to the virtual ideal control state, bus jitter noise elimination or a safety reset command is issued.

[0033] In one embodiment of the present invention, the environment adaptive sensitivity adjustment module is used to expand the tolerance range of the monitoring operator for non-critical action state transitions and shorten the discrete detection cycle of the core limit state node when the noise level fed back by the environment sensing module exceeds a preset threshold.

[0034] The following describes in more detail a collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs) with reference to Examples 2-6.

[0035] Example 2 To verify the spatiotemporal self-healing capability of this monitoring system in response to sudden external dynamic load interference, this embodiment uses an automated battery swapping station deployed along a long-distance oil and gas pipeline as a typical scenario, introducing a wind and sand shear environment with continuous wind speeds greater than 12 m / s. Under this condition, affected by nonlinear aerodynamic loads and intermittent frictional fluctuations of the guide rails, the lifting control module and the battery swapping control module inside the charging station experience asymmetrical response delays. That is, before the lifting platform reaches the preset alignment posture, the preset rigid timer has already triggered the battery swapping robotic arm to perform insertion and removal actions. This timing misalignment caused by environmental variables exceeds the real-time detection range of the discrete limit sensor. As the action sequence progresses, the battery swapping robotic arm may cut in with a centimeter-level spatial gap, causing a risk of rigid collision and physical engagement failure of the battery interface.

[0036] When the system faces the aforementioned interference conditions, the state-space multidimensional perception mapping module synchronously acquires the real-time current data and position signals of each heterogeneous execution module, calculates the current differential characteristics and position differential characteristics, and constructs a high-dimensional state vector. The semantic bias analysis adjudication module calculates the high-dimensional state vector. Compared with the pre-set ideal model The Euclidean distance between them outputs a bias gradient that reflects the change in the entropy of the execution sequence. When the deviation gradient When the preset evolution boundary is exceeded, the collaborative monitoring and control module reconstructs the control bus topology and adjusts the startup timing of subsequent execution modules.

[0037] Specifically, the collaborative monitoring and control module extracts the travel feedback residual from the non-converged execution module (i.e., the lifting control module), and converts the travel feedback residual into logic damping timing parameters in the internal register of the background controller. The logic damping timing parameters It exhibits a monotonically positive correlation with the travel feedback residual. Subsequently, the collaborative monitoring and control module will assign logic damping timing parameters... Write the state machine of the subsequent execution module to dynamically configure the suspension delay time, so that the control instructions of the subsequent execution module produce discrete logic yielding.

[0038] The causal bridging mechanism for converting spatial dimension travel feedback residuals into temporal dimension logical damping timing parameters is as follows: The backend controller uses a built-in spatiotemporal transformation mapping program to multiply the remaining geometrical spatial travel of the preceding push component by the reciprocal of the maximum permissible physical propulsion speed of the actuator under the current variable load environment, thereby obtaining the minimum physical safety time required to eliminate the spatial residuals; subsequently, the logical damping timing parameters are generated using this physical safety time as a lower bound. Furthermore, the suspend delay timer of the subsequent execution module's action state machine is rewritten. Through this timing extension mechanism in the digital control logic, the subsequent execution mechanism generates discrete yielding on the time axis, thereby avoiding entering the spatial overlapping interference region and achieving spatiotemporal decoupling.

[0039] Furthermore, the state-space multidimensional perception mapping module constructs a high-dimensional state vector. Subsequently, the semantic bias analysis adjudication module continuously calculates its correlation with the ideal model. The Euclidean distance between them. When the deviation gradient When E exceeds the preset evolution boundary, the background controller triggers a hardware core interrupt and calls the bus configuration subroutine to rewrite the data frame priority allocation table in the controller local area network bus application layer object dictionary. The original high-priority subsequent execution module control instruction data frame identifier is changed to a low-priority asynchronous waiting frame identifier, thereby adjusting the data flow topology while maintaining the existing hardware communication link unchanged.

[0040] Simultaneously, the background controller extracts the residual travel feedback from the preceding non-converged execution module and multiplies it by the damping conversion coefficient stored in non-volatile memory to obtain the logic damping timing parameters in milliseconds. The damping conversion coefficient is used to map the travel displacement deviation in the spatial geometric dimension to a time delay sequence. Its value is determined based on the maximum allowable propulsion speed of the actuator under extreme interference conditions. In this embodiment, each millimeter of travel feedback residual corresponds to a 5ms time delay, i.e., the damping conversion coefficient is 5ms / mm.

