Adsorption state judgment method and system based on negative pressure dynamic threshold and delay filtering

Abstract

The application discloses an adsorption state judgment method and system based on negative pressure dynamic threshold and delay filtering, relates to the technical field of unmanned aerial vehicle wall operation control, and aims to solve the technical problems of overshoot misjudgment, poor adaptability in multiple scenes and lack of full-cycle safety monitoring of the existing fixed threshold instantaneous judgment method. The application automatically identifies the wall material through multi-source sensing data, calculates the dynamic adsorption threshold and leakage threshold suitable for the current operation condition in combination with the real-time load of the unmanned aerial vehicle, the environmental pressure and the adsorber wear condition, performs three-level progressive delay filtering processing on the negative pressure signal to complete the adsorption success judgment, continuously monitors the negative pressure state throughout the operation, distinguishes the two fault types of slow leakage and sudden failure and triggers the emergency evacuation. The application significantly reduces the adsorption state misjudgment rate, improves the universality and safety of the unmanned aerial vehicle wall operation, and is suitable for high-altitude wall operation scenes such as building inspection and bridge detection.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) wall-mounted operation control technology, specifically to a method and system for determining adsorption state based on negative pressure dynamic threshold and time delay filtering. Background Technology

[0002] With the continuous growth in demand for high-altitude wall operations such as building inspection, bridge inspection, and exterior wall cleaning, negative pressure adsorption drones have become an important technological development direction in the field of high-altitude operations due to their advantages such as no need for rope traction, wide operating range, and high maneuverability. Accurate judgment of the adsorption state is a core prerequisite for ensuring operational safety. Currently, commercial negative pressure adsorption drones generally use a single fixed threshold instantaneous judgment method to detect the adsorption state. This method uses a negative pressure sensor installed in the air circuit to collect the air pressure inside the suction cup. When the detected value reaches a factory-preset uniform fixed threshold, adsorption is immediately determined to be successful. Although it has the characteristics of simple implementation and fast response speed, it has several insurmountable technical defects: First, after the vacuum pump starts, the negative pressure inside the suction cup will undergo an inherent process of "rapid rise - overshoot - oscillation decay - stabilization". Instantaneous judgment is prone to misjudging adsorption success at the overshoot moment, while the actual negative pressure subsequently falls below the threshold, directly causing a desorption and fall accident. Second, the airtightness and adsorption characteristics of different wall materials vary significantly. Smooth walls have a fast negative pressure rise but a low stability value, while rough walls have a low negative pressure... While the initial rise is slow, the stability value is high. However, negative pressure leakage is severe on porous walls, and a single fixed threshold cannot be adapted to all scenarios. If the threshold is too high, adsorption success cannot be determined on rough walls, while if the threshold is too low, frequent misjudgments will occur on smooth walls. At the same time, this method cannot effectively filter the measurement noise of the negative pressure sensor itself, as well as the instantaneous fluctuations caused by airflow disturbances and slight deformation of the suction cup, which are prone to being misjudged as adsorption failure, leading to unnecessary operation interruptions. More importantly, the existing technology only makes a single judgment when adsorption is established, lacking a continuous monitoring mechanism throughout the entire operation cycle. It cannot detect sudden situations such as suction cup leakage or partial detachment during operation in a timely manner, posing a great safety hazard. The above problems have become the core technical bottlenecks restricting the safe, reliable, and widespread application of negative pressure adsorption drones in complex environments. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for judging adsorption status based on negative pressure dynamic threshold and time delay filtering, which solves the problems of misjudging adsorption success due to the different air tightness and adsorption characteristics of different wall materials, as well as the inability to effectively filter the measurement noise of the negative pressure sensor itself and the instantaneous fluctuations caused by airflow disturbance and slight deformation of the suction cup, which are easy to misjudge as adsorption failure and cause unnecessary work interruption.

[0004] This invention is achieved through the following technical solution:

[0005] A method for determining the adsorption state based on negative pressure dynamic threshold and time delay filtering is applied to the wall operation control of a negative pressure adsorption UAV, including the following steps:

[0006] S1. Collect multi-source perception data of the current working wall surface using multiple sensors carried by the drone, and automatically identify the current wall surface material based on the multi-source perception data;

[0007] S2. Based on the identified current wall material, combined with the drone's real-time payload, current ambient air pressure, and number of times the suction cup has been used, calculate the dynamic stable adsorption threshold and dynamic leakage threshold that are suitable for the current operating conditions, and at the same time retrieve the maximum allowable fluctuation value corresponding to the current material.

[0008] S3. Collect the original negative pressure signal inside the suction cup, perform three-level progressive delay filtering on the original negative pressure signal, and determine the adsorption success based on the dynamic stable adsorption threshold and the maximum allowable fluctuation value.

[0009] S4. After successful adsorption is determined, the negative pressure signal is continuously collected and filtered during the operation. Adsorption failure is judged based on the dynamic leakage threshold and the characteristics of negative pressure signal change. If adsorption failure is determined, the emergency evacuation mechanism is immediately triggered.

[0010] Furthermore, the automatic identification of the current wall material based on multi-source sensing data in S1 specifically includes:

[0011] S11. Acquire wall images using a monocular camera and extract multiple visual features; measure the wall reflection intensity using lidar to obtain laser features, and combine them to form a multi-dimensional feature vector;

[0012] S12. Calculate the distance between the multidimensional feature vector of the current wall surface and the multidimensional feature vectors of all samples in the pre-established material feature library, select the preset number of samples with the closest distance, and determine the material of the current wall surface through a voting method; wherein the material feature library pre-stores the multidimensional feature vectors of multiple typical wall surface samples;

[0013] Several visual features include gray-level mean, gray-level variance, edge density, and gradient orientation histogram features.

[0014] Furthermore, S2 specifically includes:

[0015] S21. Based on the identified current wall material, retrieve the corresponding material basic stability adsorption threshold, basic leakage threshold and maximum allowable fluctuation value from the pre-established basic threshold database;

[0016] S22. Calculate the drone weight correction factor, ambient air pressure correction factor, and suction cup wear correction factor respectively; wherein the drone weight correction factor is determined based on the ratio of the drone's current actual weight to the standard weight, the ambient air pressure correction factor is determined based on the ratio of the current ambient atmospheric pressure to the standard atmospheric pressure, and the suction cup wear correction factor is determined based on the number of times the suction cup has been used;

[0017] S23. The dynamic stable adsorption threshold is calculated based on the product of the material's basic stable adsorption threshold and three correction coefficients; the dynamic stable adsorption threshold is multiplied by a preset leakage ratio coefficient to obtain the dynamic leakage threshold.

[0018] Furthermore,

[0019] The first stage of the three-stage progressive delay filtering process in S3 is a sliding window mean filter, specifically:

[0020] The original negative pressure signal inside the suction cup is acquired at a preset sampling frequency. The original negative pressure signal is then processed by a moving average of a preset window size to obtain a filtered negative pressure value, which is used for subsequent rising edge detection.

[0021] Furthermore,

[0022] The second stage of the three-stage progressive delay filtering process in S3 is rising edge detection and delay start, specifically:

[0023] When the filtered negative pressure value first reaches the first preset proportion of the dynamic stable adsorption threshold, it is marked as adsorption begins and a delay timer of the first preset duration is started; if the filtered negative pressure value is detected to decrease and fall below the second preset proportion of the dynamic stable adsorption threshold during the delay period, the delay timer is reset and the first-level sliding window mean filtering is re-executed.

[0024] Furthermore, the third stage of the three-stage progressive delay filtering process in S3 is stability verification, specifically:

[0025] After the first preset timer expires, the system enters the second preset time stability verification phase. During this phase, the filtered negative pressure value is collected, and its average value and standard deviation are calculated. When the average value is greater than or equal to the dynamic stability adsorption threshold and the standard deviation is less than or equal to the maximum allowable fluctuation value corresponding to the current material, the adsorption is finally determined to be successful. If the conditions are not met, all timers are reset, and the first-level sliding window mean filtering is re-executed.

[0026] Furthermore, the adsorption fault judgment based on the dynamic leakage threshold and negative pressure signal change characteristics in S4 specifically includes:

[0027] S41. Adsorption slow leakage judgment: When the filtered negative pressure value of a preset number of consecutive sampling points is lower than the dynamic leakage threshold, it is judged as adsorption slow leakage;

[0028] S42. Judgment of sudden adsorption failure: When the negative pressure value after filtering drops sharply beyond the preset pressure drop threshold within the third preset time period, it is judged as a sudden adsorption failure.

[0029] Furthermore, the emergency evacuation mechanism triggered in S4 is as follows: when it is determined that there is slow leakage of adsorption or sudden failure of adsorption, the drone is controlled to smoothly detach from the working wall and the current wall operation task is terminated.