[0041] During the yielding period, the collaborative monitoring and control module activates the sliding mode control operator. The sliding mode control operator uses the differential data of the drive loop current, the differential data of the position feedback, and their first-order time derivatives as multidimensional system state variables, and uses a discrete-time sliding surface control law to solve the sliding mode control equations online, outputting nonlinear dynamic deviation compensation parameters.

[0042] In the specific calculation process, the position feedback differential data, the first-order time derivative of the position feedback differential data, and the drive loop current differential data are used as the first system state variables, the second system state variables, and the third system state variables, respectively. The background controller calls the built-in linear combination program to multiply the first system state variable by the first sliding surface coefficient, the second system state variable by the second sliding surface coefficient, and then perform an algebraic summation with the third system state variable to obtain the discrete sliding surface function value of the current sampling period. Subsequently, a reaching law relationship is established according to the discrete-time equivalent control theory, and the control output increment is solved through the discrete difference equation of the system state space. The control output increment is the nonlinear dynamic deviation compensation parameter.

[0043] The background controller has a preset control saturation limit corresponding to the maximum safe contact force threshold. When the nonlinear dynamic deviation compensation parameter obtained online exceeds the control saturation limit, the collaborative monitoring and control module truncates it to the control saturation limit. Subsequently, the nonlinear dynamic deviation compensation parameter with truncation constraint is written into the control port register of the current execution module, and the control sampling window is shortened and the output drive frequency is increased according to the parameter, thereby limiting the torque over-limit of the end effector and reducing the response delay of the compensation command.

[0044] The collaborative monitoring and control module continuously monitors the state transition feedback of the executed actions. When it detects that the stroke feedback residual of a non-converged execution module has converged to the 0.05mm boundary, the system triggers a hardware timer interrupt and clears the suspended delay state in the state machine of the subsequent execution module within the next 10μs control cycle, releasing the timing constraints. Afterward, the subsequent execution module performs insertion and removal actions based on the corrected high-frequency control parameters, enabling the heterogeneous execution mechanism to cross the nonlinear interference region caused by the external environment and gradually converge to the ideal model. Defined default execution surface.

[0045] The shadow state machine redundancy monitoring module independently maintains a virtual ideal control state synchronized with the actual power replenishment process in the background. Within each 1ms sampling period, it extracts the state machine transition code, position encoder discrete pulse count value, and drive current discrete differential value from the control bus, combining them to construct a multi-dimensional time-series state hash matrix as the semantic fingerprint of actual operation; simultaneously, based on the ideal model... Output the hexadecimal ideal state control word as the ideal semantic fingerprint.

[0046] To achieve same-dimensional alignment and comparison of two types of semantic fingerprints, the backend controller has a built-in feature dimensionality reduction mapping program. This program first extracts the feature values ​​from the multi-dimensional time-series state hash matrix, and performs linear weighted summation based on a preset weight matrix to compress the high-dimensional matrix into a single-value scalar. Then, it uses a preset nonlinear mapping interval to convert the single-value scalar into a fixed-length binary encoded string, and further converts it into the actual running state control word in standard hexadecimal format.

[0047] Based on this, the system calls the bitwise XOR comparison operator to compare the actual operating state control word with the ideal state control word bit by bit. When the XOR result of three consecutive sampling cycles is not zero and the duration exceeds 5ms, the background controller determines that the actual operating state stream does not match the ideal state, immediately blocks the pulse width modulation output port, cuts off the drive circuit current, and issues a safety reset command to put the system into a safety reset state.

[0048] In the safe reset state, the background controller controls each heterogeneous execution module to retreat to the safe gap standby position and continuously executes the bus data cleaning program to eliminate occasional communication noise. When the bus communication stabilizes and the semantic fingerprint of the actual running state flow is consistent with the ideal semantic fingerprint corresponding to the safe gap standby position, the background controller sends a re-initialization control signal through the return loop to release the reset lock and reactivate the power replenishment control timing. This restores the automated power replenishment operation while ensuring physical collision avoidance safety, forming a monitoring closed loop with self-healing retry capability.

[0049] Example 3 A hardware-in-the-loop (HIL) physics simulation test bench is currently being built. The test bench integrates a lifting platform with independent servo drives, a rotating landing pad, and a battery-swapping robotic arm. The environmental simulation unit applies random aerodynamic load disturbances with a frequency range of 0.5Hz to 5.0Hz and an amplitude range of 10N to 50N to the lifting platform guide rails as a background noise source.

[0050] The sampling period for the position encoders and current sensors of each heterogeneous execution module is set to 1 ms. The selection of the sampling period is determined based on the physical boundary conditions between the transient deviation capture accuracy and the controller bus bandwidth load. When the aerodynamic load disturbance frequency reaches the upper limit boundary of 5.0 Hz, the 1 ms sampling period satisfies the Nyquist sampling theorem requirement and can provide a corresponding phase margin.

[0051] To quantitatively verify the technical superiority and spatiotemporal consistency self-healing capability of the present invention under extreme variable load conditions in the field, this embodiment simultaneously constructs two sets of existing technology control groups for linkage testing: Control group 1 adopts the motion following collision avoidance control mechanism disclosed in CN114803384B, which implements following control based on one-dimensional calculation of spatial relative distance and static multi-segment linear deceleration function. Control group 2 adopts the cross-nest power replenishment control mechanism disclosed in CN121704289A. This mechanism is based on fixed action step sequence hard push, hardware departure limit state cascade verification and background task urgency game scheduling to realize control logic entry.

[0052] During the test, the environmental simulation unit output stepped disturbance loads with equivalent wind speeds of 8.2 m / s, 14.5 m / s and 18.6 m / s respectively. The multi-channel data acquisition unit simultaneously recorded the position feedback data and time sequence compensation data of each group under the multidimensional interference condition.

[0053] When the equivalent wind speed is 8.2 m / s, in control group one, the lifting control module measured a stroke feedback residual of 2.3 mm. The battery swapping robotic arm started the insertion and removal action according to the static timing sequence, generating a contact stress peak of 45 N. In control group two, since the multi-point physical sensors had not yet shown a large-scale signal jump, the battery swapping robotic arm started normally after the discrete limit node closed, generating a contact stress peak of 32 N.

[0054] In the experimental group, the state-space multidimensional sensing mapping module extracts current difference features and position difference features to construct a high-dimensional state vector. The semantic bias analysis adjudication module calculates the high-dimensional state vector. Compared with the ideal model The Euclidean distance between them yields the bias gradient. The value is 0.15. The collaborative monitoring and control module extracts the 2.3mm stroke feedback residual and converts it into logic damping timing parameters. The calculated result is 12.5ms. Subsequently, the logic damping timing parameters for 12.5ms are... The state machine of the subsequent execution module is written, causing the control commands of the battery swapping robotic arm to generate a discrete logic yield of 12.5ms. Finally, the peak contact stress of the battery swapping robotic arm in the test group was measured to be 12N.

[0055] When the equivalent wind speed output by the environmental simulation unit increases to 14.5 m / s, the stroke feedback residual of control group 1 increases to 11.6 mm, the battery swapping robotic arm is forced to engage, the peak contact stress reaches 385 N, and mechanical locking occurs.

[0056] Under the corresponding operating conditions, the test group measured the deviation gradient. The value reached 0.68 and exceeded the preset evolution boundary. The collaborative monitoring and control module calculated the logic damping timing parameters. The yield period was 48.2 ms. Within this 48.2 ms yield period, the collaborative monitoring and control module activated the sliding mode control operator, outputting nonlinear dynamic deviation compensation parameters through the discrete-time sliding surface control law. Since the solution result exceeded the preset control saturation limit, the collaborative monitoring and control module truncated it to the control saturation limit and wrote it into the control port register of the current execution module. Simultaneously, it increased the output drive frequency, driving the lifting control module to cross the frictional abrupt change region. Finally, the peak contact stress of the test group was measured to remain at 27 N.