[0030] Furthermore,

[0031] The specific process of controlling the drone to smoothly detach from the working wall is as follows: maintain the current attitude of the drone and gradually reduce the power of the vacuum pump, while activating the flight propulsion system to provide auxiliary lift that matches the current adsorption force. After the adsorption force between the drone and the working wall is completely released and the flight attitude is stable, control the drone to detach from the working wall and fly to the preset safe area to hover.

[0032] Furthermore, the adsorption state determination system based on negative pressure dynamic threshold and time delay filtering includes:

[0033] The wall material recognition module is used to collect multi-source sensing data and identify the current wall material;

[0034] The dynamic threshold calculation module communicates with the wall material identification module and is used to calculate the dynamic stable adsorption threshold and dynamic leakage threshold based on the identified wall material and real-time operating parameters, and retrieve the maximum allowable fluctuation value corresponding to the current material.

[0035] The three-stage progressive delay filtering module is connected to the dynamic threshold calculation module and the negative pressure sensor respectively. It is used to perform three-stage progressive delay filtering on the original negative pressure signal and to determine the adsorption success based on the dynamic stable adsorption threshold and the maximum allowable fluctuation value.

[0036] The adsorption status continuous monitoring module is connected to the dynamic threshold calculation module and the three-level progressive delay filtering module. It is used to monitor the negative pressure status after filtering throughout the entire operation cycle and determine adsorption faults. When a fault is determined, an emergency evacuation mechanism is triggered.

[0037] The beneficial effects of this invention are as follows:

[0038] 1. This invention employs a three-stage progressive delay filtering judgment technology, effectively solving the core problem that traditional fixed-threshold instantaneous judgment methods cannot distinguish between negative pressure overshoot oscillation and the true stable state during the initial startup of a vacuum pump. First, high-frequency noise and instantaneous airflow disturbances from the sensor are eliminated through sliding window mean filtering. Then, overshoot interference during the rapid rise of negative pressure is filtered through rising edge detection and delayed startup. Finally, the continuous stability of the negative pressure is verified through stability checking. This not only completely avoids desorption and falling accidents caused by misjudging adsorption success at the overshoot moment, but also significantly shortens unnecessary judgment delays, achieving an optimal balance between the accuracy of adsorption state judgment and response speed.

[0039] 2. This invention constructs a multi-dimensional dynamic threshold correction system based on multi-source sensing, breaking through the technical bottleneck that traditional single fixed thresholds cannot adapt to complex operating environments. The system first identifies and matches the corresponding basic threshold by wall material recognition, and then introduces real-time correction coefficients from three dimensions: UAV real-time load, ambient air pressure, and suction cup wear, to dynamically adjust the basic threshold. This not only solves the problem of threshold adaptation for different walls, but also compensates for the decrease in vacuum pump efficiency in high-altitude areas, the increase in adsorption force demand due to increased load, and the decrease in sealing performance caused by suction cup aging. This ensures that the adsorption judgment threshold always accurately matches the actual operating conditions, greatly improving the versatility and robustness of UAV wall operations.

[0040] 3. This invention employs a multi-source sensing wall material recognition scheme that integrates a monocular camera and LiDAR, providing a reliable foundation for dynamic threshold calculation. By extracting visual features such as the mean grayscale value, grayscale variance, edge density, and gradient direction histogram of the wall image, and combining them with the wall reflection intensity measured by LiDAR, a multi-dimensional feature vector is constructed. Then, the K-nearest neighbor classification algorithm is used to achieve rapid online material recognition. This not only compensates for the blind spots of a single sensor but also achieves high-precision classification without complex model training. Furthermore, the classification accuracy can be continuously improved as the sample library expands, ensuring that the subsequent dynamic threshold can accurately match the airtightness and adsorption characteristics of the current wall. Attached Figure Description

[0041] Figure 1 This is the overall logic diagram of the present invention;

[0042] Figure 2 This is a schematic diagram of the overall process of the present invention;

[0043] Figure 3 This is a schematic diagram of the three-stage progressive delay filtering process of the present invention. Detailed Implementation

[0044] The present invention will be further described in detail below with reference to the embodiments and accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0045] See the example. Figures 1 to 3:

[0046] A method for determining the adsorption state based on negative pressure dynamic threshold and time delay filtering is applied to the wall operation control of negative pressure adsorption UAVs. Its features include the following steps:

[0047] S1. Collect multi-source perception data of the current working wall surface using multiple sensors carried by the drone, and automatically identify the current wall surface material based on the multi-source perception data;

[0048] S2. Based on the identified current wall material, combined with the drone's real-time payload, current ambient air pressure, and number of times the suction cup has been used, calculate the dynamic stable adsorption threshold and dynamic leakage threshold that are suitable for the current operating conditions, and at the same time retrieve the maximum allowable fluctuation value corresponding to the current material.