[0057] When the equivalent wind speed output by the environmental simulation unit was further increased to 18.6 m / s, the experimental group measured the deviation gradient. E reaches 0.92. The collaborative monitoring and control module will adjust the logic damping timing parameters. The maximum hardware hang time limit set by the system is locked at 60.0ms, and the discrete-time sliding surface control law continuously outputs the control saturation limit parameter.

[0058] At the end of the 60.0 ms yield cycle, the stroke feedback residual of the non-converged execution module converged to 0.08 mm, and the instantaneous insertion / extraction contact stress was measured to be 42 N. The results show that, under the synergistic control of multi-dimensional feature perception and state machine timing extension, this invention can transform spatial-physical conflicts caused by external environmental variables into buffer gaps on the logical time axis, enabling heterogeneous actuators to converge to the ideal model even under extreme perturbation environments. Defined security execution surface.

[0059] Example 4 This embodiment combines Figures 1 to 6 The present invention describes the collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs).

[0060] like Figure 1As shown, the collaborative monitoring system for the automated inspection and recharging process of UAVs includes a heterogeneous execution module, a state space multidimensional perception mapping module, a semantic deviation analysis and adjudication module, a collaborative monitoring and control module, an execution module control port register, an environmental perception module, an environmental adaptive sensitivity adjustment module, and a shadow state machine redundancy monitoring module.

[0061] The heterogeneous execution module provides current and position signals; the state-space multidimensional perception mapping module acquires these signals and constructs a high-dimensional state vector; the semantic deviation analysis and adjudication module calculates the deviation gradient based on the Euclidean distance between the high-dimensional state vector and the ideal model; and the collaborative monitoring and control module reconstructs the control bus topology based on the deviation gradient and adjusts the subsequent execution timing, while simultaneously outputting dynamic deviation compensation parameters and logical backoff parameters to the execution module control port register. The execution module control port register is used to write control instructions and release timing constraints when conditions are met.

[0062] The environmental perception module is used to acquire the environmental noise level and output the environmental noise characteristics to the environmental adaptive sensitivity adjustment module; the environmental adaptive sensitivity adjustment module dynamically optimizes the sensitivity window of the monitoring operator and applies the adjustment result to the collaborative monitoring and control module.

[0063] In addition, the semantic deviation analysis and adjudication module outputs the actual operating state stream to the shadow state machine redundancy monitoring module. The shadow state machine redundancy monitoring module runs the virtual ideal process in parallel and performs semantic fingerprint verification. When a semantic fingerprint mismatch is detected, the system enters a safe reset state to perform noise removal or issue a safe reset command; when bus communication stabilizes and the semantic fingerprint is rematched, the system sends a reinitialization control signal to the heterogeneous execution module to reactivate the power replenishment control timing.

[0064] like Figure 2 As shown, the figure constructs a three-dimensional topological space evolution diagram with deviation gradient, energy replenishment operation evolution cycle, and contact stress as dimensions. The preset ideal state trajectory always remains on the reference plane where the contact stress is 0.

[0065] The sequential control control group, used as the comparison object, adopts the control mechanism disclosed in CN114803384B. As the energy replenishment operation cycle progresses and the deviation gradient increases, its control method based on one-dimensional spatial distance calculation cannot perceive the transient response delay caused by nonlinear disturbances, resulting in mismatch of action timing, rapid increase in contact stress, and significant deviation of the trajectory towards the high-stress region.

[0066] In contrast, the collaborative monitoring system of this invention completes the control bus topology reconstruction and activates the sliding mode deviation compensation mechanism within the yielding cycle. Under the same deviation gradient and operation cycle conditions, it only produces a small amount of contact stress change, exhibiting low contact stress and self-healing synchronization trajectory characteristics.

[0067] like Figure 3 As shown, the horizontal axis represents the external equivalent wind speed (m / s), and the vertical axis represents the peak contact stress (N).

[0068] When the external equivalent wind speed is 8.2 m / s, the peak contact stress measured in the sequential control control group using the CN114803384B control mechanism is 45 N, while the peak contact stress measured in the collaborative monitoring test group of this invention is 12 N.

[0069] When the external equivalent wind speed increases to 14.5 m / s, the control group 1 cannot identify the nonlinear cumulative error in the action sequence, which leads to the forced advancement of the subsequent mechanism, the peak contact stress increases to 385 N and mechanical lock-up occurs; under the corresponding conditions, the peak contact stress of the test group of this invention is only 27 N.