[0049] S3. Collect the original negative pressure signal inside the suction cup, perform three-level progressive delay filtering on the original negative pressure signal, and determine the adsorption success based on the dynamic stable adsorption threshold and the maximum allowable fluctuation value.

[0050] S4. After successful adsorption is determined, the negative pressure signal is continuously collected and filtered during the operation. Adsorption failure is judged based on the dynamic leakage threshold and the characteristics of negative pressure signal change. If adsorption failure is determined, the emergency evacuation mechanism is immediately triggered.

[0051] Furthermore, the automatic identification of the current wall material based on multi-source sensing data in S1 specifically includes:

[0052] S11. Acquire wall images using a monocular camera and extract multiple visual features; measure the wall reflection intensity using lidar to obtain laser features, and combine them to form a multi-dimensional feature vector;

[0053] S12. Calculate the distance between the multidimensional feature vector of the current wall surface and the multidimensional feature vectors of all samples in the pre-established material feature library, select the preset number of samples with the closest distance, and determine the material of the current wall surface through a voting method; wherein the material feature library pre-stores the multidimensional feature vectors of multiple typical wall surface samples;

[0054] Several visual features include gray-level mean, gray-level variance, edge density, and gradient orientation histogram features.

[0055] In one embodiment, when the drone approaches the wall to be worked on, it acquires images of the wall using a monocular camera and extracts visual features that comprehensively characterize the wall's texture and roughness, such as grayscale mean, grayscale variance, edge density, and gradient direction histogram. Simultaneously, it measures the wall's reflectance intensity using lidar to obtain laser features reflecting the wall's physical properties. These features are combined to form a 5-dimensional feature vector. The Euclidean distance between this multi-dimensional feature vector and the multi-dimensional feature vectors of all samples in a pre-established material feature library is calculated. A predetermined number of samples with the closest distance are selected, and a voting method is used to determine the current wall material. The material feature library pre-stores multi-dimensional feature vectors of various typical wall samples. The fusion of multi-source features can compensate for the blind spots of a single sensor. The K-nearest neighbor classification algorithm can achieve fast online classification without complex training, and the classification accuracy can be further improved as the sample library expands. This ensures that the subsequent dynamic threshold can accurately match the current wall's airtightness and adsorption characteristics, laying a reliable foundation for the entire adsorption state judgment system.

[0056] Furthermore, S2 specifically includes:

[0057] S21. Based on the identified current wall material, retrieve the corresponding material basic stability adsorption threshold, basic leakage threshold and maximum allowable fluctuation value from the pre-established basic threshold database;

[0058] S22. Calculate the drone weight correction factor, ambient air pressure correction factor, and suction cup wear correction factor respectively; wherein the drone weight correction factor is determined based on the ratio of the drone's current actual weight to the standard weight, the ambient air pressure correction factor is determined based on the ratio of the current ambient atmospheric pressure to the standard atmospheric pressure, and the suction cup wear correction factor is determined based on the number of times the suction cup has been used;

[0059] S23. The dynamic stable adsorption threshold is calculated based on the product of the material's basic stable adsorption threshold and three correction coefficients; the dynamic stable adsorption threshold is multiplied by a preset leakage ratio coefficient to obtain the dynamic leakage threshold.

[0060] By introducing multi-dimensional real-time correction coefficients to dynamically adjust the material baseline threshold, the adsorption judgment threshold can accurately adapt to constantly changing operating conditions such as drone real-time payload, ambient air pressure, and suction cup wear. This solves the problems of poor adaptability and inability to cover complex operating scenarios with traditional fixed thresholds, significantly improving the accuracy and versatility of adsorption state judgment. In one embodiment, based on the identified current wall material, the corresponding material baseline stable adsorption threshold is retrieved from a pre-established baseline threshold database. The basic leakage threshold and maximum permissible fluctuation value; calculate the UAV weight correction factor respectively. Ambient air pressure correction factor and suction cup wear correction factor K w ;

[0061] The drone weight correction factor K m The ratio of the actual weight m of the drone to its standard weight m0 is used for calculation. The formula is as follows: The drone weight correction factor can compensate for the increased adsorption force required due to the increased payload.