[0070] When the external equivalent wind speed further increased to 18.6 m / s, the peak contact stress of the test group of this invention was 42 N. Since the control group 1 had already experienced mechanical locking in the previous stage, and its control system and transmission mechanism failed, its subsequent data curves are not extended in the figure.

[0071] like Figure 4 As shown, the horizontal axis represents the sampling time sequence S1 to S6, and the vertical axis represents the speed characteristic value of the actuator (m / s). A horizontal dashed line of 0.01m / s is set as the steady-state determination threshold in the figure.

[0072] The sequential control control group 1, which adopts the CN114803384B control mechanism, exhibits significant velocity oscillations under variable load. Its velocity curve fluctuates continuously at high frequency and cannot converge to below the steady-state threshold within a finite time. This indicates that the traditional linear follower control method has insufficient stability under nonlinear disturbance environments.

[0073] The velocity characteristic curve of the experimental group of this invention shows a monotonically decreasing trend starting from S1, and crosses the steady-state threshold line in the S3-S4 interval; it maintains a stable convergence state in the S4-S6 stage. This result indicates that the present invention effectively suppresses the sudden impact of the action-interleaving stage through sliding mode approach control and iterative convergence mechanism.

[0074] like Figure 5 As shown, the horizontal axis represents time (ms), and the vertical axis represents the position difference eigenvalues.

[0075] When the equivalent wind speed reaches 14.5 m / s and the background noise frequency reaches the upper limit of 5.0 Hz, the step sequence hard propulsion control group II, which adopts the CN121704289A control mechanism, experiences bus jitter and sensor signal jumps due to the asymmetric deformation of mechanical components. Its control mechanism cannot achieve global spatiotemporal collaborative perception, resulting in a significant phase shift between the ideal model position reference signal and the actual position feedback signal, and a significant increase in the root mean square tracking error.

[0076] In contrast, the experimental group of this invention constructs a high-dimensional state vector and implements global pacing audit. The actual position feedback curve and the ideal model position reference curve are highly coincident, with a root mean square error of only 0.0827, effectively eliminating the phase mismatch problem caused by environmental disturbances.

[0077] like Figure 6 As shown, the horizontal axis represents the time series (ms), and the vertical axis represents the differential current signal of the drive circuit (A).

[0078] The second control group using the CN121704289A control mechanism suffers from a lack of global synchronization and coordination of control data due to the independent sensing links of each hardware sensor. To maintain the alignment of discrete limit nodes, the controller frequently issues compensation commands, resulting in significant spikes and large oscillations in the drive current curve. The current value fluctuates wildly between -20A and 20A, posing a high risk of electrical overload.

[0079] The collaborative monitoring system of this invention constrains the control output signal through control bus topology reconstruction and sliding mode approach control, significantly reducing the amplitude of the drive current curve, concentrating the overall fluctuation in the region near zero, and exhibiting a smooth convergence trend. This demonstrates that the invention can effectively suppress high-frequency blind response and improve the electrical safety and logic state self-healing capability of heterogeneous actuators under extreme operating conditions.

[0080] Example 5 To further verify the control stability of this monitoring system when it is subjected to sudden changes in internal physical properties caused by extreme external thermodynamic environment, this embodiment introduces an extreme cold region condition where the ambient temperature drops from 25°C to -35°C in the same long-distance oil and gas pipeline scenario as in Embodiment 1, and superimposes a local shear wind field interference with a wind speed of 15m / s.

[0081] During the landing of the UAV onto the rack of the refueling mothership, the landing gear's physical contact posture initially deviates. At this time, the low temperature causes the viscosity of the guide rail lubricating oil to increase, and this, combined with the alternating wind load stress, causes intermittent sluggish friction in the internal slider. As a result, during the lifting platform's upward pushing of the UAV for centering, its vertical drive circuit generates transient current surges, leading to a spatiotemporal decoupling execution deviation between the pushing mechanism and the end-effector battery swapping robotic arm. If the system cannot detect this deviation in time and implement closed-loop control, the battery swapping connector may rigidly connect under conditions that do not meet the coaxiality boundary conditions, causing deformation of the connector terminals or damage to the inspected assets.