[0062] Ambient air pressure correction factor K p The ratio of the current ambient atmospheric pressure P to the standard atmospheric pressure P0 is used for determination, and the calculation formula is as follows: The environmental pressure correction coefficient can compensate for the decrease in vacuum pump efficiency caused by the reduction in air pressure at high altitudes.

[0063] Suction cup wear correction factor K w The number of times the suction cup is used, n, is determined by the formula K. w =1+k×n (k is the preset wear coefficient), the suction cup wear correction coefficient can compensate for the decrease in sealing performance caused by the aging of the suction cup;

[0064] Based on the material-based stable adsorption threshold P base The dynamic stable adsorption threshold P is calculated by multiplying it by the three correction factors. stable The calculation formula is P stable =K m ×K p ×K w ×P base The dynamic stable adsorption threshold P stable The dynamic leakage threshold P is obtained by multiplying by a preset leakage ratio coefficient. leak The calculation formula is P leak =α×P stable (α is the preset leakage ratio coefficient);

[0065] The synergistic effect of the multi-dimensional correction coefficients ensures that the dynamic threshold always matches the current actual operating conditions precisely, avoiding both delays or failures in adsorption success determination caused by excessively high thresholds and misjudgments caused by excessively low thresholds.

[0066] Furthermore, the first stage of the three-stage progressive delay filtering process in S3 is a sliding window mean filter, specifically:

[0067] The original negative pressure signal inside the suction cup is acquired at a preset sampling frequency. The original negative pressure signal is then processed by a moving average of a preset window size to obtain a filtered negative pressure value, which is used for subsequent rising edge detection.

[0068] The original negative pressure signal P inside the suction cup is acquired at a preset sampling frequency. raw The original negative pressure signal is subjected to a moving average processing with a preset window size to obtain the filtered negative pressure value P. filtered (l), the calculation formula is: (N is the preset window size), used for subsequent rising edge detection. The sliding window mean filter can smooth high-frequency glitches and instantaneous fluctuations in the original signal, retain the overall trend of negative pressure change, and will not excessively delay the response speed of the negative pressure signal. At the same time, it can effectively filter out the measurement noise of the sensor itself and the instantaneous negative pressure fluctuations caused by airflow disturbances and slight deformation of the suction cup, ensuring that the subsequent judgment logic can be based on the real negative pressure change trend.

[0069] Furthermore, the second stage of the three-stage progressive delay filtering process in S3 is rising edge detection and delayed start, specifically as follows:

[0070] When the filtered negative pressure value first reaches the first preset proportion of the dynamic stable adsorption threshold, it is marked as adsorption begins and a delay timer of the first preset duration is started; if the filtered negative pressure value is detected to decrease and fall below the second preset proportion of the dynamic stable adsorption threshold during the delay period, the delay timer is reset and the first-level sliding window mean filtering is re-executed.

[0071] The rapid rise and overshoot interference during the initial startup of the filter vacuum pump is eliminated, preventing misjudgment of successful adsorption at the overshoot moment. This solves the most prominent overshoot misjudgment problem of traditional instantaneous judgment methods, fundamentally reducing the risk of desorption and fall due to misjudgment. In a specific embodiment, when the filtered negative pressure value first reaches the dynamic stable adsorption threshold P... stable When the first preset ratio is reached, adsorption begins and a delay timer for a first preset duration is started; during the delay period, the trend of the filtered negative pressure value is continuously monitored. If the negative pressure value decreases and falls below the dynamic stable adsorption threshold P, the timer will detect the change. stable If the second preset ratio indicates that the current negative pressure rise is an overshoot oscillation during the initial startup of the vacuum pump rather than a true stable rise, the delay timer is immediately reset, and the first-stage sliding window mean filtering is re-executed. This system can accurately identify the overshoot phase during the negative pressure rise process. By delaying and allowing the negative pressure to complete the inherent process of "rapid rise - overshoot - oscillation decay - stabilization," it avoids misjudging adsorption success at the overshoot peak. Simultaneously, by detecting the negative pressure decline trend, the judgment process is reset in a timely manner, ensuring that only a true and continuous negative pressure rise can proceed to the subsequent stability verification stage.