[0082] For the aforementioned variable load conditions, the collaborative monitoring system for the automated inspection and recharging process of the UAV establishes a closed-loop information processing link through a state-space multi-dimensional perception mapping module, a semantic deviation analysis and adjudication module, and a collaborative monitoring and control module. The system synchronously reads the feedback current sequence of the platform's vertical drive motor and the output pulse position sequence of the absolute encoder via a data bus with a 1ms sampling period.

[0083] The state space multidimensional perception mapping module first reads the current motor feedback current value and performs differential calculation with the standard no-load current reference value preset in the read-only memory to obtain the current differential signal characterizing the variable load resistance of the motor; at the same time, it reads the actual pulse position value of the absolute encoder and performs differential calculation with the target pulse displacement value of the corresponding timing node in the state machine of the background controller to obtain the displacement differential signal characterizing the displacement gap of the actuator.

[0084] Subsequently, the state-space multidimensional perception mapping module establishes a floating-point array storage space, writes the current differential signal into the first scalar index bit, and writes the displacement differential signal into the second scalar index bit, thereby converting the multidimensional physical state of the heterogeneous execution unit into a high-dimensional state vector. .

[0085] To determine the preset evolution boundary values ​​in the semantic deviation analysis adjudication module, the system first performs offline engineering calibration. Under the standard operating conditions of no wind and no resistance on the simulation test bench, the system repeatedly performs 500 complete alignment and battery swapping operations, and records the ideal state vector generated in each cycle in real time.

[0086] The system performs Euclidean distance matrix clustering analysis on 500 sets of ideal state vectors, further calculates the covariance envelope radius characterizing the spatial clustering density of the ideal alignment pose, and determines the maximum fluctuation value of 0.035 as the system's preset evolution boundary. This preset evolution boundary corresponds to the physical displacement limit window allowed when the lifting mechanism is in a safe docking pose.

[0087] Under the current cold-region shear wind field conditions, the semantic deviation analysis adjudication module calls the built-in distance operator to read the high-dimensional state vector. And calculate its comparison with an ideal model stored in non-volatile memory. The Euclidean distance between them. The microprocessor outputs the deviation gradient, which characterizes the change in the current sequence entropy value of the system, through the square root operation of the sum of the squares of the array elements. .

[0088] As the interference from the variable load continues to accumulate, the deviation gradient... It continues to increase. When it reaches 0.052 and exceeds the preset evolution boundary of 0.035, the collaborative monitoring and control module rewrites the hard interrupt control word in situ, triggering real-time reconstruction of the data flow topology of the monitoring system.

[0089] At this point, the collaborative monitoring and control module suspends the downlink output register of the battery swapping robotic arm controller, cuts off the original sequential action timing, and reads the stroke feedback residual corresponding to the current pace of the lifting mechanism. Subsequently, the stroke feedback residual is multiplied by the damping conversion factor stored in the controller's read-only memory to convert it into logical damping timing parameters. .

[0090] Transformed logic damping timing parameters The time delay is measured in milliseconds and written as a dynamic delay constant into the timer register of the state machine of the subsequent execution unit, so that the action instructions of the battery swapping robotic arm generate discrete time backoff and realize the active extension of the execution timing.

[0091] During the yielding period, the collaborative monitoring and control module continuously reads the differential current data and position differential data of the drive circuit, and uses the sliding mode control operator to solve the discrete-time sliding surface control law online to obtain the nonlinear dynamic deviation compensation parameters.

[0092] Subsequently, the nonlinear dynamic deviation compensation parameters are converted into pulse width modulation signal duty cycle correction values ​​and written into the lifting mechanism drive port, thereby driving the lifting control module to continuously correct the current position deviation.

[0093] When the yielding cycle ends, the stroke feedback residual of the lifting control module converges to within the 0.05mm closed boundary. The system triggers a hardware timer interrupt, clears the state machine's suspended control word, and restores the timing sequence of subsequent battery swapping robotic arm actions. The battery swapping robotic arm then crosses the pose fluctuation zone to initiate the plugging and unplugging action, thereby completing the mapping of physical space deviation to logical time yielding and achieving stable energy replenishment control under extreme cold conditions.

[0094] Example 6 Based on the verification of external environmental interference in the aforementioned embodiments, in order to further verify the long-term adaptive capability of this monitoring system to the degradation of the mechanical characteristics of heterogeneous actuators, this embodiment introduces a mechanical aging scenario in which the long-term operation of the power station equipment leads to wear of the drive motor bearings and increased transmission clearance.

[0095] In this scenario, as the mechanical characteristic parameters of the key actuators gradually deviate from the factory baseline values, the preset ideal model in the state-space multidimensional perception mapping module... A persistent, entrenched deviation develops between the actual response and the actual response of the actuator. As operating time increases, this entrenched deviation may cause the system to frequently trigger false alarms and unnecessary logical failovers, thereby reducing the efficiency of recharging operations.

[0096] To eliminate the impact of static model drift caused by equipment aging on collaborative monitoring logic, this invention forcibly initiates the baseline model adaptive reconstruction procedure before the system enters the high-frequency power replenishment mode.

[0097] Specifically, the system calls upon historical data from the entire energy replenishment operation process, using the current feedback sequence and position feedback sequence collected by each heterogeneous execution module within 50 consecutive energy replenishment cycles as steady-state time-series samples. Subsequently, the first derivative of the sample sequence is extracted as a velocity feature, and the time interval where the velocity feature is below 0.01 m / s is defined as the steady-state operation interval.

[0098] Within the steady-state operating range, the system calculates the average current value. and the average displacement deviation And the least squares method is used for the ideal model. The feature weights in the model are iteratively updated. After the update is complete, the new model parameters are written to the baseline model configuration area in non-volatile memory, thus forming the reconstructed ideal model. .

[0099] To verify the stability of the model reconstruction results, the system further introduces a working condition offset compensation factor and calculates the high-dimensional state vector of the current operating cycle in real time. Compared with the reconstructed ideal model The corresponding covariance matrix .

[0100] Wherein, the covariance matrix The diagonal elements are used to characterize the fluctuation intensity of response characteristics in each dimension, while the off-diagonal elements are used to characterize the degree of physical coupling between different actuators.

[0101] When the covariance matrix If the fluctuation amplitude in any dimension exceeds 1.5 times the preset evolution boundary and the duration exceeds 500ms, the collaborative monitoring and control module determines that there is a sudden change in the physical characteristics of the system and executes the model anomaly recovery process.

[0102] In the model anomaly recovery process, the collaborative monitoring and control module first resets the feature fusion gain of the state space multidimensional perception mapping module and initializes the state weights of all execution modules to the mean distribution; then it starts the three-step iterative convergence operator to perform model recalibration.

[0103] The first step is to establish a linear regression model using the current motor feedback current value and the corresponding stroke feedback residual, and to preliminarily correct the linear correlation coefficient between current and stroke. The second step is to correct the logic damping timing parameters based on the updated linear correlation coefficient. The transformation mapping curve enables the logical yielding mechanism to adapt to the current mechanical wear state; The third step is to detect the peak contact stress when the end of the power replenishment robot arm engages with the battery terminal, and determine whether it converges back to the preset safe execution surface, in order to verify the effectiveness of the model reconstruction results.

[0104] When the peak contact stress meets the preset safety execution surface requirements, the system confirms that the model reconstruction is complete and uses the updated parameters as the new operating benchmark.

[0105] Through the aforementioned adaptive reconstruction and dynamic calibration mechanism, the system can continuously compensate for the physical characteristic drift caused by equipment aging, enabling the sensing benchmark of the monitoring system to evolve synchronously with the mechanical wear state, avoiding logical judgment drift caused by equipment aging, thereby ensuring that heterogeneous actuators maintain stable spatiotemporal synchronization consistency and energy replenishment coordination capabilities throughout their entire life cycle.