[0072] Furthermore, the third stage of the three-stage progressive delay filtering process in S3 is stability verification, specifically:

[0073] After the first preset timer expires, the system enters the second preset time stability verification phase. During this phase, the filtered negative pressure value is collected, and its average value and standard deviation are calculated. When the average value is greater than or equal to the dynamic stability adsorption threshold and the standard deviation is less than or equal to the maximum allowable fluctuation value corresponding to the current material, the adsorption is finally determined to be successful. If the conditions are not met, all timers are reset, and the first-level sliding window mean filtering is re-executed.

[0074] The stability of negative pressure is verified by calculating the average and standard deviation of the negative pressure values ​​during the stability verification phase. Its core advantage lies in ensuring that adsorption is only considered successful when the negative pressure reaches and stabilizes above the dynamic stability adsorption threshold, further improving the accuracy and reliability of the judgment and completely solving the problem that traditional instantaneous judgment methods cannot identify negative pressure stability. The example is as follows: After the first preset timer delay ends, the stability verification phase begins with a second preset time. All filtered negative pressure values ​​are collected during this phase, and the average and standard deviation of the filtered negative pressure values ​​are calculated. When the average value is greater than or equal to the dynamic stability adsorption threshold P... stable Furthermore, if the standard deviation is less than or equal to the maximum allowable fluctuation value corresponding to the current material, it indicates that the negative pressure has reached the pressure level required for stable adsorption and the fluctuation is within the allowable range, providing continuous and reliable adsorption force, and adsorption is ultimately determined to be successful. If any of the above conditions are not met, it indicates that the negative pressure has not yet stabilized or there are continuous fluctuations, and adsorption safety cannot be guaranteed. All timers are immediately reset, and the first-level sliding window mean filtering is re-executed. The average value verification ensures that the overall negative pressure has reached the pressure requirements for adsorption, and the standard deviation verification ensures that the stability of the negative pressure meets the operational requirements. The combination of the two can effectively distinguish between the brief drop after negative pressure overshoot and the true stable state, avoiding adsorption failure caused by unstable negative pressure.

[0075] Furthermore, the adsorption fault judgment based on the dynamic leakage threshold and negative pressure signal change characteristics in S4 specifically includes:

[0076] S41. Adsorption slow leakage judgment: When the filtered negative pressure value of a preset number of consecutive sampling points is lower than the dynamic leakage threshold, it is judged as adsorption slow leakage;

[0077] S42. Judgment of sudden adsorption failure: When the negative pressure value after filtering drops sharply beyond the preset pressure drop threshold within the third preset time period, it is judged as a sudden adsorption failure.

[0078] By distinguishing between two different fault types—slow adsorption leakage and sudden failure—a graded fault assessment is achieved. The core advantage lies in improving the targetedness and timeliness of fault response, avoiding inappropriate emergency handling due to misjudgment of fault type, and simultaneously covering adsorption safety monitoring throughout the entire operation cycle. In one embodiment: after successful adsorption is determined, negative pressure signals are continuously collected and filtered throughout the entire operation process for adsorption fault assessment. Fault assessment is performed when the filtered negative pressure values ​​at a consecutive preset number of sampling points are all below the dynamic leakage threshold P. leak When the suction cup experiences a slow leak, the fault is typically caused by minor wear, foreign object ingress, or partial seal failure, and the negative pressure drop is relatively gradual. Conversely, if the filtered negative pressure drops sharply beyond a preset pressure drop threshold within a third preset time period, it is considered a sudden suction failure. This fault is usually caused by large-scale suction cup detachment, pipe rupture, or sudden vacuum pump shutdown, and the negative pressure drop is extremely rapid. This tiered fault assessment system accurately identifies fault types based on different characteristics of negative pressure changes, providing a scientific basis for subsequent emergency response. It avoids overly aggressive emergency measures that could lead to operational interruptions for slow leaks, while ensuring a rapid response to sudden failures, minimizing safety risks.

[0079] Furthermore, the emergency evacuation mechanism triggered in S4 is as follows: when it is determined that there is slow leakage of adsorption or sudden failure of adsorption, the drone is controlled to smoothly detach from the working wall and the current wall operation task is terminated.

[0080] A unified, progressive emergency evacuation mechanism is triggered for different types of adsorption failures. This mechanism ensures the smooth detachment of the drone from the work surface, preventing loss of control due to sudden adsorption loss and effectively preventing drone crashes or collisions caused by adsorption failure, thus ensuring operational safety. In one embodiment, when a slow adsorption leak or sudden adsorption failure is detected, an adsorption failure state is immediately triggered, and the emergency evacuation mechanism is activated. The drone maintains its current attitude while the vacuum pump power is gradually reduced. Simultaneously, the flight propulsion system is activated to provide auxiliary lift matching the current adsorption force. Once the adsorption force between the drone and the wall is completely released and the flight attitude is stable, the drone smoothly detaches from the work surface, flies to a preset safe area, and hovers, terminating the current wall operation. This progressive evacuation process ensures that the drone maintains a balance of forces throughout the evacuation process, preventing loss of control and crashes due to sudden loss of adsorption force, maximizing drone safety, and avoiding property damage and personal injury caused by crashes.

[0081] Furthermore, the specific process of controlling the drone to smoothly detach from the working wall is as follows: maintain the current attitude of the drone and gradually reduce the power of the vacuum pump, while activating the flight propulsion system to provide auxiliary lift that matches the current adsorption force. After the adsorption force between the drone and the working wall is completely released and the flight attitude is stable, control the drone to detach from the working wall and fly to the preset safe area to hover.

[0082] Furthermore, the adsorption state determination system based on negative pressure dynamic threshold and time delay filtering includes:

[0083] The wall material recognition module is used to collect multi-source sensing data and identify the current wall material;

[0084] The dynamic threshold calculation module communicates with the wall material identification module and is used to calculate the dynamic stable adsorption threshold and dynamic leakage threshold based on the identified wall material and real-time operating parameters, and retrieve the maximum allowable fluctuation value corresponding to the current material.

[0085] The three-stage progressive delay filtering module is connected to the dynamic threshold calculation module and the negative pressure sensor respectively. It is used to perform three-stage progressive delay filtering on the original negative pressure signal and to determine the adsorption success based on the dynamic stable adsorption threshold and the maximum allowable fluctuation value.

[0086] The adsorption status continuous monitoring module is connected to the dynamic threshold calculation module and the three-level progressive delay filtering module. It is used to monitor the negative pressure status after filtering throughout the entire operation cycle and determine adsorption faults. When a fault is determined, an emergency evacuation mechanism is triggered.

[0087] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A method for determining adsorption state based on negative pressure dynamic threshold and time delay filtering, applied to wall operation control of a negative pressure adsorption UAV, characterized in that, Includes the following steps: S1. Collect multi-source perception data of the current working wall surface using multiple sensors carried by the drone, and automatically identify the current wall surface material based on the multi-source perception data; S2. Based on the identified current wall material, combined with the drone's real-time payload, current ambient air pressure, and number of times the suction cup has been used, calculate the dynamic stable adsorption threshold and dynamic leakage threshold that are suitable for the current operating conditions, and at the same time retrieve the maximum allowable fluctuation value corresponding to the current material. S3. Collect the original negative pressure signal inside the suction cup, perform three-level progressive delay filtering on the original negative pressure signal, and determine the adsorption success based on the dynamic stable adsorption threshold and the maximum allowable fluctuation value. S4. After successful adsorption is determined, the negative pressure signal is continuously collected and filtered during the operation. Adsorption failure is judged based on the dynamic leakage threshold and the characteristics of negative pressure signal change. If adsorption failure is determined, the emergency evacuation mechanism is immediately triggered.

2. The adsorption state determination method based on negative pressure dynamic threshold and time delay filtering according to claim 1, characterized in that, S1's automatic identification of the current wall material based on multi-source sensing data specifically includes: S11. Acquire wall images using a monocular camera and extract multiple visual features; measure the wall reflection intensity using lidar to obtain laser features, and combine them to form a multi-dimensional feature vector; S12. Calculate the distance between the multidimensional feature vector of the current wall surface and the multidimensional feature vectors of all samples in the pre-established material feature library, select the preset number of samples with the closest distance, and determine the material of the current wall surface through a voting method; wherein the material feature library pre-stores the multidimensional feature vectors of multiple typical wall surface samples; Several visual features include gray-level mean, gray-level variance, edge density, and gradient orientation histogram features.

3. The adsorption state determination method based on negative pressure dynamic threshold and time delay filtering according to claim 1, characterized in that, S2 specifically includes: S21. Based on the identified current wall material, retrieve the corresponding material basic stability adsorption threshold, basic leakage threshold and maximum allowable fluctuation value from the pre-established basic threshold database; S22. Calculate the drone weight correction factor, ambient air pressure correction factor, and suction cup wear correction factor respectively; wherein the drone weight correction factor is determined based on the ratio of the drone's current actual weight to the standard weight, the ambient air pressure correction factor is determined based on the ratio of the current ambient atmospheric pressure to the standard atmospheric pressure, and the suction cup wear correction factor is determined based on the number of times the suction cup has been used; S23. The dynamic stable adsorption threshold is calculated based on the product of the material's basic stable adsorption threshold and three correction coefficients; the dynamic stable adsorption threshold is multiplied by a preset leakage ratio coefficient to obtain the dynamic leakage threshold.