[0106] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0107] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0108] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs), characterized in that, include: The state space multidimensional perception mapping module is used to collect current signals and position signals from heterogeneous execution modules, and construct a high-dimensional state vector based on the current signals and position signals. The semantic deviation analysis and adjudication module is used to calculate the Euclidean distance between the high-dimensional state vector and the ideal model state vector to output the deviation gradient. The deviation gradient is used to characterize the spatiotemporal offset state of the execution sequence and triggers the collaborative monitoring and control module to execute the control strategy when the preset evolution boundary is exceeded. The collaborative monitoring and control module is used to reconstruct the control bus topology and regulate the startup timing of subsequent execution modules when the deviation gradient exceeds a preset evolution boundary. The collaborative monitoring and control module is further used to generate logic damping timing parameters based on the travel feedback residual of the unconverged execution modules. The background controller uses a built-in spatiotemporal transformation mapping program to multiply the remaining geometrical space travel of the preceding execution component by the reciprocal of the maximum permissible physical propulsion speed of the actuator under the current variable load environment to obtain the minimum physical safety time required to eliminate the spatial residual, and uses this minimum physical safety time as a lower bound. The logic damping timing parameters are generated to convert the spatial dimension travel feedback residual into a time dimension logic damping timing parameter; the logic damping timing parameters are written into the state machine of the subsequent execution module to generate discrete logic yield; the sliding mode control operator is activated during the yielding period to solve the sliding mode control equation online and output nonlinear dynamic deviation compensation parameters limited by the maximum threshold of the safe contact force; the nonlinear dynamic deviation compensation parameters are written into the control port register of the current execution module; when the travel feedback residual converges to the 0.05mm boundary, the timing constraints of the state machine of the subsequent execution module are released in the next control period; The shadow state machine redundancy monitoring module is used to run the virtual ideal process in parallel and verify the actual running state flow with the semantic fingerprint. When there is a semantic mismatch, the system enters a safe reset state. An environment-adaptive sensitivity adjustment module is used to dynamically adjust the sensitivity window of the monitoring operator based on the noise level fed back by the environment sensing module.

2. The collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The heterogeneous execution module includes a lifting control module, a rotation control module, and a battery swapping control module; the state space multidimensional perception mapping module is used to synchronously acquire the real-time current data and position signals of each heterogeneous execution module, and construct the high-dimensional state vector through differential processing.

3. The collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The semantic deviation analysis adjudication module is also used in the deviation gradient When E exceeds the preset evolution boundary, the deviation gradient will be... The evolutionary characteristics of E are compared with a preset interference model library to identify the type of interference source in the current external environment.

4. The collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The logic damping timing parameters It has a monotonically positive correlation with the travel feedback residual of the unconverged execution module, and is used to dynamically configure the suspension delay time of the state machine of the subsequent execution module.

5. The collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The sliding mode control operator uses the drive loop current differential data, position feedback differential data, and the first-order time derivative of the position feedback differential data as multidimensional system state variables to solve the sliding mode control equations online.

6. The collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The control saturation limit corresponding to the maximum threshold of the safe contact force is preset by the background controller; when the nonlinear dynamic deviation compensation parameter exceeds the control saturation limit, the collaborative monitoring and control module will limit the output control parameter to the control saturation limit.

7. A collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The collaborative monitoring and control module writes the nonlinear dynamic deviation compensation parameters into the control port register of the current execution module by: shortening the control sampling window of the current execution module and increasing the output drive frequency according to the nonlinear dynamic deviation compensation parameters.

8. A collaborative monitoring system for the automated inspection and recharging process of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The collaborative monitoring and control module releases timing constraints by triggering a microsecond-level hardware timer interrupt when the travel feedback residual reaches the 0.05mm boundary, and clearing the suspended delay state in the state machine of the subsequent execution module in the next control cycle.

9. A collaborative monitoring system for the automated inspection and recharging process of a UAV according to claim 1, characterized in that, The shadow state machine redundancy monitoring module is used to maintain a virtual ideal control state synchronized with the actual power replenishment process and convert the control bus data stream into an operational semantic fingerprint. When the semantic fingerprint of the actual operational state stream does not match the semantic fingerprint corresponding to the virtual ideal control state, bus jitter noise elimination or a safety reset command is issued.

10. A collaborative monitoring system for the automated inspection and recharging process of a UAV according to claim 1, characterized in that, The environmental adaptive sensitivity adjustment module is used to expand the tolerance range of the monitoring operator for non-critical action state transitions and shorten the discrete detection cycle of the core limit state node when the noise level fed back by the environmental sensing module exceeds a preset threshold.

Images