4. The adsorption state determination method based on negative pressure dynamic threshold and time delay filtering according to claim 1, characterized in that, The first stage of the three-stage progressive delay filtering process in S3 is a sliding window mean filter, specifically: The original negative pressure signal inside the suction cup is acquired at a preset sampling frequency. The original negative pressure signal is then processed by a moving average of a preset window size to obtain a filtered negative pressure value, which is used for subsequent rising edge detection.

5. The adsorption state determination method based on negative pressure dynamic threshold and time delay filtering according to claim 4, characterized in that, The second stage of the three-stage progressive delay filtering process in S3 is rising edge detection and delay start, specifically: When the filtered negative pressure value first reaches the first preset ratio of the dynamic stable adsorption threshold, it is marked as adsorption begins and a delay timer of the first preset duration is started. If, during the delay period, a decrease in the filtered negative pressure value is detected and it falls below the second preset ratio of the dynamic stable adsorption threshold, the delay timer is reset and the first-level sliding window mean filtering is re-executed.

6. The adsorption state determination method based on negative pressure dynamic threshold and time delay filtering according to claim 5, characterized in that, The third stage of the three-stage progressive delay filtering process in S3 is stability verification, specifically: After the first preset timer expires, the system enters the second preset time stability verification stage. During this stage, the filtered negative pressure value is collected, and its average value and standard deviation are calculated. When the average value is greater than or equal to the dynamic stability adsorption threshold and the standard deviation is less than or equal to the maximum allowable fluctuation value corresponding to the current material, the adsorption is finally determined to be successful. If the conditions are not met, all timers are reset, and the first-level sliding window mean filtering is re-executed.

7. The adsorption state determination method based on negative pressure dynamic threshold and time delay filtering according to claim 1, characterized in that, The adsorption fault judgment in S4 based on the dynamic leakage threshold and negative pressure signal change characteristics specifically includes: S41. Adsorption slow leakage judgment: When the filtered negative pressure value of a preset number of consecutive sampling points is lower than the dynamic leakage threshold, it is judged as adsorption slow leakage; S42. Judgment of sudden adsorption failure: When the negative pressure value after filtering drops sharply beyond the preset pressure drop threshold within the third preset time period, it is judged as a sudden adsorption failure.

8. The adsorption state determination method based on negative pressure dynamic threshold and time delay filtering according to claim 7, characterized in that, The emergency evacuation mechanism in S4 is triggered as follows: when it is determined that there is slow leakage of adsorption or sudden failure of adsorption, the drone is controlled to smoothly detach from the working wall and the current wall operation task is terminated.

9. The adsorption state determination method based on negative pressure dynamic threshold and time delay filtering according to claim 8, characterized in that, The specific process of controlling the drone to smoothly detach from the working wall is as follows: maintain the current attitude of the drone and gradually reduce the power of the vacuum pump, while activating the flight propulsion system to provide auxiliary lift that matches the current adsorption force. After the adsorption force between the drone and the working wall is completely released and the flight attitude is stable, control the drone to detach from the working wall and fly to the preset safe area to hover.

10. An adsorption state determination system based on negative pressure dynamic threshold and time delay filtering, characterized in that, include: The wall material recognition module is used to collect multi-source sensing data and identify the current wall material; The dynamic threshold calculation module communicates with the wall material identification module and is used to calculate the dynamic stable adsorption threshold and dynamic leakage threshold based on the identified wall material and real-time operating parameters, and retrieve the maximum allowable fluctuation value corresponding to the current material. The three-stage progressive delay filtering module is connected to the dynamic threshold calculation module and the negative pressure sensor respectively. It is used to perform three-stage progressive delay filtering on the original negative pressure signal and to determine the adsorption success based on the dynamic stable adsorption threshold and the maximum allowable fluctuation value. The adsorption status continuous monitoring module is connected to the dynamic threshold calculation module and the three-level progressive delay filtering module. It is used to monitor the negative pressure status after filtering throughout the entire operation cycle and determine adsorption faults. When a fault is determined, an emergency evacuation mechanism is triggered.

Images